Introduction

No background required. This book assumes no biology, no chemistry, and no medical training. Every organ, every cell, every hormone, and every drug is explained the moment it appears. If you can read a newspaper, you can read this.

This book is education, not medical advice. It explains how diseases and treatments work. It cannot examine you, it does not know your history, and it is not a substitute for a clinician. Never start, stop, or change a medication because of something you read here. If something in a chapter sounds like you, that is a reason to see a doctor, not a reason to self-treat.

What this book is

Almost everyone will get one of the diseases in this book, or love someone who does. Diabetes, high blood pressure, heart disease, cancer, depression, asthma, dementia: between them they account for most of the illness and most of the deaths on earth. And yet the way they are usually explained to the people living with them is a handful of instructions with the reasoning removed. Take this pill. Cut the salt. Watch your sugar. Come back in three months.

This book puts the reasoning back. It takes the diseases people actually get and explains each one all the way down: what is physically going wrong inside the body, why that produces the symptoms it produces, why a particular pill helps, what the pill is doing at the level of molecules, what it costs you in side effects, and which of the things a person does themselves genuinely change the outcome rather than just sounding healthy.

The premise is one sentence:

A disease stops being frightening in the way ignorance is frightening once you can say what it is doing, why, and what each intervention is for.

That is not a promise that understanding cures anything. Understanding a stroke does not unblock an artery. But almost every chronic disease in this book is managed, day to day, by the person who has it, and people manage badly what they have only been told to do rather than shown how it works. Someone who knows that metformin works mostly by telling the liver to stop pouring glucose into the blood overnight has a reason to take it at night with food. Someone who knows that the damage from high blood pressure is silent and cumulative has a reason to keep taking a pill that makes them feel no different at all.

What counts as "common and famous"

Two kinds of disease made it into these pages.

The common ones, meaning the diseases that produce most of the world's illness and death: heart disease and stroke, the cancers, chronic lung disease, diabetes and obesity, kidney disease, dementia, depression and anxiety, the ordinary infections. These are the ones you or your family will most likely meet.

The famous ones, meaning the diseases that shaped history, medicine, or public imagination even when they are now rare in wealthy countries: tuberculosis, HIV, malaria, cholera, measles, polio, sickle cell disease, cystic fibrosis, Huntington's disease. Some of these are still mass killers somewhere else on earth, which is itself one of the book's subjects.

What is deliberately left out: the very rare, the very specialised, and anything where a short explanation would be more dangerous than no explanation. The last part of the book explains how to think about the rest.

Every chapter answers the same twelve questions

The disease chapters are built to a fixed template, so that once you have read one you can navigate any of them, and so that two diseases can be compared honestly rather than through whichever facts happened to be interesting.

1. What it is. The definition, in plain words, and the exact numbers or findings that define it, because most modern diagnoses are a threshold on a measurement.
2. The history. When people first recognised it, what they thought it was, and the discovery that changed everything.
3. What actually goes wrong. The mechanism, built up from the normal biology first. You cannot understand insulin resistance without first understanding what insulin does when everything is working.
4. What it does to the body. Organ by organ, and on what timeline.
5. Is it deadly? Honest numbers on mortality and prognosis, and what changes them.
6. Is it contagious? How it spreads, or a clear statement that it does not, because this is one of the most common and most damaging misunderstandings.
7. Who gets it. Age, sex, genes, family history, ancestry, country, income, and the difference between a risk factor and a cause.
8. Treatment, and how the treatments actually work. Each drug class or procedure explained mechanically: what it binds to, what it blocks, what changes as a result.
9. What treatment costs. Side effects, the ones that are common and tolerable and the ones that are rare and serious, plus the honest trade-off each drug represents.
10. What the person can do. Diet, movement, sleep, monitoring, adherence, vaccination, and which of these has evidence behind it rather than folklore.
11. Living with it. What the disease does to a life, not just to an organ.
12. What's next. The research that is likely to change the picture, and the hype that probably will not.

Each chapter ends with sources and notes, and a "👉" pointing to the next one.

The order things are explained in

The book is built in three layers.

First, the body itself (Chapters 1 to 11). You cannot understand a broken system without knowing the working one, so these chapters build you from the bottom up: what a protein, a fat, a vitamin, and an enzyme actually are; what a cell contains and how often each tissue replaces itself; then every major organ with what it does, how much spare capacity it holds, how long it lasts, what wears it out, and what protects it. Then four chapters on what food, movement, sleep, stress, and time actually do inside you, mechanically, and what the evidence supports.

Second, how disease works (Chapters 12 to 17). What a disease even is, how the immune system defends the body, how infections move between people, how genes cause and modify illness, how a drug travels through the body and produces both its effect and its side effects, and how anyone knows any of this.

Third, the diseases themselves (Chapters 18 to 60), grouped by the system they attack: metabolism and circulation, cancer, infection, brain and mind, airways and immunity, inherited disease, the wearing body, and a final group covering the conditions that are extremely common and rarely explained: anaemia, eyes and ears and teeth, skin, blood clots, sleep disorders, ADHD and autism and eating disorders, the urinary and gynaecological conditions, and injury.

Then the wider view (Chapters 61 to 65). Disease is not distributed randomly across the world. What kills a 60-year-old in Japan is not what kills a 60-year-old in Nigeria, and the reasons are historical, economic, and biological in that order. These chapters cover who gets sick and where, why some genetic diseases concentrate in some ancestries, which preventive measures actually move national death rates, a concrete practical playbook, and what the next fifty years plausibly hold.

Finally, two reference chapters decode the drugs: the GLP-1 class in full, and how to read any drug name, lab value, or medical term you have never seen.

If you want the practical version immediately, Chapter 64 is the operational summary, and it links back to the evidence for everything it recommends.

On numbers

Health statistics move. Prevalence counts get revised, mortality falls in one region and rises in another, and two reputable agencies will give you two different numbers for the same disease because they define it differently. This book uses figures from named sources (the World Health Organization, the International Diabetes Federation, the Global Burden of Disease study, national statistics agencies, and the primary medical literature) and treats them as representative rather than exact. Where the science is genuinely unsettled, the text says so instead of picking a side.

Let's start with the thing all of this happens to: the body itself. 👉

How to read this book

A few notes to make the tour easier.

The safety rule, first

This book explains diseases and drugs in enough detail that it would be easy to mistake it for clinical guidance. It is not.

Never act on this book alone. Doses, thresholds, and drug choices in these pages are illustrative, chosen to explain a mechanism. Real prescribing depends on your kidney function, your other medicines, your pregnancy status, your age, and a dozen other things a book cannot know. Use the book to understand what your clinician is doing and to ask better questions. Do not use it to overrule them.

Two situations override everything else in this book: new, severe, or sudden symptoms (chest pain, one-sided weakness, difficulty breathing, sudden severe headache, confusion, a stiff neck with fever, thoughts of suicide) mean emergency care now, not reading. And anything that changes fast deserves a real examination, because speed is itself a diagnostic clue.

Every disease chapter has the same twelve sections

The order is fixed, so you can jump straight to the part you want:

1. What it is
2. The history
3. What actually goes wrong
4. What it does to the body
5. Is it deadly?
6. Is it contagious?
7. Who gets it
8. Treatment, and how it works
9. What treatment costs
10. What the person can do
11. Living with it
12. What's next

Then sources and notes, then a 👉 arrow to the next chapter.

The first seventeen chapters are foundations rather than diseases, so they do not follow that template. Chapters 1 to 7 build the body itself: the molecules, the cells, and each organ with what it does, how much reserve it has, what damages it, and what protects it. Chapters 8 to 11 cover what food, movement, sleep, stress, and time actually do inside you. Chapters 12 to 17 cover how disease, immunity, genes, drugs, and evidence work. The last two parts, on populations and on drug terminology, do not follow the template either.

Boxes and signs to watch for

Don't be confused: ... boxes untangle two things people routinely mix up, like type 1 and type 2 diabetes, a heart attack and cardiac arrest, or HIV and AIDS. Mixing these up is not a small error. It changes what you think the disease is.

In short: lines sit directly under most section headings and compress that section into one sentence, so you can skim a chapter and stop where it matters.

Bold marks a technical term the first time it is defined. Tables compare things that are genuinely comparable. A "👉" at the end of a chapter points to the next.

If a term or a drug name is unfamiliar

Two places to look. The glossary defines every term the book uses. Chapter 67 is a decoder: it explains why a drug's name ending tells you its class, how medical words are assembled from about forty Greek and Latin pieces, how to convert between the two systems of lab units, and what the frequency words on a package leaflet actually mean numerically. If you have ever wondered why GLP-1 is called GLP-1, that is where the answer is, alongside Chapter 66, which covers that drug class in full.

Four words this book uses precisely

In short: Acute means fast rather than severe, chronic means long-lasting rather than mild, and a risk factor is not a cause.

Ordinary conversation treats these as synonyms. Medicine does not, and the difference matters in almost every chapter.

Two more pairs are worth fixing in your head now, because they recur constantly:

• Sign vs symptom. A symptom is what the patient feels (pain, nausea). A sign is what someone else can measure (a fever, a murmur, a blood result). Many of the most dangerous diseases in this book, hypertension above all, have signs long before they have symptoms. That is exactly why they are dangerous.
• Incidence vs prevalence. Incidence is how many new cases appear in a period. Prevalence is how many people have it right now. A disease that is easy to catch and quick to resolve has high incidence and low prevalence. A disease that is hard to catch and impossible to cure, like HIV in the treatment era, can have falling incidence and rising prevalence at the same time. Headlines confuse these two constantly.

On the numbers in the tables

In short: Global figures are modelled estimates, relative risk is meaningless without the baseline, and survival statistics describe groups rather than individuals.

Every chapter carries numbers: how many people have the disease, how many die of it, how much a treatment reduces risk. Three cautions.

Global figures are estimates, not counts. Nobody counted 589 million people with diabetes. That number comes from surveys of samples, extrapolated with a model. It is the best available answer and it is not precise to the last million.

Relative risk is not absolute risk. "Doubles your risk" is meaningless without the starting number. Doubling a 1-in-10,000 risk is a rounding error. Doubling a 1-in-4 risk is a catastrophe. Chapter 17 explains how to read this properly, and every chapter afterwards tries to give absolute numbers.

Survival numbers describe groups, not you. A "five-year survival of 20 percent" is a property of a large population diagnosed years ago, under older treatments, across all ages and stages. It tells you about the disease. It does not tell any individual their future.

Read in any order

The chapters are grouped by body system, but each stands alone. If you came here for one disease, go straight to it; it will link back to whichever foundation chapter it needs. If you are reading start to finish, the foundation chapters will make every disease chapter shorter and clearer.

If you want the practical version first, go to Chapter 64, which is the operational summary: what to do, what to measure, what to have screened and when, what the warning signs are, and what to ignore.

Use the search box (or press S) to jump to any disease, drug, or term.

With that, we start with the thing all of this happens to. 👉

The Body in Outline

TL;DR. You are roughly 37 trillion cells organised into about 200 cell types, arranged into four tissue types, assembled into around 78 organs, working in 11 systems. The whole thing runs on about 100 watts, roughly a bright old-fashioned light bulb, and holds itself within a set of narrow chemical ranges every second of your life without asking you. The single most useful idea in this chapter is reserve capacity: nearly every organ is built with far more capacity than daily life requires, which is why you can lose one kidney, most of a liver, or half a lung and feel completely normal. That reserve is also why the diseases in this book are silent for so long. By the time you feel an organ failing, most of its spare capacity is already gone.

Key takeaways

• Reserve capacity is the reason chronic disease is silent. Kidneys give no symptoms until roughly 80 to 90 percent of function is lost. The liver can lose most of its mass and still work. Symptoms are a late alarm, not an early one.
• Your body is about 60 percent water, 16 percent protein, 15 to 25 percent fat, and 6 percent minerals, and every one of those fractions is doing a job.
• You carry roughly as many bacterial cells as human ones, not ten times as many. That widely repeated 10:1 figure was an estimate from 1972 and was corrected in 2016.
• Every organ is a trade-off between capacity and cost. Building spare capacity takes energy, so evolution supplied enough for a hard life at 30, not a long life at 80.
• Almost all body regulation runs on negative feedback: something senses a value, compares it to a target, and acts to reduce the difference.
• Organs differ enormously in whether they can repair themselves. Liver regenerates, skin and gut replace themselves constantly, and heart muscle and brain cells largely do not.

What you are made of

In short: Water, protein, fat, and minerals in fairly predictable proportions, and knowing the proportions explains a surprising amount.

Two things in that table are worth pausing on.

Fat is not inert. The chapter on obesity explains that fat tissue is a functioning endocrine organ that secretes hormones and inflammatory signals. Below a certain body fat percentage, roughly 3 to 5 percent in men, the body starts failing: hormone production drops, immunity falls, and menstruation stops in women well before that point. Fat is a required organ that can also become a disease.

The carbohydrate store is tiny. You carry about 100 g of glycogen in the liver and 400 g in muscle, together perhaps 2,000 calories, roughly a day of fuel. Fat stores run to tens of thousands of calories. This asymmetry explains why fasting works, why marathon runners "hit the wall" at about two hours, and why the body defends blood glucose so aggressively: the buffer is small and the brain cannot run on fat directly.

The scale of the thing

In short: The numbers are worth knowing because they make the later chapters concrete rather than abstract.

Don't be confused: you are not "mostly bacteria." The claim that bacteria outnumber your own cells ten to one comes from a back-of-envelope estimate published in 1972 and repeated for forty years. A careful 2016 recount put the ratio at close to 1:1, around 38 trillion bacteria to 37 trillion human cells, and noted that a single bowel movement meaningfully changes it. Bacteria are genuinely important, and the drama of the old figure was not real.

The eleven systems

In short: Eleven systems, each with a job, and most disease is one of them failing in a way that stresses the others.

The division is a teaching convenience, not a fact about the body. The kidney is a urinary organ that also controls blood pressure, makes a hormone that drives red cell production, and activates vitamin D. The gut contains more neurons than the spinal cord and produces most of the body's serotonin. Fat tissue is an endocrine gland. Bone is an endocrine organ too. Every time this book says a disease of one system causes problems in another, it is because the systems were never really separate.

Reserve capacity, and why it explains so much

In short: Organs are built with several times the capacity daily life needs, which is why losing most of one causes no symptoms until very late.

This is the single most important concept for understanding why the diseases in this book behave the way they do.

Read that table twice, because it explains a pattern that runs through the entire book:

The disease starts long before the symptom. Type 2 diabetes was developing for a decade before diagnosis. Kidney disease is silent until stage 4. Parkinson's has destroyed most of a brain nucleus before the first tremor. High blood pressure is damaging arteries for twenty years without producing a single sensation.

Which means symptoms are a bad early-warning system and measurements are a good one. This is the entire argument for blood pressure checks, HbA1c, kidney function tests, and screening programmes, and it is why the practical chapter at the end of this book (Chapter 64) is built around measurement rather than around how you feel.

And it explains why reserve is worth building deliberately. Peak bone mass, peak muscle mass, and peak aerobic fitness are all reached in early adulthood and decline afterwards. The higher the peak, the longer it takes to fall below the threshold where function is lost. That is the physiological argument for exercise in your thirties: you are not treating a problem, you are raising the ceiling you will spend the next fifty years descending from.

How the body holds itself steady

In short: Nearly all regulation is negative feedback, and disease is usually that loop breaking.

Your body holds dozens of values inside narrow ranges: core temperature near 37 degrees Celsius, blood pH between 7.35 and 7.45, blood glucose between roughly 70 and 140 mg/dL, plus sodium, potassium, calcium, oxygen, carbon dioxide, blood pressure, and water.

Every one of these is held by negative feedback: a sensor measures the value, a controller compares it to a target, and an effector acts to reduce the difference. When the value goes up, the response pushes it down.

A worked example, blood glucose after a meal:

1. Glucose rises as food is absorbed.
2. Sensor: beta cells in the pancreas detect the rise.
3. Controller: those cells decide how much insulin to release.
4. Effector: muscle, fat, and liver cells take glucose out of the blood.
5. Glucose falls, insulin release slows, and the loop settles.

Break any of the three parts and you get a disease. Chapter 12 develops this properly, and Chapter 18 shows exactly what happens when the sensor is destroyed (type 1) versus when the effector stops listening (type 2).

Positive feedback exists too, and is rarer because it is unstable by nature. The body uses it where an event needs to be fast and complete: blood clotting, where each activated factor activates more; labour contractions; and the nerve impulse itself. Positive feedback loops that run without an off switch are dangerous, which is why disseminated intravascular coagulation and cytokine storms are emergencies.

Repair: what heals and what does not

In short: Tissues that replace themselves constantly heal well and are more cancer-prone; tissues that do not, scar instead.

Body tissues fall into three groups by how they respond to damage, and this determines whether an injury heals invisibly or leaves permanent loss.

That third row is why several of the most feared diseases in this book are feared: they destroy tissue that cannot come back. It is also why prevention matters more for those organs than for any others, and why a treatment that arrives in the first hour of a heart attack or stroke is worth so much more than one that arrives on day two.

Scar tissue is not a smaller version of the original. It is collagen: strong, but not contractile, not conductive, and not able to do the job. A scarred heart pumps less and conducts electricity unpredictably. A scarred liver obstructs blood flow. A scarred lung is stiff. Fibrosis is one of the most common final pathways in this book.

What "normal" means

In short: A normal range is a statistical statement about a reference population, not a boundary between health and illness.

A laboratory reference range is usually built by measuring a healthy population and taking the middle 95 percent. Two consequences follow immediately.

By construction, 1 in 20 healthy people falls outside a normal range. Run 20 unrelated tests on a completely healthy person and the odds of at least one abnormal result are better than even. This is why comprehensive "full body" blood panels in people without symptoms generate so much follow-up and so little benefit.

Reference ranges depend on the reference population. They vary with age, sex, pregnancy, altitude, and, as Chapter 62 explains, sometimes with ancestry in ways that have been handled badly. A haemoglobin that is normal in Denver is anaemic at sea level, because altitude drives red cell production up.

And two more things a number cannot tell you: your own baseline (a creatinine of 1.1 is fine if you have always been 1.1 and alarming if you were 0.7 last year), and the trend, which is almost always more informative than a single value.

What this part covers

In short: Six more chapters: the molecules, the cells, the organs in three groups, and then what food, movement, sleep, and time actually do to all of it.

The rest of this part builds the body from the bottom up, because every disease chapter later assumes it.

• Chapter 2: what a protein, an amino acid, a fat, a carbohydrate, a vitamin, and an enzyme actually are, and what each does.
• Chapter 3: the cell, its parts, how often each tissue replaces itself, and why that matters.
• Chapters 4 to 7: the organs one at a time. For each: what it does, how much capacity it has, how long it lasts, what damages it, what protects it, and what the early warning signs are.
• Chapters 8 to 10: what food, movement, sleep, stress, and mood actually do inside the body, mechanically.
• Chapter 11: what ageing is, how fast each system declines, and what changes the slope.

Sources and notes

Cell count of about 37 trillion: Bianconi et al., Annals of Human Biology, 2013. Bacteria to human cell ratio close to 1:1: Sender, Fuchs, and Milo, PLoS Biology and Cell, 2016, correcting the widely cited 10:1 figure that originated with Luckey, 1972. Body composition percentages are standard reference-man figures and vary with age, sex, and adiposity. Neuron count of about 86 billion: Azevedo, Herculano-Houzel et al., Journal of Comparative Neurology, 2009. Alveolar surface area, vessel length, and blood volume figures are standard physiology (Guyton and Hall; West, Respiratory Physiology). Liver regeneration from about 25 percent residual mass is established in living-donor transplantation practice. Thresholds for symptom onset (kidney, pancreatic beta cell, substantia nigra) are cited in their respective chapters. Reference range construction as the central 95 percent of a healthy population is standard clinical chemistry.

Open questions. Total organ and cell-type counts depend entirely on definitions and are not settled numbers. How much cardiomyocyte and neuronal renewal occurs in adult humans is an active area of research, with current estimates of heart muscle turnover around 1 percent a year in young adults, falling with age.

Next: the molecules everything above is built from, defined properly. 👉

The Molecules of Life

TL;DR. Six classes of molecule do everything. Water is the medium. Proteins are the machines and the structure, built from 20 amino acids of which 9 must come from food. Carbohydrates are the fast fuel. Fats are the dense fuel, the membranes, and the raw material for several hormones. Nucleic acids hold the instructions. Vitamins and minerals are the small parts that let the machines work, needed in milligram quantities and catastrophic when absent. Nearly every nutrition argument you have ever heard is an argument about the middle three, and most of them make more sense once you know what the molecules actually are.

Key takeaways

• A protein is a chain of amino acids folded into a shape, and the shape is the function. Change the shape and you change the job, which is what a mutation does.
• Nine amino acids are essential, meaning your body cannot make them and they must come from food. The other eleven you can build yourself.
• You recycle roughly 250 to 300 grams of protein a day internally, far more than you eat, which is why protein needs are lower than the supplement industry implies and higher than the official minimum for older and active people.
• ATP is the body's energy currency, and you make and spend roughly your own body weight in it every day by recycling the same molecules thousands of times.
• Enzymes are proteins that speed up specific reactions, often by factors of millions. Almost every drug that ends in "-inhibitor" is blocking one.
• Vitamins are not fuel. They are parts, mostly enzyme helpers, which is why more than enough gives no extra benefit and several are toxic in excess.

Water

In short: About 60 percent of you, and it is the reason biochemistry works at all rather than a passive filler.

Water is the solvent every reaction in your body happens in, and its unusual properties are load-bearing:

• It dissolves charged and polar things (salts, sugars, most proteins) and repels greasy things (fats). That single split is what makes cell membranes possible: a membrane forms spontaneously because fat molecules cluster away from water. Life is organised by what water will and will not mix with.
• It carries an enormous amount of heat per degree, so your temperature moves slowly rather than swinging with every meal or effort. Sweat exploits the same property: evaporating water removes a lot of heat.
• It transports. Blood is mostly water, and so is the fluid between cells, urine, and the contents of the gut.

Water balance is tightly controlled. You lose roughly 2 to 2.5 litres a day through urine, skin, breath, and stool, and you replace it from drinks, food (fruit and vegetables are mostly water), and the water your own metabolism produces. Thirst is triggered by a rise in blood concentration of about 1 to 2 percent, which is sensitive, and it becomes less reliable with age, which is why dehydration is common in older people.

The electrolytes dissolved in that water are as important as the water itself. Sodium, potassium, chloride, calcium, magnesium, and phosphate carry electrical charge, and nerves and muscle work by moving them across membranes. Get them wrong and the consequences are immediate: low sodium causes confusion and seizures, high or low potassium stops the heart.

Don't be confused: drinking more water is not automatically better. Kidneys excrete excess readily, so extra water mostly produces extra urine. Drinking a very large volume quickly can dilute blood sodium dangerously, which has killed people in endurance events and in water drinking contests. The useful rule is to drink to thirst plus a little more in heat, illness, or hard exercise, and to check urine colour rather than counting glasses.

Proteins and amino acids

In short: Chains of amino acids folded into shapes, and they do nearly every active job in the body.

What an amino acid is

An amino acid is a small molecule with three parts: an amino group at one end, an acid group at the other, and a side chain in the middle that makes it different from the other 19. The side chain is everything: it decides whether that position on a protein is greasy or water-loving, positively or negatively charged, bulky or small, reactive or inert.

Amino acids link end to end by peptide bonds. A few linked together is a peptide; dozens to thousands is a protein. GLP-1 (Chapter 66) is 30 amino acids, so it is a peptide. Haemoglobin is 574, so it is a protein.

The twenty, and the nine that matter for your diet

Some of the second column are conditionally essential: during illness, injury, rapid growth, or in premature infants, the body cannot make enough and they must be supplied. Glutamine and arginine are the usual examples.

Protein quality means how well a food's amino acid profile matches what you need. Animal proteins (eggs, dairy, meat, fish) contain all nine essential amino acids in useful proportions. Most single plant foods are low in one or two: grains are low in lysine, legumes are low in methionine. Eating a variety across the day solves this completely, and the old advice to carefully combine proteins within a single meal turned out to be unnecessary, because the body maintains a free amino acid pool that lasts hours.

From chain to shape to job

A protein chain folds, spontaneously and reproducibly, into a specific three-dimensional shape driven mainly by which side chains want to avoid water. That shape creates surfaces and pockets that fit other molecules. The shape is the function.

That is why heat, extreme pH, and some chemicals denature proteins, unfolding them irreversibly: a cooked egg white cannot be uncooked. It is also why a single amino acid substitution can be catastrophic, as in sickle cell disease (Chapter 48), and why misfolding causes disease directly in Alzheimer's, Parkinson's, and the prion diseases.

What proteins actually do

How much protein you need, honestly

The official minimum (RDA) is 0.8 g per kg of body weight per day, which is set as the amount that prevents deficiency in almost everyone, not the amount that is optimal.

The evidence for higher intakes is strongest in two groups:

Two practical points. Distribution matters more than people expect: roughly 25 to 40 g in each of three or four meals stimulates muscle protein synthesis better than the same total eaten mostly at dinner. And you recycle far more protein than you eat: about 250 to 300 g a day is broken down and rebuilt internally, against a dietary intake of perhaps 60 to 100 g. Your body is a recycling operation with a top-up, not a throughput system.

Carbohydrates

In short: The fast fuel, in three sizes, and the only class of nutrient with no essential requirement.

Carbohydrates are sugars, in chains of different lengths:

• Monosaccharides, single sugars: glucose (the universal fuel), fructose (fruit sugar, handled mainly by the liver), galactose.
• Disaccharides, two joined: sucrose (table sugar: glucose plus fructose), lactose (milk sugar: glucose plus galactose, and the one most adults worldwide cannot digest, Chapter 62), maltose.
• Polysaccharides, long chains: starch in plants, glycogen in your liver and muscle, and fibre, which you cannot digest at all.

Glucose is the reference fuel. Every cell can burn it, the brain relies on it almost exclusively under normal conditions (it can switch partly to ketones during fasting), and red blood cells can use nothing else because they have no mitochondria. This is why blood glucose is defended so tightly and why Chapter 18 exists.

Fibre is a carbohydrate you cannot digest, and it matters anyway.

Fermentable fibre is eaten by colon bacteria, which produce short-chain fatty acids that feed the cells lining the colon directly. Most people eat around half the recommended 25 to 30 g a day, and low fibre intake is among the most consistent dietary associations with colorectal cancer, cardiovascular disease, and mortality.

There is no essential carbohydrate. The body can make all the glucose it needs from amino acids and from the glycerol backbone of fats. This is the physiological basis for low-carbohydrate diets being survivable, and it is not an argument that they are superior, only that carbohydrate, uniquely among the three macronutrients, has no minimum requirement.

Fats

In short: Dense fuel, every cell membrane, several hormones, and the transport system for four vitamins.

Fatty acids are chains of carbon atoms with an acid group at one end. Three or so attached to a glycerol backbone make a triglyceride, which is what dietary fat and body fat mostly are.

The classification everyone argues about is just chemistry:

Two fatty acids are essential: linoleic acid (omega-6) and alpha-linolenic acid (omega-3). Your body cannot make either. From them it builds the longer-chain omega-3s EPA and DHA, though the conversion is inefficient, which is why oily fish is the reliable source. DHA is a major structural component of brain and retinal membranes.

Fats do four things beyond fuel:

1. Membranes. Every cell is wrapped in a double layer of phospholipids, fat molecules with a water-loving head and water-fearing tails. Cholesterol sits among them controlling fluidity.
2. Hormone precursor. Cholesterol is the raw material for cortisol, aldosterone, oestrogen, testosterone, and vitamin D. This is why cholesterol is essential and why the goal is controlling the particles that deposit it in artery walls (Chapter 21), not eliminating it.
3. Vitamin transport. Vitamins A, D, E, and K dissolve only in fat, so a fat-free meal absorbs them poorly, and conditions that block fat absorption cause deficiency in all four.
4. Signalling. Prostaglandins and related molecules, built from fatty acids, mediate inflammation, pain, fever, and clotting. This is exactly what aspirin and the other NSAIDs block (Chapter 16).

Nucleic acids

In short: DNA is the archive, RNA is the working copy, and the whole system exists to specify proteins.

DNA is a four-letter code (A, C, G, T) in a double helix, about 3 billion letters long, packed into 46 chromosomes in nearly every cell. Roughly 20,000 stretches of it are genes.

RNA is the single-stranded working copy. Messenger RNA carries a gene's instructions out of the nucleus, transfer RNA brings the matching amino acid, and ribosomal RNA forms part of the machine that links them.

The flow is: DNA to RNA to protein. That is the reason the previous section exists: nearly all of your genetic information is instructions for building proteins, and this book's genetic diseases (Chapter 15, Chapter 49) are all cases of one protein being wrong or missing.

Two practical consequences appear repeatedly. mRNA vaccines (Chapter 65) simply deliver the middle step, letting your own cells make one protein for the immune system to learn. And antibiotics and antivirals frequently work by attacking the bacterial or viral version of this machinery, which is different enough from ours to target.

ATP: the energy currency

In short: The molecule that carries energy from fuel to work, made and spent so fast that you turn over roughly your own body weight in it daily.

Burning fuel does not directly power anything. Energy from food is used to attach a third phosphate group to a molecule called ADP, making ATP (adenosine triphosphate). Snapping that phosphate off releases energy exactly where it is needed: to contract a muscle fibre, pump an ion across a membrane, or build a protein.

Three ways to make it, and knowing them explains a great deal about exercise (Chapter 9):

You hold only about 250 g of ATP at any moment and use roughly 50 to 75 kg of it a day, meaning each molecule is recharged hundreds or thousands of times daily. Mitochondria do most of that work, which is why mitochondrial density in muscle is one of the main things that improves with endurance training, and why mitochondrial dysfunction appears in the ageing chapter.

Don't be confused: lactate is not what makes your muscles sore. Lactate rises during hard exercise and clears within an hour, and it is a fuel that other tissues burn rather than a waste product. The soreness that peaks a day or two later is microscopic muscle damage and inflammation, which is a different process entirely.

Vitamins

In short: Thirteen small molecules you cannot make, needed in tiny amounts, mostly acting as helpers that let enzymes work.

A vitamin is an organic compound required in small amounts that the body cannot synthesise adequately. Most act as coenzymes: small helper molecules an enzyme needs in order to function. That is why they are needed in milligram or microgram quantities, and why having more than enough does nothing extra.

Fat-soluble (A, D, E, K) versus water-soluble (B group and C) is the distinction that matters practically. Water-soluble vitamins are excreted when in excess, so toxicity is rare (B6 is the notable exception). Fat-soluble vitamins accumulate, so excess can be genuinely harmful: too much vitamin A causes liver damage and birth defects, and very high vitamin D causes dangerous blood calcium.

The honest position on supplements: correcting a genuine deficiency works and matters enormously. Taking vitamins without a deficiency has repeatedly failed to show benefit in large trials, and several have caused harm (Chapter 63). The common real deficiencies worth knowing about are iron, vitamin D, B12, iodine, and folate.

Minerals

In short: Inorganic elements you must eat, some in grams and some in micrograms, and the ones you have heard least about are often the ones with the largest population effect.

Putting it together: what a meal becomes

In short: Protein becomes amino acids for rebuilding, carbohydrate becomes glucose for immediate fuel, fat becomes fatty acids for storage and membranes, and nothing is stored as what it arrived as.

A plate of chicken, rice, olive oil, and broccoli, followed through the body:

1. Chicken protein is cut by stomach acid and enzymes into amino acids, absorbed in the small intestine, and enters a shared pool. Some rebuilds muscle and enzymes; the rest is stripped of its nitrogen (excreted as urea by the kidneys) and burned or stored.
2. Rice starch is cut into glucose, absorbed, and raises blood glucose, which triggers insulin. Glucose is burned immediately, stored as glycogen until those stores are full, and converted to fat beyond that.
3. Olive oil is emulsified by bile, cut by lipase, absorbed, reassembled into triglycerides, packaged into particles, and distributed for burning, storage, or membrane building.
4. Broccoli fibre is not digested at all. It slows the absorption of everything above, adds bulk, and is fermented in the colon.
5. The vitamins and minerals in all of it are absorbed separately and used as parts, not fuel.

Two things follow. Your body does not store food as food; everything is broken to components and rebuilt, which is why "you are what you eat" is chemically true and nutritionally misleading. And the same calories behave differently depending on what they arrive with: fibre, protein, and fat all slow the glucose rise from the same amount of starch, which is most of why whole foods and refined ones produce different responses (Chapter 8).

Sources and notes

Standard biochemistry (Lehninger, Principles of Biochemistry; Berg, Tymoczko, and Stryer, Biochemistry). Essential amino acid list and protein RDA of 0.8 g/kg/day: WHO/FAO/UNU protein and amino acid requirements report and national dietary reference intakes. Higher protein targets for older and training adults: Bauer et al., JAMDA, 2013 (PROT-AGE), and Morton et al., British Journal of Sports Medicine, 2018, which found benefit plateauing around 1.6 g/kg/day. Whole-body protein turnover of roughly 250 to 300 g/day is a standard tracer-study figure. ATP turnover of 50 to 75 kg/day is a commonly cited calculation from total daily energy expenditure. Lactate as fuel rather than waste: Brooks, Cell Metabolism, 2018. Fibre intake and outcomes: Reynolds et al., The Lancet, 2019. Trans fat elimination outcomes: national policy evaluations. Vitamin and mineral functions and deficiency syndromes are standard nutrition references.

Open questions. Optimal protein intake across the lifespan is still argued, particularly the balance between muscle preservation and the possibility that high protein intake affects longevity pathways. The health effects of specific saturated fatty acids, and whether the food matrix matters more than the fatty acid profile, remain actively debated.

Next: how these molecules are organised into the smallest living unit, and how often each part of you is replaced. 👉

Cells, Tissues, and How Often You Are Replaced

TL;DR. A cell is a bag of water wrapped in a fat membrane, containing a library (the nucleus), power stations (mitochondria), factories (ribosomes), a packaging department (the Golgi), and a recycling plant (lysosomes). Most of your body is replaced continuously and at wildly different speeds: the lining of your gut is renewed every few days, your skin every month, your red blood cells every four months, your skeleton about every ten years, and the neurons in your cortex and most of your heart muscle essentially never. That single table explains why chemotherapy makes hair fall out, why a gut infection resolves in days and a heart attack leaves a permanent scar, and why the tissues that renew fastest are the ones that most often turn cancerous.

Key takeaways

• The cell membrane is a double layer of fat molecules, which forms spontaneously because their tails avoid water. Every drug, nutrient, and signal has to get past it, and how it does so determines a great deal of pharmacology.
• Mitochondria have their own DNA, inherited only from your mother, because they descend from bacteria absorbed by an ancestral cell around 1.5 billion years ago.
• Turnover rates vary by a factor of a thousand between tissues, and that variation predicts healing, chemotherapy side effects, and cancer risk.
• Apoptosis is controlled cell suicide, and it is as important as cell division. Too little produces cancer; too much produces neurodegeneration.
• Most cells can divide only a limited number of times (the Hayflick limit) because their chromosome end-caps shorten each division. Cancer cells escape this.
• There are only four tissue types, and every organ in your body is a specific arrangement of them.

The cell, part by part

In short: Ten components, each with a job, and knowing them makes most disease mechanisms in this book legible.

The membrane, and why it explains so much pharmacology

A phospholipid has a water-loving head and two water-fearing tails. Drop a lot of them into water and they arrange themselves automatically into a double sheet, tails inward, heads facing the water on both sides. No instruction is needed; it is simply the lowest-energy arrangement. That is the cell membrane, and it is the reason a cell can exist as a separate compartment at all.

The practical consequences run through the rest of the book:

• Greasy molecules cross easily; charged ones do not. So fat-soluble drugs reach the brain and water-soluble ones often do not (Chapter 16).
• Anything the cell needs that cannot cross by itself needs a door: a channel, a pump, or a transporter. Those doors are drug targets, which is what SGLT2 inhibitors and SSRIs are blocking.
• Receptors sit in the membrane with one part outside and one inside, converting an external signal into an internal action. This is how insulin, adrenaline, and most hormones work without ever entering the cell.

Mitochondria, the former bacteria

Mitochondria have two membranes, their own circular DNA, and their own ribosomes that look bacterial. The explanation, now well supported, is that an ancestral cell engulfed a bacterium around 1.5 billion years ago and the two never separated.

Three consequences that matter medically:

• You inherit mitochondria only from your mother, since the egg supplies them and sperm contribute essentially none. Mitochondrial diseases therefore follow a distinctive inheritance pattern (Chapter 15).
• Some antibiotics affect them. Because mitochondrial ribosomes resemble bacterial ones, drugs such as linezolid and the aminoglycosides can interfere with them, which is part of why they cause the toxicities they do.
• Tissues with high energy demand have the most. Heart muscle cells are roughly a third mitochondria by volume. Endurance training increases mitochondrial number and size in muscle, which is a large part of what "getting fitter" physically means.

How often you are replaced

In short: Turnover ranges from days to never, and where a tissue sits on that scale predicts how it heals, how it responds to chemotherapy, and how likely it is to become cancerous.

Three clinical patterns fall straight out of this table.

Chemotherapy side effects are predictable from it. Classical chemotherapy kills dividing cells. The tissues at the top of the table divide fastest, so those are the ones that suffer: bone marrow, gut lining, hair. There is nothing arbitrary about the side effect list (Chapter 26).

Cancer risk tracks division. Every division is an opportunity to copy DNA wrongly. Tissues that divide constantly accumulate more mutations, and the lifetime cancer risk of different tissues correlates strongly with their total stem cell divisions. This is why colon, skin, and blood cancers are common and cancers of heart muscle are vanishingly rare.

Permanent damage happens where turnover is zero. Heart, brain, retina, and inner ear are the organs where an insult leaves lasting loss, and they are the organs where prevention and speed of treatment matter most.

Cell division, and the brakes on it

In short: A tightly checkpointed cycle, and cancer is what happens when the checkpoints fail.

A cell that is going to divide runs a four-phase cell cycle: it grows (G1), copies its entire 3-billion-letter genome (S), grows and prepares (G2), then divides (M, mitosis). Most cells in your body are not in the cycle at all; they sit in a resting state called G0 and may never leave it.

The cycle has checkpoints: quality control gates that halt progress if the DNA is damaged or incompletely copied. The protein p53 is the most famous of these gatekeepers, halting the cycle for repair and, if repair fails, triggering the cell to destroy itself. It is disabled in roughly half of all human cancers (Chapter 24), which is exactly what you would expect from a gate that stops damaged cells from multiplying.

The Hayflick limit and telomeres

Normal human cells grown in a dish divide only about 40 to 60 times and then stop permanently, a state called senescence. Leonard Hayflick demonstrated this in 1961, overturning a widespread belief that cells were immortal in culture.

The mechanism is telomeres: repetitive DNA caps on chromosome ends that shorten slightly with each division because the copying machinery cannot quite finish the end. When they get too short, the cell stops dividing. Telomeres are a division counter.

Two escapes from this exist, and both are consequential:

• Stem cells and germ cells express telomerase, an enzyme that rebuilds the caps, which is how they keep dividing for a lifetime.
• Most cancers reactivate telomerase, which is one of the required steps toward becoming a clinically significant tumour (Chapter 24).

Senescent cells do not simply sit quietly. They accumulate with age and secrete inflammatory signals that damage the tissue around them, which connects this section to Chapter 11 and to the drugs being developed to clear them.

Apoptosis: the cell suicide programme

Apoptosis is orderly, programmed cell death. The cell shrinks, chops up its own DNA, packages itself into neat fragments, and is quietly eaten by neighbours without provoking inflammation. It is used constantly: to sculpt fingers from a webbed embryonic hand, to delete immune cells that would attack you (Chapter 13), to remove cells with irreparable DNA damage, and to keep tissue sizes stable. Something on the order of tens of billions of your cells do this every day.

Necrosis is the opposite: uncontrolled death from injury, in which the cell bursts and spills its contents, triggering inflammation. That is what happens in a heart attack, a burn, or a crush injury, and the inflammation it provokes causes much of the secondary damage.

The balance matters clinically in both directions:

• Too little apoptosis allows damaged cells to survive and accumulate. This is one of the hallmarks of cancer.
• Too much apoptosis removes cells that should have stayed, which contributes to neurodegeneration and to some forms of organ failure.

Stem cells

In short: Cells that can both renew themselves and produce specialised cells, and they are the reason renewable tissues are renewable.

A stem cell has two properties: it can divide to make more of itself, and it can produce daughter cells that specialise into a working cell type.

The adult multipotent ones do the everyday work of the turnover table above. Blood-forming stem cells in the marrow produce roughly 2 million red cells every second, plus all the white cells and platelets, for a lifetime. This is what a bone marrow transplant transfers (Chapter 26), and what gene therapy for sickle cell disease modifies (Chapter 48).

Don't be confused: most "stem cell treatments" sold commercially are not treatments. Legitimate, proven stem cell therapy is a short list: blood and marrow transplantation, some skin and corneal grafting, and a small number of approved gene therapies. Clinics selling injections of "stem cells" for arthritis, autism, ageing, or almost anything else are overwhelmingly selling an unproven product, sometimes a harmful one, and regulators in several countries have acted against them.

The four tissue types

In short: Every organ in your body is built from just four kinds of tissue arranged differently.

Epithelium deserves particular attention because so much of this book is about it. It is a barrier, so it separates inside from outside. It is exposed, so it meets carcinogens, acid, microbes, and mechanical stress. And it renews constantly, so it divides a great deal. Barrier, exposure, and division together are why carcinomas dominate the cancer chapters.

The basement membrane beneath epithelium matters more than its obscurity suggests: it is the line that defines whether a cancer is "in situ" (contained, curable by removal) or "invasive" (through the line, able to reach blood and lymphatic vessels and spread). One layer of protein, and it is the difference between two prognoses.

From tissue to organ

In short: An organ is a specific arrangement of the four tissues around a job, and its structure is dictated by the job.

Take the small intestine as a worked example. It needs to absorb nutrients, so it needs an enormous surface area, a blood supply to carry absorbed material away, muscle to move contents along, and nerves to coordinate it:

• Epithelium lines the inside, thrown into folds, then finger-like villi, then microscopic microvilli on each cell, multiplying surface area to roughly 30 square metres.
• Connective tissue underneath carries blood vessels and lymphatics to collect what is absorbed.
• Smooth muscle in two layers, circular and longitudinal, produces the wave of contraction that moves contents along.
• Nervous tissue forms its own network in the wall, containing more neurons than the spinal cord, which is why the gut coordinates digestion largely without instruction from the brain.

Every organ in the next four chapters is the same exercise: four tissue types, arranged to accomplish something.

Why this chapter matters for the rest of the book

In short: Six recurring disease mechanisms are all cell-level problems, and recognising them makes the disease chapters shorter.

1. A protein is wrong or missing because a gene is mutated: sickle cell, cystic fibrosis, haemophilia.
2. Cells divide when they should not, because the brakes failed: all cancer.
3. Cells die that cannot be replaced: heart attack, stroke, neurodegeneration, hearing loss.
4. Repair overshoots into scar: cirrhosis, pulmonary fibrosis, chronic kidney disease.
5. Cells accumulate something they cannot clear: amyloid in Alzheimer's, lipid in atherosclerosis, iron in haemochromatosis, misfolded protein in Parkinson's.
6. The immune system attacks its own cells: type 1 diabetes, rheumatoid arthritis, multiple sclerosis.

Almost every disease in this book is one of those six, or a combination.

Sources and notes

Standard cell biology (Alberts et al., Molecular Biology of the Cell; Lodish, Molecular Cell Biology). Turnover rates: Spalding et al., Cell, 2005 (carbon-14 dating of human cell ages) and Nature, 2008 (adipocyte turnover); Bergmann et al., Science, 2009, for cardiomyocyte renewal of roughly 1 percent per year at age 25 falling to 0.45 percent at 75; Sender and Milo, Nature Medicine, 2021, for whole-body cellular turnover mass. Hayflick limit: Hayflick and Moorhead, Experimental Cell Research, 1961. Telomerase: Greider and Blackburn, with Szostak awarded the 2009 Nobel Prize. Induced pluripotent stem cells: Takahashi and Yamanaka, Cell, 2006, Nobel Prize 2012. Small intestinal surface area of roughly 30 m²: Helander and Fändriks, Scandinavian Journal of Gastroenterology, 2014, correcting the frequently repeated "tennis court" figure. Cancer risk correlating with stem cell divisions: Tomasetti and Vogelstein, Science, 2015, with the qualifications noted in Chapter 24. Unproven commercial stem cell clinics: FDA and international regulator statements.

Open questions. Whether meaningful neurogenesis occurs in the adult human hippocampus is genuinely contested, with well-conducted studies reaching opposite conclusions. The extent of adult heart muscle renewal, and whether it can be therapeutically increased, is unresolved.

Next: the organs, starting with the pump, the bellows, and the fluid they move. 👉

The Heart, the Lungs, and the Blood

TL;DR. A fist-sized muscular pump moves about 5 litres of blood a minute at rest and up to five times that under load, pushing it through roughly 100,000 km of vessels and returning it every minute. The lungs present 70 to 100 square metres of surface, a membrane less than a thousandth of a millimetre thick, across which oxygen enters and carbon dioxide leaves 20,000 times a day. Both systems have enormous reserve and one shared vulnerability: their working parts do not regenerate. Heart muscle lost to a heart attack becomes scar, and alveoli destroyed by smoke are gone permanently. Everything you can do for these organs is protection and capacity-building, because there is no repair option.

Key takeaways

• The heart beats roughly 2.5 to 3 billion times in a lifetime without a single scheduled maintenance stop, and it is the only muscle that generates its own electrical rhythm.
• The heart feeds itself between beats. Coronary arteries fill during relaxation, not contraction, which is why a very fast heart rate can starve the heart muscle.
• Aerobic fitness (VO2max) is one of the strongest predictors of mortality ever measured, outperforming smoking, diabetes, and hypertension in some cohorts.
• The endothelium, a single-cell layer lining every vessel, weighs about a kilogram in total and is where atherosclerosis begins. It is a functioning organ, not a pipe surface.
• Lungs have no way to regrow alveoli. Emphysema is permanent, which is what makes stopping smoking urgent rather than merely advisable.
• Veins hold about two-thirds of your blood volume at any moment, acting as a reservoir the body can draw on.

The heart

In short: A 300-gram pump with its own electrical system and its own blood supply, which fills between beats and cannot repair itself.

What it is

Two pumps side by side in one organ. The right side takes oxygen-poor blood returning from the body and pushes it a short distance to the lungs at low pressure. The left side takes oxygen-rich blood from the lungs and pushes it to the entire body at high pressure, which is why its wall is roughly three times thicker.

Its own electricity

The heart does not need a nerve signal to beat. A cluster of specialised cells, the sinoatrial node, depolarises spontaneously 60 to 100 times a minute and sets the rhythm. The impulse spreads across the atria, pauses at the atrioventricular node (giving the atria time to finish emptying into the ventricles), then races down a fast conduction pathway so both ventricles contract almost simultaneously.

Nerves and hormones modulate this rate but do not create it. A heart removed from the body and supplied with oxygen and nutrients will keep beating, which is what makes transplantation possible.

Three things follow that appear later:

• Damage to the conduction pathway causes heart block, and a pacemaker substitutes for it.
• Chaotic atrial electrical activity is atrial fibrillation (Chapter 21).
• Chaotic ventricular activity is ventricular fibrillation, which pumps no blood at all, and a defibrillator works by depolarising the whole muscle at once so the sinoatrial node can restart in an orderly way.

It feeds itself between beats

The heart receives no oxygen from the blood passing through its chambers. It has its own supply: two coronary arteries branching off the aorta immediately above the aortic valve.

Crucially, they fill during diastole, the relaxation phase, because during contraction the muscle squeezes its own vessels shut. Two consequences:

• A very fast heart rate shortens diastole disproportionately, reducing the heart's own blood supply exactly when its demand is highest. This is why a racing heart can provoke chest pain in someone with narrowed coronaries, and part of why beta blockers help angina.
• Diastolic blood pressure matters for coronary perfusion, which is one reason very low diastolic pressure in someone with stiff arteries is not automatically good.

Reserve and durability

The heart's reserve is in its capacity to increase output, not in spare tissue. A fit person can raise cardiac output fivefold; an unfit or failing heart cannot, which is felt as breathlessness on exertion long before anything is felt at rest.

What ages it:

• Maximum heart rate falls by roughly 0.7 beats per minute per year, regardless of fitness. This is one of the few genuinely fixed declines.
• The ventricle stiffens, filling less easily, which contributes to heart failure with preserved ejection fraction (Chapter 21).
• Valves thicken and calcify, and aortic stenosis becomes common in the eighties.
• Heart muscle does not regenerate meaningfully. Turnover is around 1 percent a year in young adults and lower with age, so any muscle lost is permanently replaced by scar.

What damages it: high blood pressure (making it pump against resistance for decades), blocked coronary arteries, diabetes, smoking, excess alcohol (which can cause cardiomyopathy directly), untreated sleep apnoea, some chemotherapy drugs (anthracyclines cause cumulative, permanent damage), and viral myocarditis.

What protects it: everything in Chapter 63, with blood pressure control, not smoking, and aerobic exercise at the top.

The blood vessels

In short: Not plumbing but an active organ, whose single-cell lining is where cardiovascular disease begins.

The endothelium is the single layer of cells lining the entire system. Laid flat it would cover several thousand square metres and it weighs roughly a kilogram. It is not a passive surface:

• It releases nitric oxide, which relaxes the vessel wall. This is what exercise improves and what smoking, high glucose, and high LDL impair.
• It controls clotting, presenting a non-stick surface until it is injured, then doing the opposite.
• It controls what crosses into tissue, tightly in the brain (the blood-brain barrier) and loosely in the liver.
• Its injury is the first step in atherosclerosis (Chapter 21).

Arterial stiffening is the most consistent vascular change with age. Elastin fragments and is replaced by stiffer collagen, so the aorta stops cushioning each beat. Systolic pressure rises, diastolic often falls, and the gap between them widens. It is not entirely inevitable: populations with low sodium intake and high physical activity show far less of it.

The blood

In short: Five litres carrying oxygen, fuel, heat, hormones, immune cells, and the clotting system, with the marrow replacing 2 million red cells every second.

Red cells are unusual. They eject their nucleus and mitochondria during development, which frees space for haemoglobin and means they cannot repair themselves or use oxygen for their own metabolism. They survive about 120 days of being squeezed through capillaries narrower than they are, then are removed by the spleen and liver, and their iron is recycled. That 120-day lifespan is why HbA1c averages three months of blood glucose.

Haemoglobin is four protein chains, each holding an iron atom that binds one oxygen molecule. Its binding is cooperative: picking up the first oxygen makes the next easier, which produces an S-shaped curve that loads oxygen efficiently in the lungs and unloads it efficiently in tissue. Warm, acidic, carbon-dioxide-rich conditions, exactly what an exercising muscle produces, shift the curve so haemoglobin releases more oxygen precisely where it is needed. It is an elegant piece of automatic control, and Chapter 48 is about what happens when one amino acid in it is wrong.

Clotting is a cascade: a dozen factors activating each other in sequence, so that a small trigger produces a large, fast response. It is deliberately positive feedback, restrained by an equally elaborate anticlotting system. Nearly every anticoagulant drug blocks one specific step in that cascade (Chapter 56).

The lungs

In short: A hundred square metres of gas-exchange surface with a self-cleaning conveyor belt, and the alveoli never grow back.

What it is

Air travels down the trachea, through roughly 23 generations of branching tubes, and ends in alveoli: 300 to 500 million tiny sacs with a combined surface area of 70 to 100 square metres, about half a tennis court, wrapped in capillaries.

Breathing is driven by the diaphragm, a dome of muscle that flattens on contraction, enlarging the chest and drawing air in by suction. Exhalation at rest is passive: the lungs recoil elastically. That elastic recoil is exactly what emphysema destroys (Chapter 45), which is why COPD patients struggle to breathe out.

Surfactant is a detergent-like substance lining the alveoli that stops them collapsing under surface tension. Premature babies have not yet made enough, which causes respiratory distress syndrome, and giving artificial surfactant is one of the great advances in neonatal care.

The self-cleaning system

You inhale roughly 11,000 litres of air a day, carrying dust, pollen, microbes, and pollution. The airways handle it with a mucociliary escalator: a layer of sticky mucus riding on a bed of cilia that beat 1,000 or more times a minute, moving the sheet upward at roughly a centimetre a minute to the throat, where it is swallowed.

Cigarette smoke paralyses and then destroys those cilia. The escalator stops. Mucus pools, bacteria colonise, and the smoker coughs because coughing is now the only clearance mechanism left. That single fact explains chronic bronchitis, why smokers get pneumonia more often, and why a "smoker's cough" is not benign (Chapter 45).

Reserve and durability

Lung reserve is large: you can lose an entire lung and function normally at rest.

What ages it:

• FEV1, the volume you can force out in one second, peaks in the mid-twenties and declines by roughly 25 to 30 mL a year afterwards. In a susceptible smoker it declines two to three times faster, which is the Fletcher-Peto curve in Chapter 45.
• The chest wall stiffens and respiratory muscles weaken.
• Alveolar walls are lost gradually even without smoking, reducing surface area.

The two-sided lesson from that decline curve: the height of your peak matters as much as the rate of your decline. Childhood respiratory infections, prematurity, and maternal smoking all lower the peak, so a person can reach the disease threshold decades earlier with an entirely normal rate of decline.

What damages the lungs: tobacco smoke above all, then household smoke from cooking fires (affecting billions of people, mostly women and children), outdoor particulate pollution, occupational dusts (silica, coal, asbestos), and repeated severe infection including tuberculosis.

What protects them: not smoking and not being around smoke, clean cooking fuel, ventilation and filtration indoors, respiratory protection at work, vaccination against influenza, pneumococcus, COVID-19, and RSV, and aerobic exercise, which does not increase lung capacity much but improves everything downstream of it.

Reading the warning signs

In short: Six symptoms from these systems mean urgent assessment, and knowing which is which is genuinely useful.

The numbers worth knowing about yourself

In short: Five measurements describe the state of these systems better than any symptom.

Don't be confused: a low resting heart rate is usually good and occasionally a problem. In a fit person, a rate in the 40s or 50s reflects a strong heart ejecting more per beat. In someone unfit, or with dizziness, fainting, or fatigue, the same number can mean the conduction system is failing. The number alone does not distinguish them; symptoms do.

Sources and notes

Cardiac and respiratory physiology figures are standard (Guyton and Hall, Textbook of Medical Physiology; West, Respiratory Physiology: The Essentials). Alveolar number and surface area estimates: Ochs et al., American Journal of Respiratory and Critical Care Medicine, 2004 (approximately 480 million alveoli on average). Cardiomyocyte turnover: Bergmann et al., Science, 2009. Maximum heart rate decline with age: Tanaka, Monahan, and Seals, JACC, 2001. FEV1 decline: Fletcher and Peto, BMJ, 1977. Cardiorespiratory fitness and mortality: Mandsager et al., JAMA Network Open, 2018, which found no upper limit to the benefit of higher fitness in a cohort of over 122,000. Endothelial mass and surface estimates vary by method and are given as approximations. Mucociliary clearance rates: standard respiratory physiology. Sex differences in heart attack presentation: multiple registry analyses cited in Chapter 21.

Open questions. Whether adult human cardiomyocyte renewal can be therapeutically increased is unresolved. The extent to which arterial stiffening is an inevitable consequence of ageing versus a consequence of lifetime sodium intake and inactivity continues to be debated.

Next: the organs that process everything you swallow and everything your cells discard. 👉

The Gut, the Liver, and the Kidneys

TL;DR. A nine-metre tube runs through you, technically outside your body, lined with about 30 square metres of absorbing surface and staffed by roughly 38 trillion bacteria. Everything absorbed from it goes first to the liver, a 1.5 kilogram chemical plant running several hundred processes, which is also the only major organ that can regrow itself. What the liver cannot dispose of goes to the kidneys, which filter your entire blood volume roughly 30 times a day, producing 180 litres of filtrate and reclaiming almost all of it. All three organs have huge reserve, all three are silent while failing, and all three are damaged by the same short list: alcohol, obesity, uncontrolled diabetes and blood pressure, and a handful of common painkillers.

Key takeaways

• The gut has its own nervous system with more neurons than the spinal cord, and it runs digestion largely without instruction from the brain.
• All blood from the intestines goes to the liver first. That single piece of plumbing explains first-pass drug metabolism, why colorectal cancer spreads to the liver, and why liver scarring causes bleeding in the oesophagus.
• The liver regenerates from as little as a quarter of its mass, which is why living donation works, and it scars instead when injury is repeated rather than single.
• You are born with all the nephrons you will ever have, roughly a million per kidney, and lose about 1 percent a year after your thirties.
• The kidneys use about 20 to 25 percent of your cardiac output despite being 0.5 percent of body weight, which is why they are so vulnerable to anything that reduces blood flow.
• Regular NSAIDs, alcohol, and untreated diabetes and hypertension damage all three of these organs, which makes the protective list short and repetitive.

The digestive tract

In short: A nine-metre tube that is topologically outside you, with each section doing one job in sequence.

The journey, section by section

The small intestine's surface area is multiplied three times over: circular folds, then finger-like villi, then microvilli on each cell, reaching roughly 30 square metres. The frequently repeated claim that it covers a tennis court comes from older estimates and was corrected downward in 2014, which does not make 30 square metres any less remarkable for something that fits in an abdomen.

The stomach's acid is a defence as much as a digestive tool. At pH 1.5 it kills most swallowed microbes, which is why reduced acid (from proton pump inhibitors, or from age) raises the risk of gut infections including C. difficile and Salmonella (Chapter 52). The stomach protects itself with a mucus and bicarbonate layer maintained by prostaglandins, and blocking those prostaglandins is exactly what NSAIDs do, which is the mechanism of drug-induced ulcers.

The gut's own brain

The enteric nervous system contains 200 to 600 million neurons in the gut wall, more than the spinal cord. It coordinates peristalsis, secretion, and blood flow autonomously; a section of intestine removed from the body will still produce coordinated waves.

It communicates with the brain constantly through the vagus nerve, mostly in the upward direction: roughly 80 to 90 percent of vagal fibres carry information from gut to brain rather than the reverse. The gut also produces around 90 to 95 percent of the body's serotonin, though that serotonin acts locally on gut motility and does not cross into the brain.

This is the real, well-established core of the "gut-brain axis." It is also the part that gets stretched furthest in popular coverage: the connection is genuine, and specific claims that a particular supplement or diet changes mood through the microbiome are mostly not established.

The microbiome

Roughly 38 trillion bacteria, overwhelmingly in the colon, weighing perhaps 200 grams in total. They do several things you cannot:

• Ferment fibre into short-chain fatty acids, which directly feed the cells lining the colon.
• Synthesise some vitamin K and several B vitamins.
• Occupy the space, competing with pathogens. Wipe them out with broad-spectrum antibiotics and C. difficile can take over (Chapter 36).
• Train the immune system, particularly in early life.

What is genuinely established: antibiotics disrupt it, diet changes it within days, low diversity is associated with several diseases, and faecal transplantation cures recurrent C. difficile infection. What is not established: that manipulating it treats obesity, depression, autism, or most of what commercial probiotics are sold for.

Care and durability

What damages the gut: smoking (raises Crohn's risk and ulcer risk), regular NSAIDs, excess alcohol, very low fibre intake, and chronic H. pylori infection.

What protects it: fibre from varied plant sources, not smoking, sensible NSAID use, and treating H. pylori when found. Bowel cancer screening from 45 is the single highest-value intervention for this organ (Chapter 25).

Warning signs that need investigation rather than reassurance: blood in stool, a persistent change in bowel habit, difficulty swallowing, unintentional weight loss, persistent vomiting, and iron-deficiency anaemia in anyone over about 40.

The liver

In short: A 1.5 kg chemical plant running several hundred processes, with a dual blood supply, no pain fibres, and a unique ability to regrow.

What it does

The number "500 functions" gets quoted often. The ones that matter clinically group into six:

The plumbing that explains everything

The liver has two blood supplies. The hepatic artery brings oxygenated blood. The portal vein brings everything absorbed from the intestines, roughly 75 percent of the liver's blood flow.

That arrangement is deliberate: the liver screens everything you absorb before it reaches the rest of you. Four consequences run through this book:

1. First-pass metabolism. A swallowed drug passes through the liver before reaching the circulation, and much of the dose may be destroyed there. This is why some drugs must be injected or placed under the tongue (Chapter 16).
2. Colorectal cancer spreads to the liver first, because that is where its venous drainage goes (Chapter 24).
3. A scarred liver obstructs that flow, so pressure backs up in the portal vein, forcing blood through small veins in the oesophagus that were never built for it. Those veins rupture, which is one of the ways cirrhosis kills (Chapter 32).
4. Gut bacteria products bypass the filter in cirrhosis and reach the brain, causing confusion.

Regeneration, and its limit

The liver is the only major internal organ that regrows. Remove up to 70 percent and the remainder expands back to functional mass within weeks. This is what makes living-donor liver transplantation possible.

But regeneration and repair are not the same thing. A single large insult is regenerated. A repeated small insult (daily alcohol, years of fat accumulation, chronic viral infection) triggers scarring instead, because the injury never stops long enough for orderly regrowth. Collagen accumulates, the architecture is destroyed, and the result is cirrhosis, which is largely irreversible.

The liver has no pain fibres in its substance. It has them only in its outer capsule. This is the main reason liver disease is silent until it is advanced: there is no early warning sensation available.

Care and durability

What damages it: alcohol, excess body fat (metabolic fatty liver disease now affects a quarter to a third of adults), chronic hepatitis B and C, paracetamol/acetaminophen overdose (the leading cause of acute liver failure in several countries), certain herbal and bodybuilding supplements, and iron or copper overload.

What protects it: limiting alcohol, maintaining a reasonable weight, hepatitis B vaccination, hepatitis C testing and cure, respecting the paracetamol maximum dose (and knowing that many combination cold remedies contain it), and scepticism about supplements, which are a growing cause of drug-induced liver injury.

Warning signs: jaundice, dark urine with pale stools, persistent itching, unexplained easy bruising, abdominal swelling, and confusion in someone with known liver disease.

The pancreas and gallbladder

In short: Two accessory organs, one making digestive enzymes and insulin, the other storing bile.

The pancreas does two unrelated jobs in one organ:

• Exocrine (about 95 percent of its mass): makes digestive enzymes for protein, fat, and starch, delivered into the duodenum. Loss of this causes fatty, floating stools and weight loss, treated with enzyme capsules taken with meals.
• Endocrine (the islets, about 1 to 2 percent): makes insulin and glucagon (Chapter 18).

It sits deep in the abdomen, which is why pancreatic cancer is usually advanced before it is found (Chapter 25). Its enzymes activating inside it rather than in the intestine is pancreatitis, and the two dominant causes are gallstones and alcohol.

The gallbladder stores and concentrates bile between meals and squeezes it out when fat arrives. Bile is not a digestive enzyme; it is a detergent that breaks fat into droplets small enough for enzymes to work on. Cholesterol precipitating out of concentrated bile forms gallstones, which are present in 10 to 15 percent of adults and silent in most of them (Chapter 52). You can live normally without a gallbladder; bile simply drips continuously instead of being released in a bolus.

The kidneys

In short: Two fist-sized filters that process the entire blood volume 30 times a day and quietly run blood pressure, red cell production, and bone chemistry as well.

The numbers

The five jobs

The kidney is usually described as a filter, which undersells it by four jobs:

1. Filtration and reabsorption. Blood is forced through a tuft of capillaries; the filtrate then travels down a tubule where the body reclaims water, glucose, sodium, and everything else worth keeping, with exquisite precision.
2. Blood pressure. By controlling how much sodium and water is retained, and by releasing renin to activate the cascade described in Chapter 20.
3. Red blood cell production. Releases erythropoietin when it senses low oxygen, telling the marrow to make more red cells. Failure of this is why kidney disease causes anaemia.
4. Bone and calcium. Performs the final activation step of vitamin D. Failure of this is why kidney disease causes bone disease and vascular calcification.
5. Acid-base balance. Excretes acid and regenerates bicarbonate, holding blood pH in a range of 0.1 units.

Every one of those jobs shows up as a symptom when kidneys fail, which is why kidney failure produces such a scattered picture (Chapter 23).

Why they are so vulnerable

Two structural facts:

• They receive a quarter of your cardiac output, so anything toxic in the blood arrives at the kidney in quantity, and anything that reduces blood flow (dehydration, low blood pressure, blood loss) starves them quickly.
• Nephrons do not regenerate. Lost ones are gone. The survivors compensate by filtering harder, and that overwork scars them in turn, which is the self-accelerating loop in Chapter 23.

Care and durability

What damages them: diabetes and high blood pressure (together the cause of most kidney disease worldwide), regular NSAID use, dehydration, repeated urinary infections and obstruction, some antibiotics and contrast agents, and, in specific populations, heat stress during heavy agricultural labour.

What protects them: controlling blood pressure and blood sugar, avoiding routine NSAIDs, staying hydrated, and knowing your sick day rules: during vomiting, diarrhoea, or fever, several common drugs (ACE inhibitors, ARBs, diuretics, SGLT2 inhibitors, metformin, NSAIDs) should usually be paused, because dehydration plus those drugs is how a stomach bug becomes acute kidney injury. Ask for that list in writing if you take any of them.

How to check: a blood creatinine (giving eGFR) and a urine albumin-to-creatinine ratio. Both are cheap, and both are needed, because either alone misses people. Annually if you have diabetes, hypertension, or cardiovascular disease.

The bladder and urinary tract

In short: A storage tank with a socially trained release mechanism, and the most common site of bacterial infection in the body.

The bladder holds 400 to 600 mL, signalling fullness at around 150 to 250 mL. Continence depends on a coordinated arrangement of smooth muscle, two sphincters, and pelvic floor muscle, under control from the spinal cord and brain.

Urinary tract infections are the commonest bacterial infection in women, mostly from gut bacteria travelling the short distance up the urethra (Chapter 59). Incontinence affects a large minority of women and men over 60 and is substantially treatable, yet is one of the most under-reported conditions in medicine because of embarrassment.

Kidney stones form when urine is concentrated enough for crystals to precipitate. The single most effective prevention is drinking enough that urine stays pale, which reduces recurrence substantially.

The pattern across all three systems

In short: The same short list of insults damages gut, liver, and kidneys, and the same short list protects them.

That repetition is not lazy writing. It is the actual finding: a small number of exposures damage most of the body's processing organs, which is why the practical chapter at the end of this book is short.

Sources and notes

Gastrointestinal, hepatic, and renal physiology are standard (Guyton and Hall; Boron and Boulpaep, Medical Physiology). Small intestinal surface area of roughly 30 to 32 m²: Helander and Fändriks, Scandinavian Journal of Gastroenterology, 2014. Enteric neuron counts and vagal afferent proportions: Furness, Nature Reviews Gastroenterology & Hepatology, 2012. Microbiome mass and cell counts: Sender, Fuchs, and Milo, 2016. Liver regeneration thresholds are established in living-donor transplantation practice. Nephron number and age-related loss: Denic et al., NEJM, 2017, and related morphometric studies. Kidney share of cardiac output and daily filtration volume are standard physiology. Gallstone prevalence of 10 to 15 percent in Western adults: epidemiological surveys.

Open questions. How much of the gut-brain axis translates into clinically useful intervention is unresolved, and most commercial claims about it outrun the evidence. Whether nephron number at birth, which varies severalfold between individuals, explains part of the variation in adult kidney disease risk is an active question.

Next: the organ that makes you a person, and the four senses that feed it. 👉

The Brain, the Nerves, and the Senses

TL;DR. Your brain is about 1.4 kilograms, roughly 2 percent of your body weight, and it consumes about 20 percent of your oxygen and glucose continuously. It contains around 86 billion neurons connected at perhaps 100 trillion synapses, and it cannot store fuel, which is why four minutes without blood flow causes permanent damage. Nerves carry signals to and from it at up to 120 metres per second along insulated fibres. The senses feed it: about 120 million light detectors in each eye, and roughly 15,000 sound-detecting hair cells in each ear that, in humans, never grow back. The recurring theme of this chapter is that the nervous system's parts are the least replaceable in the body, and that its most important property, the ability to rewire, is the reason recovery from damage is possible at all.

Key takeaways

• Four minutes. That is roughly how long brain tissue survives without blood flow, and it is why stroke and cardiac arrest are measured in minutes rather than hours.
• Neurons do not meaningfully regenerate, but connections do rewire. Neuroplasticity is the mechanism behind all rehabilitation, and it is use-dependent, which is why rehabilitation is a treatment rather than a comfort.
• Myelin insulation speeds nerve signals roughly a hundredfold. Losing it is multiple sclerosis (Chapter 40).
• Inner ear hair cells never regenerate in humans. Noise damage is permanent and entirely preventable, and untreated hearing loss is the single largest modifiable midlife risk factor for dementia.
• The lens of the eye never sheds cells. It only adds them, which is why everyone becomes long-sighted after about 45 and why cataract is essentially universal if you live long enough.
• The blood-brain barrier protects the brain and blocks most drugs, which is why brain infections and brain tumours are so much harder to treat than the same problems elsewhere.

The brain

In short: An organ that cannot store fuel, cannot replace its cells, and compensates by being able to rewire the connections between them.

The numbers

The fuel problem is the central vulnerability. Muscle stores glycogen; the liver stores glycogen; the brain stores almost nothing. It requires continuous delivery, which is why an interruption is catastrophic within minutes and why Chapter 22 is organised entirely around speed. During prolonged fasting the brain can shift partly to ketones made by the liver from fat, which reduces but never eliminates its glucose requirement.

The main parts, and what they do

How a neuron works

A neuron receives signals on branching dendrites, sums them at its cell body, and if the total crosses a threshold, fires an all-or-nothing electrical pulse down its axon. At the end, the pulse triggers release of a neurotransmitter across a tiny gap, the synapse, onto the next cell.

Three things follow:

• The signal is electrical within a cell and chemical between cells. Nearly every drug acting on the brain works at the chemical step: SSRIs block serotonin reuptake, opioids activate opioid receptors, benzodiazepines enhance GABA, antipsychotics block dopamine receptors.
• Excitation and inhibition must balance. Glutamate excites, GABA inhibits. Tip the balance toward excitation and neurons fire in synchrony, which is a seizure (Chapter 39).
• Myelin makes it fast. Support cells wrap axons in a fatty insulating sheath with gaps, so the impulse jumps between gaps rather than travelling continuously. This raises conduction speed from about 1 metre per second to up to 120. Stripping it is what multiple sclerosis does.

Plasticity: the one form of repair available

Adult neurons in the cortex are not replaced. What the brain can do instead is rewire: strengthen some synapses, weaken others, grow new dendritic branches, and recruit neighbouring regions to take over lost functions.

This is not a metaphor. It is the mechanism behind:

• Learning and memory, which physically consist of changed synaptic strength.
• Recovery after stroke, where surviving regions progressively take over. It is use-dependent, which is precisely why intensive, repetitive, task-specific rehabilitation works and passive rest does not (Chapter 22).
• Chronic pain, where repeated pain signalling sensitises the system until it fires without a stimulus (Chapter 40).
• Cognitive reserve, the observation that people with more education and more cognitively demanding lives sustain more Alzheimer pathology before showing symptoms (Chapter 37).

Plasticity is greatest in childhood and never disappears. It is the physiological basis for the claim, which is true, that it is never too late to start.

The blood-brain barrier

Brain capillaries are sealed with unusually tight junctions and wrapped by support cells, forming a barrier that admits oxygen, glucose, and small fat-soluble molecules while excluding most large or water-soluble ones, including most drugs and most immune cells.

The trade-off is stark. It protects the brain from circulating toxins and infections, and it is the main reason brain infections, brain tumours, and neurological diseases are so hard to treat: the drug cannot get in. Levodopa exists in its particular form precisely because dopamine cannot cross it and its precursor can (Chapter 38).

Care and durability

What ages the brain: total brain volume declines slowly from around the age of 30 to 40, with frontal and hippocampal regions affected earliest. Processing speed declines steadily. Crystallised knowledge (vocabulary, accumulated expertise) is largely preserved or improves into the seventies. So "getting slower" and "knowing less" are different, and only the first is typical.

What damages it: high blood pressure (the largest vascular contributor), smoking, diabetes, excess alcohol, repeated head injury, air pollution, untreated hearing loss, social isolation, poor sleep, and untreated depression. That list is the modifiable dementia risk factor list from Chapter 37, and it is strikingly similar to the cardiovascular list, which is not a coincidence.

What protects it: cardiovascular risk control, physical exercise (the most consistent protective association), cognitive and social engagement, treating hearing and vision loss, protecting the head, sleep, and treating depression.

Warning signs: sudden weakness, speech difficulty, or facial droop mean stroke and emergency care now. Progressive memory loss affecting daily function is different from ordinary forgetfulness and warrants assessment. New severe headache, especially sudden and maximal within seconds, is an emergency.

The peripheral nerves

In short: The cabling, which unlike the brain can regrow slowly, and which fails from the longest fibres inward.

Nerves outside the brain and spinal cord come in three functional types: motor (brain to muscle), sensory (body to brain), and autonomic (automatic control of organs, split into sympathetic "fight or flight" and parasympathetic "rest and digest").

Peripheral nerves can regenerate, unlike central ones, at roughly 1 mm per day. A cut nerve in the wrist may take months to a year to restore function, and often does so incompletely.

The longest fibres fail first, which is why diabetic and other neuropathies start in the toes and creep upward in a stocking pattern before appearing in the fingertips (Chapter 40). A neuron whose axon runs from the spinal cord to the foot has to maintain a cellular process a metre long, which is metabolically demanding and the first thing to fail when metabolism is impaired.

The autonomic nervous system deserves specific mention because it runs everything you do not think about: heart rate, blood pressure, digestion, sweating, pupil size, bladder function, sexual function. Damage to it (from diabetes, Parkinson's, or amyloidosis) causes dizziness on standing, gut paralysis, and bladder dysfunction, which are frequently misattributed.

The eye

In short: A camera whose film cannot be replaced and whose lens never stops thickening, which makes two age-related changes universal.

How it works

Light passes through the cornea (which does most of the focusing), through the pupil, through the lens (which fine-tunes focus), and lands on the retina, a sheet of neural tissue that is technically an outgrowth of the brain.

The retina contains two detector types:

Almost all of your detailed vision comes from the fovea, an area about 1.5 mm across. Everything else is periphery, which is why losing the macula (age-related macular degeneration) destroys reading and face recognition while leaving navigation intact, and why glaucoma, which destroys the periphery first, can advance a long way before it is noticed.

The two universal changes

Presbyopia. The lens grows throughout life, adding layers and never shedding cells, so it becomes progressively stiffer. By around 45 it can no longer change shape enough to focus close up. This happens to everyone. Reading glasses are not a failure of eye care; they are an inevitability of lens biology.

Cataract. The same never-replaced lens proteins accumulate damage over decades, and the lens clouds. It is essentially universal with sufficient age, and it is the leading cause of blindness worldwide, almost entirely because of unequal access to a 20-minute operation that replaces the lens and restores vision (Chapter 54).

Care and durability

What damages the eye: ultraviolet light (cataract, and eyelid cancers), smoking (a major risk factor for macular degeneration), diabetes (retinopathy, a leading cause of working-age blindness), raised eye pressure (glaucoma), and, in children, insufficient time outdoors, which is now the best-supported factor in the global rise of short-sightedness.

What protects it: sunglasses with UV protection, not smoking, glucose and blood pressure control, regular eye examinations (which detect glaucoma before any symptom), and, for children, roughly two hours a day outdoors, which trials show meaningfully reduces the onset of myopia.

Warning signs that are emergencies: sudden loss of vision, sudden onset of many new floaters or flashes (possible retinal detachment), a curtain across the visual field, sudden painful red eye with visual loss, and double vision of sudden onset.

The ear

In short: Fifteen thousand hair cells per ear that never grow back, which makes noise damage permanent and hearing protection a lifelong investment.

Sound reaches the eardrum, is amplified about twentyfold by three tiny bones (the smallest in the body), and enters the fluid-filled cochlea, a spiral in which different frequencies displace different positions along a membrane. Roughly 15,000 hair cells sit along it and convert mechanical displacement into nerve signals.

Those hair cells do not regenerate in mammals. Birds and fish regrow theirs, which is why the research field exists, and humans do not. Every one destroyed by noise, disease, or certain drugs is permanently gone.

Noise damage is cumulative and dose-dependent. Sustained exposure above roughly 85 decibels causes progressive loss, and the exposure limit halves for every 3 dB increase. Personal audio at high volume, power tools, live music, and industrial noise all contribute, and the damage is painless and invisible until enough cells are gone.

High frequencies go first, which is why the earliest symptom is not "everything is quieter" but "I can hear you but I cannot make out the words," particularly in a noisy room. Consonants carry most of the information in speech and they are high-frequency.

Two consequences worth stating plainly:

• Untreated hearing loss is the largest modifiable midlife risk factor for dementia (Chapter 37), and a randomised trial found hearing aids slowed cognitive decline in older adults at higher risk.
• The average person waits about a decade between noticing hearing difficulty and doing something about it, during which social withdrawal and its consequences accumulate.

The vestibular system shares the inner ear: three fluid-filled loops at right angles detecting rotation, plus two organs detecting linear acceleration and gravity. Its failure causes vertigo, and the commonest cause, benign paroxysmal positional vertigo, is produced by displaced crystals and cured in minutes by a positioning manoeuvre, which is one of the highest-value interventions in medicine relative to its cost.

Smell and taste

In short: The one sensory system that does regenerate, and an early warning indicator for two major diseases.

Smell receptors are neurons directly exposed to the outside air, and uniquely among human neurons they are replaced throughout life, roughly every one to two months.

Taste detects only five qualities (sweet, salty, sour, bitter, umami). Nearly everything people call taste is actually smell, arriving at the nose from the back of the mouth, which is why food is tasteless with a blocked nose.

Loss of smell is clinically informative out of proportion to its inconvenience:

• It appears years before motor symptoms in Parkinson's disease (Chapter 38) and early in Alzheimer's.
• It was a distinctive early feature of COVID-19.
• It causes real harm: undetected gas leaks and spoiled food, plus loss of appetite and a measurable effect on mood and quality of life that is routinely underestimated.

Sources and notes

Neuron count of about 86 billion: Azevedo, Herculano-Houzel et al., Journal of Comparative Neurology, 2009. Brain energy consumption, blood flow, and ischaemic tolerance are standard physiology (Kandel et al., Principles of Neural Science). Nerve conduction velocities and peripheral regeneration rate of approximately 1 mm/day are standard neurology. Photoreceptor counts: Curcio et al., Journal of Comparative Neurology, 1990. Cochlear hair cell numbers and the absence of mammalian regeneration are standard otology. Noise exposure limits: NIOSH and WHO occupational standards. Hearing aids and cognitive decline: the ACHIEVE trial, Lin et al., The Lancet, 2023. Time outdoors and myopia incidence: He et al., JAMA, 2015, and subsequent cluster-randomised trials. Olfactory neuron turnover and its role as a prodromal marker in Parkinson's: Doty, Nature Reviews Neurology, 2012.

Open questions. Whether adult humans produce meaningful numbers of new hippocampal neurons is genuinely contested, with careful studies reaching opposite conclusions. Restoring hair cell regeneration in mammals remains an unsolved research goal.

Next: the frame that holds you up, the covering that keeps the world out, and the glands that set the pace. 👉

Bones, Muscles, Skin, and Glands

TL;DR. Your skeleton is not scaffolding but living tissue that is completely rebuilt roughly every decade, and it doubles as a calcium bank and a blood cell factory. Your muscle is about a third of your body mass, is your largest glucose sink, and shrinks by 3 to 8 percent per decade after 30 unless you actively resist. Your skin is your largest organ, replaces itself monthly, and is the barrier that makes everything else possible. And a handful of glands weighing a few grams in total set your metabolic rate, your stress response, your growth, and your fertility using messages measured in billionths of a gram. The theme of this chapter is use it or lose it: bone and muscle are the two tissues in your body whose maintenance is directly commanded by mechanical load, which means the instruction to keep them is something you send yourself.

Key takeaways

• Bone is rebuilt continuously, with the whole skeleton replaced roughly every 10 years. It responds directly to load, which is why weightlessness and bed rest cause rapid loss.
• Peak bone mass is reached by the late twenties and everything afterwards is decline from that peak, which is why what you do at 20 matters at 75.
• Muscle is lost at 3 to 8 percent per decade after 30, accelerating after 60, and resistance training reverses a large part of it at any age, including in the nineties.
• Muscle is your largest glucose sink. Losing it worsens blood sugar control independently of anything else, which links this chapter directly to diabetes.
• Skin is a barrier, a thermostat, an immune organ, and a vitamin factory, and its outermost layer is entirely dead cells.
• Hormones work at concentrations of parts per billion, which is why a gland weighing 25 grams can control your whole metabolic rate.

Bone

In short: Living, constantly rebuilt tissue that responds to mechanical load, stores 99 percent of your calcium, and manufactures your blood.

What it actually is

Bone is roughly two-thirds mineral (calcium phosphate crystals, providing stiffness) and one-third protein (mostly collagen, providing tensile strength and preventing shattering). That composite is why bone is both hard and slightly flexible: mineral alone would be brittle like chalk, collagen alone would be rubbery.

Two forms:

• Cortical (compact) bone: the dense outer shell, about 80 percent of skeletal mass. Strong, slow to turn over.
• Trabecular (spongy) bone: an internal honeycomb, about 20 percent of mass but with far more surface area, and therefore metabolically much more active. It is lost first in osteoporosis, which is why the spine, hip, and wrist, which are trabecular-rich, are the classic fracture sites (Chapter 50).

It is rebuilt constantly

Two cell types work in opposition, continuously, at millions of microscopic sites:

• Osteoclasts dissolve old bone.
• Osteoblasts lay down new bone.

The whole skeleton is replaced roughly every ten years. Every osteoporosis drug acts on this balance: bisphosphonates and denosumab suppress the demolition crew, while teriparatide and romosozumab stimulate the building crew.

It responds directly to load

Bone adapts to the forces placed on it, laying down more material where stress is highest. The consequences are dramatic in both directions:

• Astronauts lose 1 to 2 percent of bone mass per month in weightlessness, which is one of the hardest problems for long-duration spaceflight.
• Extended bed rest causes comparable loss.
• The racquet arm of a lifelong tennis player has measurably thicker, denser bone than the other arm, in the same person with the same genes and the same diet.

Which is why the exercise recommendation for bone is specific: impact and resistance, not swimming or cycling. Bone responds to force, and the forces that matter are ground impact and muscle pulling on bone.

The timeline that matters

The practical implication is unusual in medicine: the most effective intervention for osteoporosis happens fifty years before the disease. Building a higher peak in adolescence and early adulthood, through nutrition, calcium, vitamin D, and above all weight-bearing activity, raises the ceiling from which the rest of life descends.

Bone's other two jobs

It is the calcium bank. Blood calcium must stay within a narrow range for nerves and muscle to work at all, and the body will dissolve bone without hesitation to maintain it. Parathyroid hormone commands that withdrawal. This is why chronic kidney disease, which disturbs calcium and phosphate handling, destroys bone (Chapter 23).

It makes your blood. Red marrow in the pelvis, spine, ribs, sternum, and the ends of long bones produces about 2 million red cells per second plus all white cells and platelets. This is what leukaemia disrupts, what chemotherapy suppresses, and what a bone marrow transplant replaces.

Care and durability

What damages bone: inactivity, smoking, excess alcohol, low body weight, corticosteroids (the commonest medical cause), oestrogen or testosterone deficiency, low calcium and vitamin D, coeliac disease, hyperthyroidism, and chronic kidney disease.

What protects it: weight-bearing and resistance exercise, adequate calcium (roughly 700 to 1,200 mg a day, ideally from food) and vitamin D, not smoking, moderate alcohol, and, after a fragility fracture, actual treatment, which most people who fracture never receive (Chapter 50).

Joints

In short: Cartilage-lined bearings lubricated by fluid, with no blood supply and therefore almost no capacity to repair.

A synovial joint has cartilage caps on the bone ends, a capsule enclosing a small volume of synovial fluid, and ligaments holding it together. The system is remarkable: the friction coefficient of healthy cartilage on cartilage is lower than ice on ice.

Cartilage has no blood supply and no nerves. It is fed by fluid moving in and out as the joint is loaded and unloaded, which is a genuine reason movement maintains joint health rather than consuming it. It also means:

• Cartilage heals very poorly. A defect does not fill in with cartilage; it fills with inferior fibrous tissue or not at all.
• Cartilage damage is painless, because there are no nerves in it. Osteoarthritis pain comes from the bone underneath, the joint lining, and the capsule, which is why X-ray appearance and pain correlate so weakly (Chapter 50).

Don't be confused: exercise does not wear out your joints. The intuition that joints are like car tyres with a fixed mileage is wrong. Cartilage requires cyclical loading to be nourished, and long-term studies of recreational runners find lower, not higher, rates of knee osteoarthritis than in sedentary people. What does damage joints is a specific injury, particularly a ligament rupture, obesity, and occupational loads involving repetitive kneeling and heavy lifting.

Muscle

In short: A third of your body mass, your largest glucose sink, and the tissue whose decline with age is both the most consequential and the most reversible.

Three kinds

How contraction works

Muscle fibres contain interleaved filaments of two proteins, actin and myosin. Myosin heads grab actin, pivot, release, and grab again, sliding the filaments past each other. Each cycle consumes one molecule of ATP. Calcium released inside the fibre is the trigger that exposes the binding sites.

Two clinically important facts fall out. Rigor mortis occurs because ATP is required to release the myosin head, so without it muscle locks. And calcium's central role is why disturbed blood calcium, potassium, and magnesium cause cramps, weakness, and, at the extreme, cardiac arrest.

Fibre types

Type II fibres are lost preferentially with age, which is why older people lose power (the ability to produce force quickly) faster than they lose strength, and why the ability to catch yourself when stumbling declines before the ability to lift a bag. This is a direct argument for including some fast, powerful movement in training rather than only slow strength work.

Sarcopenia: the decline that matters most

Muscle mass declines by roughly 3 to 8 percent per decade after 30, accelerating after 60, and strength declines faster than mass because fibre quality and neural drive also decline.

The consequences reach much further than appearance:

And it is substantially reversible. Resistance training produces measurable gains in strength and mass in people in their eighties and nineties, including in nursing home residents. The adaptation machinery does not switch off with age; it becomes less sensitive, which means the stimulus needs to be adequate and the protein intake needs to be a little higher (Chapter 2).

Care and durability

What damages muscle: disuse (the largest single factor, and bed rest can cost several percent of mass in a week), inadequate protein, corticosteroids, some cholesterol drugs in a minority, chronic inflammatory disease, and rapid weight loss without resistance training, which is a real concern with GLP-1 drugs (Chapter 66).

What protects it: resistance training two or three times a week, adequate protein spread across meals, and staying out of bed during illness wherever safely possible. Early mobilisation after surgery and in intensive care is a treatment for exactly this reason.

Skin

In short: The largest organ, a barrier that replaces itself monthly, and simultaneously a thermostat, an immune organ, a sensor, and a vitamin factory.

The structure

The outermost layer being dead is the point. Dead, keratin-packed cells embedded in lipid form a waterproof, chemically resistant sheet. When that lipid matrix is disrupted, as in eczema, water escapes and allergens get in, which is why moisturisers are a treatment rather than a cosmetic and why infant eczema is a route to food allergy (Chapter 46).

Skin covers about 1.5 to 2 square metres, weighs 3 to 5 kg, and the epidermis is completely replaced roughly every 4 to 6 weeks.

Its five jobs

1. Barrier, as above.
2. Temperature control. Blood vessels dilate to dump heat and constrict to conserve it, and 2 to 4 million sweat glands can produce over a litre an hour in the heat. Evaporation is the only mechanism that works when air temperature exceeds body temperature, which is why humidity makes heat dangerous (Chapter 60).
3. Sensation. Distinct receptors for light touch, pressure, vibration, temperature, and pain.
4. Immunity. Resident immune cells sample everything crossing the surface, which is why skin is where many allergies begin.
5. Vitamin D synthesis. UVB converts a cholesterol derivative in the skin into vitamin D precursor. This is why deficiency tracks latitude, season, skin pigmentation, clothing, and indoor living, and why supplementation is recommended in many countries during winter.

Care and durability

What ages skin: ultraviolet light does most of it. Compare sun-exposed forearm skin to skin on the same person's buttock at 70 and the difference is almost entirely UV, not time. Photoageing degrades collagen and elastin, producing wrinkling, thinning, irregular pigmentation, and, cumulatively, skin cancer. Smoking adds substantially.

What protects it: sun protection (shade, clothing, and sunscreen, applied at the quantity people almost never actually use), not smoking, and treating the barrier with emollients if it is compromised.

What to watch: a new or changing mole, particularly one that is asymmetric, irregularly bordered, multi-coloured, growing, or simply different from your others; a sore that does not heal in a month; and any lesion that bleeds repeatedly (Chapter 55).

The endocrine glands

In short: A few dozen grams of tissue producing messages at parts-per-billion concentrations that set your metabolic rate, stress response, growth, and fertility.

A hormone is a chemical message released into the blood by one tissue to act on another. The system is slow compared with nerves (seconds to hours rather than milliseconds) and it reaches everywhere at once. Because receptors are extraordinarily sensitive, hormones work at concentrations of nanograms per litre.

Note the last two rows: fat and bone are endocrine organs, which was not recognised until relatively recently and which reframes both obesity and kidney bone disease.

Nearly all of it runs on negative feedback through three levels: hypothalamus instructs pituitary, pituitary instructs gland, gland's hormone switches off both levels above. That is why two blood tests can localise a problem, and it is developed properly in Chapter 51.

The immune organs

In short: A distributed organ with no single location, and one part of it deliberately shrinks away after childhood.

The immune system's tissues are scattered by design:

• Bone marrow: makes all immune cells.
• Thymus: where T cells learn not to attack you. It is largest in childhood and progressively replaced by fat from adolescence onward, which is one of the reasons immunity declines with age.
• Spleen: filters blood, removes old red cells, and clears encapsulated bacteria. Losing it, whether surgically or functionally as in sickle cell disease, leaves lifelong vulnerability to specific infections (Chapter 48).
• Lymph nodes: several hundred filtering stations where immune cells meet what has been collected from tissue. Swollen nodes mean the local station is working.
• Mucosal tissue in gut, airways, and elsewhere, which is where most immune activity actually happens.

The lymphatic system doubles as drainage: fluid leaking out of capillaries into tissue is collected and returned to the bloodstream. When that drainage is blocked, typically after lymph node surgery or radiotherapy, the result is lymphoedema, permanent swelling of a limb, which is one of the reasons cancer surgery has moved toward removing fewer nodes (Chapter 26).

The reproductive organs

In short: Organs whose hormones affect bone, muscle, mood, and cardiovascular risk throughout life, not only fertility.

The reproductive systems matter in a general health book mostly for what their hormones do elsewhere.

Oestrogen maintains bone density (which is why loss at menopause causes rapid bone loss), affects blood vessel function, influences fat distribution, and contributes to the lower rate of cardiovascular disease in women before menopause.

Testosterone maintains muscle and bone mass, red cell production, libido, and mood. It declines gradually with age, roughly 1 percent a year after 30 to 40, which is a far more gradual change than menopause and is not equivalent to it.

Two things worth stating because they are commonly misunderstood. Women are born with all the eggs they will ever have, and their number and quality decline from birth, which is the basis of age-related fertility decline and of the rise in chromosomal abnormalities with maternal age. Men produce sperm continuously, roughly 1,500 per second, with a production cycle of about two to three months, which is why factors affecting sperm quality often reverse within a few months.

The unifying principle of this chapter

In short: Bone and muscle are maintained on demand, so the demand has to come from you.

Almost every organ in the preceding chapters maintains itself automatically. Bone and muscle do not. They are expensive tissues, and the body will happily dismantle them if they are not being used, because in evolutionary terms carrying unused muscle and bone is a waste of calories.

That is the biological reason behind advice that otherwise sounds like exhortation:

• Bone is maintained by impact and by muscle pulling on it.
• Muscle is maintained by being asked to produce near-maximal force.
• Neither responds to intention, only to load.

And it is why disuse is not neutral. Two weeks of bed rest in an older adult can cost several percent of leg muscle mass, which may take months to rebuild, and some of which is never rebuilt. That is the physiological argument for getting people out of hospital beds, and for not treating "rest" as automatically therapeutic.

Sources and notes

Standard physiology and anatomy references (Guyton and Hall; Marieb and Hoehn). Skeletal turnover of roughly 10 percent a year: bone remodelling literature. Peak bone mass timing and adolescent accrual: Bailey et al. and NIH consensus statements. Astronaut bone loss of 1 to 2 percent a month: NASA and spaceflight physiology studies. Tennis player bone asymmetry: Jones et al., Journal of Bone and Joint Surgery, 1977, and later replications. Sarcopenia rates of 3 to 8 percent per decade: standard geriatrics literature (Volpi, Nazemi, and Fujita, Current Opinion in Clinical Nutrition, 2004). Resistance training in the very old: Fiatarone et al., JAMA, 1990, and NEJM, 1994. Grip strength and mortality: Leong et al., The Lancet, 2015 (PURE study). Running and knee osteoarthritis: Alentorn-Geli et al., JOSPT, 2017, meta-analysis. Skin surface area, epidermal turnover, and sweat rates are standard dermatology and physiology. Thymic involution: standard immunology. Testosterone decline with age: Harman et al., Journal of Clinical Endocrinology & Metabolism, 2001 (Baltimore Longitudinal Study).

Open questions. How much of age-related muscle loss is inevitable biology versus accumulated disuse is genuinely debated, with masters athlete data suggesting the disuse component is larger than once assumed. Whether cartilage can be regenerated therapeutically remains unsolved.

Next: what happens to all of this when you eat. 👉

What Food Actually Does

TL;DR. Food does four things: supplies energy, supplies building material, supplies the small parts (vitamins and minerals) that let the machinery run, and feeds the bacteria in your colon. Most nutrition arguments are about the first, most nutrition failures are about the third, and the fourth was ignored for a century. The honest state of the evidence is that dietary patterns beat individual nutrients, that the strongest findings are unglamorous (more fibre, more plants, less ultra-processed food, less sugary drink, less alcohol), and that almost every confident, specific claim you have heard about a single food is weaker than it sounds. This chapter explains what actually happens inside you when you eat, so you can judge the claims yourself.

Key takeaways

• Your resting metabolism is 60 to 70 percent of your daily energy use. Exercise is a smaller share than people assume, which is why exercise is excellent for health and a weak tool for weight loss on its own.
• Protein costs more to process than it delivers in appetite, keeps you full longer, and preserves muscle during weight loss. It is the macronutrient with the clearest practical case.
• Fibre is the most consistently under-eaten thing in modern diets, at roughly half the recommended intake in most countries, and it is among the most consistently protective.
• The same calories behave differently depending on what they arrive with. Fibre, protein, and fat all flatten the glucose response to the same amount of starch.
• In a tightly controlled trial, people ate about 500 more calories a day on an ultra-processed diet than on a minimally processed one matched for sugar, fat, fibre, and macronutrients.
• Alcohol has no safe threshold for cancer risk, and the apparent heart benefit of moderate drinking has largely dissolved under better methods.

Where your energy actually goes

In short: Most of your daily calorie burn happens while you are doing nothing, which is why diet dominates weight and exercise dominates health.

Two consequences follow immediately.

Exercise is a poor weight-loss tool and an excellent health tool. A 45-minute run might burn 400 calories, which is a large muffin. That is why exercise trials for weight loss consistently disappoint, and why the same trials show substantial improvements in blood pressure, insulin sensitivity, cardiorespiratory fitness, and mortality. Those are different outcomes and they should not be conflated.

The thermic effect differs by macronutrient, which is a genuine, if modest, reason protein helps with weight management:

So 100 calories of protein delivers roughly 75 usable calories, while 100 calories of fat delivers about 98. It is not a large enough effect to build a diet on, and it is real.

Protein

In short: The macronutrient with the clearest practical case, and the one most people over 60 do not get enough of.

Chapter 2 explains what a protein and an amino acid are. What matters practically:

How much. The official minimum is 0.8 g per kg of body weight per day, set to prevent deficiency rather than to optimise anything.

Why more with age. Older muscle shows anabolic resistance: it needs a bigger protein dose to trigger the same amount of rebuilding. Combined with typically lower intake and less activity, this is a substantial contributor to the sarcopenia in Chapter 7.

Distribution matters. Roughly 25 to 40 g per meal, three or four times a day, stimulates muscle protein synthesis better than the same total concentrated at dinner, which is how most people actually eat.

What 30 g of protein looks like: 120 g chicken breast, 150 g salmon, 4 large eggs, 170 g Greek yoghurt plus a handful of nuts, 200 g firm tofu, 250 g cooked lentils, 85 g hard cheese, or a standard scoop and a half of whey powder.

Plant versus animal. Animal proteins contain all nine essential amino acids in useful proportions. Individual plant foods are usually low in one or two, and eating a variety across the day resolves it completely. Plant-based eaters need slightly more total protein, since digestibility is somewhat lower, and need a reliable B12 source, which is not optional.

Carbohydrate

In short: The fastest fuel, the one with no minimum requirement, and the one whose effect depends far more on what it arrives with than on how much of it there is.

Carbohydrate is sugars in chains. What determines its effect on you is how quickly it becomes glucose in your blood.

What flattens the glucose response to the same amount of starch:

This is most of why "whole foods" and "refined foods" produce different responses. It is not a mystical property; it is physical structure and what else is on the plate.

Fibre deserves its own paragraph because it is the single most under-eaten component of modern diets. Recommended intake is 25 to 30 g a day; typical intake in most high-income countries is 12 to 18 g. A large meta-analysis found that people in the highest fibre intake groups had roughly 15 to 30 percent lower all-cause and cardiovascular mortality and lower rates of colorectal cancer, type 2 diabetes, and stroke, with a dose-response relationship. Almost no other single dietary change has that breadth of association.

Practically: legumes, whole grains, vegetables, fruit with skins, nuts and seeds. Increase it gradually, because a sudden jump causes bloating while gut bacteria adjust.

Sugar. The specific problems are that liquid sugar produces almost no satiety, that added sugar displaces more useful food, and that it causes dental caries directly (Chapter 54). Sugar-sweetened drinks are the most consistently implicated single dietary item across the obesity, diabetes, and dental literature.

Fat

In short: Essential for membranes, hormones, and four vitamins, and the type matters more than the total.

Total fat intake turned out to matter far less than the low-fat era assumed. What has held up is the type, and specifically what a given fat replaces in the diet:

Two fatty acids are essential: linoleic acid (omega-6) and alpha-linolenic acid (omega-3). The long-chain omega-3s EPA and DHA are made from the latter inefficiently, which is why oily fish (salmon, mackerel, sardines, herring) is the reliable source, and algae-derived supplements are the plant-based equivalent.

Vitamins and minerals: the deficiencies that actually happen

In short: Five deficiencies account for most of the real-world burden, and none of them are what supplement marketing focuses on.

Everything else is mostly noise. Large randomised trials of multivitamins, antioxidants, vitamin D in unselected populations, and fish oil in primary prevention have repeatedly found no benefit, and several found harm (Chapter 63). Correcting a real deficiency works; topping up someone who is replete does not.

Ultra-processed food

In short: The most interesting recent finding in nutrition, with a genuine causal experiment behind it and an unresolved mechanism.

The NOVA classification sorts food by extent of processing rather than by nutrients:

The experiment that matters. Kevin Hall's group at the US National Institutes of Health admitted volunteers to a metabolic ward and fed them either ultra-processed or minimally processed diets, matched for calories offered, sugar, fat, fibre, sodium, and macronutrients, with unlimited eating. On the ultra-processed diet people ate about 500 more calories a day and gained weight; on the minimally processed diet they lost it. Same people, crossed over, same nutrients on paper, opposite outcomes.

That is a genuine causal demonstration, and the mechanism is not settled. Candidates include energy density, softer texture allowing faster eating, low satiety per calorie, additives affecting appetite signalling, and effects on the microbiome. Large observational studies associate ultra-processed intake with obesity, cardiovascular disease, type 2 diabetes, and mortality, with the usual confounding caveats.

The honest position: the finding is real and the mechanism is unknown, so the practical advice ("reduce the share of your diet that is ultra-processed") is well founded while any specific claim about which additive is responsible is not.

Alcohol

In short: A toxin your liver processes at about one drink an hour, with a group 1 carcinogen classification and no safe threshold for cancer.

What happens mechanically. Alcohol is absorbed rapidly, largely from the small intestine. The liver converts it with alcohol dehydrogenase into acetaldehyde, which is toxic and carcinogenic, then with aldehyde dehydrogenase into harmless acetate. The first step is fast and the second is the bottleneck, so acetaldehyde accumulates. Your liver handles roughly one standard drink per hour, and nothing speeds this up: not coffee, not cold showers, not food, though food slows absorption.

Why the East Asian flushing response matters. Roughly a third of people of East Asian descent carry an ALDH2 variant that slows the second step, so acetaldehyde builds up rapidly, causing flushing, nausea, and palpitations. Those who drink despite it have substantially raised oesophageal cancer risk, because they are marinating their oesophagus in a carcinogen (Chapter 62).

What it does over time: liver disease (Chapter 32), seven cancers (mouth, throat, oesophagus, liver, colorectum, breast), hypertension, atrial fibrillation, cardiomyopathy, pancreatitis, brain volume loss, disrupted sleep architecture, and dependence (Chapter 43).

The "moderate drinking is protective" story. Observational studies for decades showed lower mortality in moderate drinkers than in non-drinkers. The comparison group was the problem: it included former heavy drinkers who quit because they were ill, and people too unwell to drink. Studies separating lifelong abstainers, and genetic (Mendelian randomisation) analyses using alcohol-metabolism variants as a natural experiment, have largely dissolved the protective effect. Cancer risk rises from low levels of intake with no threshold, which is why several countries have revised their guidance sharply downward.

Meal timing and fasting

In short: Time-restricted eating and intermittent fasting work mainly by reducing total intake, and trials comparing them to ordinary calorie restriction generally find them equivalent.

Various patterns have been tested: 16:8 time-restricted eating, alternate-day fasting, 5:2. The consistent finding across randomised trials is that when calories are matched, the fasting pattern produces about the same weight loss and metabolic improvement as continuous restriction.

That is not a criticism. If a structured eating window makes it easier for you to eat less, it is a useful tool, and adherence is the binding constraint in every dietary intervention. It is a criticism of the claim that fasting has a unique metabolic magic independent of intake, which the trials do not support.

Two genuine caveats: very late eating is associated with worse glucose handling, since insulin sensitivity follows a daily rhythm and is lower at night. And prolonged fasting is not appropriate for people on insulin or sulfonylureas, in pregnancy, with a history of eating disorder, or who are underweight.

Hydration

In short: Drink to thirst plus a margin in heat and illness, and judge by urine colour rather than by counting glasses.

You lose roughly 2 to 2.5 litres a day and replace it from drinks, food (fruit and vegetables are mostly water), and metabolic water. The "eight glasses a day" rule has no clear origin in evidence and ignores that food supplies perhaps 20 to 30 percent of intake.

Practical markers: pale straw urine is about right; consistently dark suggests too little; consistently colourless suggests more than needed. Thirst is a reliable signal in healthy adults and becomes less reliable with age, which is why dehydration is common in older people and a frequent contributor to hospital admissions, falls, and acute kidney injury.

Drinking far too much matters too. Excessive intake in a short time can dilute blood sodium dangerously, which has killed endurance athletes and participants in drinking contests.

What a good diet actually looks like

In short: The evidence converges on a pattern rather than a rule set, and the pattern is unexciting.

Stripped of everything contested, the areas of genuine agreement across dietary guidelines, cohort studies, and the few long randomised trials:

And the meta-point: almost every diet that works does so by increasing satiety per calorie and reducing intake, whether it calls itself low-carbohydrate, Mediterranean, plant-based, or high-protein. Trials comparing named diets against each other consistently find small differences between them and large differences between people who adhere and people who do not. Choose the one you can sustain.

Six food claims worth dismissing

In short: These recur constantly and none survives contact with the evidence.

1. "Detox" diets and cleanses. Your liver and kidneys do this continuously and no commercial product improves on them. No detox product has ever identified which toxin it removes.
2. "Eating frequently stokes your metabolism." The thermic effect depends on total intake, not how it is divided. Meal frequency trials show no metabolic advantage.
3. "Alkaline diets change your blood pH." Blood pH is held between 7.35 and 7.45 by mechanisms that will dissolve your bones before letting it drift. Food changes urine pH, not blood pH.
4. "Superfoods." No single food has properties that matter against the background of a whole diet. The term is marketing, and it is banned in advertising in the EU without specific authorised health claims.
5. "Food intolerance test panels" (IgG testing). These measure exposure, not intolerance. Every major allergy body advises against them, and they generate long unnecessary exclusion lists (Chapter 46).
6. "Sugar feeds cancer." All cells use glucose, cancer cells included; starving yourself of carbohydrate does not selectively starve a tumour and malnutrition worsens cancer outcomes.

Don't be confused: "processed" is not a synonym for unhealthy. Freezing, pasteurising, canning, fermenting, and milling are all processing, and several of them make food safer and more nutritious. Frozen vegetables are often higher in vitamins than "fresh" produce that has spent a week in transit. The NOVA group that matters is group 4, industrial formulations, not processing in general.

Sources and notes

Energy expenditure components: standard nutrition texts and doubly-labelled-water studies. Thermic effect by macronutrient: Westerterp, Nutrition & Metabolism, 2004. Protein requirement figures: WHO/FAO/UNU; PROT-AGE recommendations (Bauer et al., JAMDA, 2013); Morton et al., BJSM, 2018. Anabolic resistance in ageing muscle: Wall, Gorissen, and van Loon. Fibre intake and outcomes: Reynolds et al., The Lancet, 2019, a series of meta-analyses. Ultra-processed food trial: Hall et al., Cell Metabolism, 2019. NOVA classification: Monteiro et al. PREDIMED: Estruch et al., NEJM, 2018. Trans fat elimination: national policy evaluations. Alcohol and Mendelian randomisation: Millwood et al., The Lancet, 2019; IARC classification of alcohol as a group 1 carcinogen. ALDH2 and oesophageal cancer: Brooks et al., PLoS Medicine, 2009. Intermittent fasting equivalence: Liu et al., NEJM, 2022, and multiple meta-analyses. IgG food testing: position statements from EAACI, AAAAI, and equivalents. Iron deficiency anaemia prevalence: WHO and Global Burden of Disease estimates.

Open questions. The mechanism by which ultra-processed food increases intake is unresolved and is the most important open question in nutrition. Optimal protein intake across the lifespan remains debated. Whether specific saturated fatty acids differ in effect, and how much the food matrix matters relative to nutrient content, is unsettled.

Next: what happens inside you when you move. 👉

What Movement Actually Does

TL;DR. Exercise is not a way of burning calories that happens to have side benefits. It is a signal that tells dozens of tissues to rebuild themselves. Contracting muscle pulls glucose out of your blood without needing insulin, builds new mitochondria, pulls on bone so bone thickens, releases signalling molecules that reach the brain and trigger the growth of new connections, and calms the low-grade inflammation underlying much of this book. If exercise came as a tablet it would be the most prescribed drug in the world by a wide margin, and the single most striking finding in the field is that aerobic fitness predicts death better than smoking, diabetes, or high blood pressure in large cohorts, with no upper limit to the benefit.

Key takeaways

• Contracting muscle takes up glucose without insulin. This is why a walk after a meal lowers blood sugar even in someone with severe insulin resistance.
• In a cohort of over 122,000 people, those with the lowest cardiorespiratory fitness had a higher risk of death than those with coronary artery disease, diabetes, or a smoking history, and benefit continued rising with fitness with no observed ceiling.
• The steepest part of the curve is at the bottom. Going from doing nothing to doing a little produces a bigger risk reduction than going from moderate to very high.
• Resistance training is not optional after 40. It is the only intervention that reliably reverses the muscle and bone loss in Chapter 7.
• Sitting is a separate risk from not exercising. Long uninterrupted sitting is associated with worse metabolic outcomes even in people who meet activity guidelines, though high activity attenuates it.
• Exercise has outcome-trial evidence in more than 25 chronic diseases, and in several it outperforms the drugs available.

What happens the moment you start moving

In short: Within seconds, muscle opens glucose doors that do not need insulin, the heart triples its output, and blood is redirected from the gut to the muscles.

That last row is the most clinically important sentence in this chapter. Normally, glucose enters muscle only when insulin instructs those transporters to move to the surface. Muscle contraction triggers the same movement through an entirely separate pathway that does not require insulin at all.

The consequences are direct:

• A walk after a meal lowers blood glucose in someone whose insulin barely works.
• Insulin sensitivity is improved for roughly 24 to 48 hours after a single session, which is why frequency matters more than duration.
• Exercise is effective in type 2 diabetes precisely because it bypasses the broken mechanism (Chapter 18).

What happens over weeks and months

In short: Every system adapts, and the adaptations are structural rather than motivational.

Muscle is an endocrine organ. Contracting muscle releases signalling molecules called myokines into the blood, which act on liver, fat, bone, blood vessels, and brain. This is the best current mechanistic explanation for why an activity performed by your legs improves your memory and your mood: the muscle is sending chemical messages, not just using energy.

The two numbers that predict how long you live

In short: Aerobic capacity and strength are among the strongest mortality predictors ever measured, and both are trainable at any age.

Aerobic fitness

VO2max is the maximum rate at which you can take in and use oxygen. It is limited mostly by how much blood your heart can pump and how well your muscles can extract oxygen from it, and it is the best single measure of cardiorespiratory fitness.

The mortality data are striking. In a cohort of over 122,000 people undergoing treadmill testing at the Cleveland Clinic:

• The difference in mortality between the lowest fitness group and the highest was larger than the effect of coronary artery disease, diabetes, smoking, or end-stage kidney disease.
• The relationship continued to improve at the highest fitness levels, with no observed ceiling, contradicting the idea that extreme fitness carries risk.
• Being unfit carried a risk comparable in magnitude to some of the most serious chronic diseases.

Association is not causation, and the reverse-causation concern is real (illness makes you unfit). But randomised trials do confirm that training improves fitness and improves the intermediate outcomes, and the size and consistency of the association is hard to explain away.

VO2max declines by roughly 10 percent per decade after 30 in sedentary people, and by roughly half that in people who keep training. That difference compounds. The practical version of this is that a 70-year-old who has trained consistently can have the aerobic capacity of a sedentary 45-year-old, and functional independence depends on staying above the threshold needed for daily tasks.

Strength

Grip strength, a cheap proxy for total muscle strength, predicts mortality across large international cohorts. In the PURE study of nearly 140,000 people across 17 countries, each 5 kg reduction in grip strength was associated with a 16 percent higher risk of death from any cause, and it predicted cardiovascular death better than systolic blood pressure did.

Sit-to-stand ability predicts the same thing more practically: whether you can rise from a chair, or from the floor, without using your hands is a genuine functional threshold, and it is what determines independence in later life.

The dose-response curve

In short: The gains are front-loaded, which means the most valuable exercise is the first hour a sedentary person does.

Plot mortality against activity and the curve drops steeply at first, then flattens.

This shape has two implications that matter for how advice should be given. Telling a completely sedentary person to do 150 minutes a week is aiming at the flat part of the curve while the steep part is available at 15 minutes a day. And there is no meaningful evidence of harm at high volumes for the general population, despite recurring headlines about extreme endurance exercise; the specific concerns (atrial fibrillation, coronary calcification in very high-volume male endurance athletes) apply to a small group and have not translated into higher mortality.

Steps. The 10,000-step target came from a 1960s Japanese pedometer marketing campaign, not from research. Accelerometer studies find benefit beginning around 2,500 to 4,000 steps a day and continuing to improve up to roughly 8,000 to 10,000 in older adults and somewhat higher in younger ones, after which it plateaus.

The four kinds of training, and what each actually gives you

In short: They are not interchangeable, and most people do only the first.

Resistance training is the one most commonly skipped and the one with the most specific irreplaceable benefit. No amount of walking or cycling will preserve muscle mass or bone density the way loading does. Trials in people in their eighties and nineties, including nursing home residents, show meaningful gains in strength, walking speed, and stair-climbing power. The machinery does not switch off; it just needs an adequate stimulus.

What "adequate" means in practice: sets taken close enough to failure that the last few repetitions are genuinely hard. Load matters less than proximity to failure for muscle growth, though heavier loads are better for bone and for maximum strength. Two sessions a week covering the major movement patterns captures most of the available benefit.

Exercise as a treatment

In short: In several conditions, exercise is not adjunctive advice but a first-line treatment with trial evidence behind it.

Sitting is a separate problem

In short: Meeting your exercise target does not fully cancel eight hours of uninterrupted sitting, though it substantially reduces it.

Muscle is metabolically active only when it is doing something. Long uninterrupted sitting reduces the activity of an enzyme that clears fat from blood, reduces glucose uptake, and is associated with worse outcomes independently of whether you exercise.

The effect is attenuated at high activity levels: analyses pooling over a million people found that roughly 60 to 75 minutes a day of moderate activity largely eliminated the excess mortality associated with long sitting. The practical implication is not to panic about desks but to break them up, since interrupting sitting every 30 to 60 minutes with a few minutes of movement improves glucose and lipid handling measurably.

A concrete weekly template

In short: One realistic pattern covering all four training types in about three hours a week.

Not a prescription, and a defensible default that most people can adapt.

Additions worth making by age: balance work daily from about 50 (standing on one leg while brushing teeth is genuinely enough to start); deliberate power work, meaning moving a moderate load quickly, from about 60, because power declines before strength and is what catches you when you stumble.

What people get wrong

In short: Six errors that cause most of the wasted effort and most of the injuries.

1. Starting at the intensity you think you should manage rather than the one you can recover from. Almost all early failure is from doing too much too soon, followed by injury or demoralisation. Progression beats intensity.
2. Doing only cardio. It leaves the muscle and bone problem completely unaddressed.
3. Treating exercise as a weight-loss tool. It is a poor one and an excellent health one, and people quit when the scale does not move.
4. Believing you must feel sore for it to work. Soreness is a marker of unaccustomed load, not of adaptation, and it fades as you adapt while the benefits continue.
5. Stretching before rather than moving before. Static stretching before exercise does not prevent injury and can transiently reduce power. A gradual warm-up does more.
6. Stopping entirely when something hurts. For most musculoskeletal pain, modified activity beats rest (Chapter 50). Complete rest deconditions fast.

Don't be confused: "no pain, no gain" is wrong, and so is "stop at the first discomfort." Effort that is hard but controlled is the stimulus. Sharp pain, joint pain, or pain that persists into the next day is a signal to change what you are doing. Muscle burn during a hard set and general fatigue afterwards are not.

If you are starting from nothing

In short: The first month is about building a habit, not a physiology, and the bar is far lower than most people assume.

• Walk daily, starting wherever you actually are, even 10 minutes. The steep part of the mortality curve is right here.
• Add two short resistance sessions a week, using bodyweight if that is what is available. Sit to stand from a chair, push against a wall or the floor, carry shopping deliberately.
• Progress by adding a little each week, roughly 10 percent, rather than jumping.
• Anchor it to something existing: after a specific meal, before a specific programme, with a specific person.
• Expect the benefits you cannot see first. Blood pressure, insulin sensitivity, mood, and sleep improve within weeks, well before anything is visible.

If you have heart disease, uncontrolled hypertension, severe lung disease, or symptoms on exertion, get assessed before starting anything vigorous. For most people, the risk of starting moderate exercise is lower than the risk of not doing so, and supervised programmes exist for exactly the higher-risk groups (cardiac and pulmonary rehabilitation), which are proven, funded in many health systems, and chronically under-attended.

Sources and notes

Cardiorespiratory fitness and mortality: Mandsager et al., JAMA Network Open, 2018 (122,007 patients, no observed ceiling of benefit). Grip strength and mortality: Leong et al., The Lancet, 2015 (PURE study, 139,691 participants across 17 countries). Dose-response for physical activity: Arem et al., JAMA Internal Medicine, 2015; Ekelund et al., BMJ, 2019. Step counts: Paluch et al., Lancet Public Health, 2022, meta-analysis of 15 cohorts. Sitting and its attenuation by activity: Ekelund et al., The Lancet, 2016, pooled analysis of over one million people. Contraction-mediated GLUT4 translocation independent of insulin: Richter and Hargreaves, Physiological Reviews, 2013. Myokines: Pedersen and Febbraio, Nature Reviews Endocrinology, 2012. Resistance training in the very old: Fiatarone et al., JAMA, 1990, and NEJM, 1994. Exercise for depression: Noetel et al., BMJ, 2024, network meta-analysis. VO2max decline with age and its attenuation by training: Fleg et al., Circulation, 2005. Falls prevention exercise: Sherrington et al., Cochrane review, 2019. Static stretching before exercise: Behm et al., systematic reviews.

Open questions. How much of the fitness-mortality association is causal versus reverse causation cannot be fully resolved without randomised trials of a size nobody will run. The optimal balance of aerobic and resistance training for longevity specifically is not established. Whether very high-volume endurance exercise carries a small cardiac risk in some individuals remains debated.

Next: the third of your life that determines much of the other two. 👉

What Sleep, Stress, and Mood Do to the Body

TL;DR. Sleep is not the absence of activity. It is a scheduled maintenance window in which the brain consolidates memory, flushes waste through a drainage system that opens mainly during sleep, and releases most of the day's growth hormone, while appetite hormones, insulin sensitivity, and immune function are all reset. Stress is a survival system designed for minutes that becomes destructive when it runs for years, and the damage it does is measurable in blood pressure, blood sugar, immune function, and belly fat. Depression is not only a mood state; it raises the risk of heart disease roughly twofold and shortens life mostly through physical illness. And loneliness has an effect on mortality comparable in size to smoking. All three of these are usually filed under "wellbeing", and all three are physiology.

Key takeaways

• Four nights of restricted sleep can reduce insulin sensitivity by around 25 percent in healthy young adults, producing a temporary state resembling early type 2 diabetes.
• The brain's waste clearance system (the glymphatic system) is far more active during sleep, which is the current leading explanation for why sleep is non-negotiable.
• Sleep loss raises ghrelin and lowers leptin, increasing appetite specifically for energy-dense food. Sleep is a weight-management variable.
• Acute stress is protective; chronic stress is corrosive. The same cortisol that saves you in an emergency causes visceral fat gain, insulin resistance, and immune suppression when it never switches off.
• Depression roughly doubles the risk of coronary heart disease and worsens outcomes after a heart attack, independently of behaviour.
• Social isolation carries a mortality risk comparable to smoking 15 cigarettes a day in meta-analysis, and larger than obesity.

Sleep: what actually happens

In short: Two very different states alternating in 90-minute cycles, each doing a different job, and losing either one has specific consequences.

The architecture

Sleep is not uniform. You cycle through stages roughly every 90 minutes, four to six times a night.

That last distribution matters practically. Going to bed two hours late costs you disproportionately more REM sleep; waking two hours early costs you the same. And alcohol suppresses REM in the first half of the night, producing the characteristic pattern of falling asleep quickly and then waking at 3 or 4 a.m. as it wears off and REM rebounds.

What sleep is for

How much you need

Seven to nine hours for most adults, with genuine individual variation. Teenagers need 8 to 10 and have a biologically delayed body clock, which is the argument behind later school start times. Older adults need roughly the same as younger ones but sleep more lightly and wake more often, so reduced sleep in later life is common rather than optimal.

Genuine short sleepers exist and are rare. Mutations in a small number of genes, including DEC2, allow people to function on 4 to 6 hours without impairment. They are perhaps 1 percent or less of the population, and almost everyone who believes they belong to this group performs worse on objective testing than they think.

Sleep debt is real and only partly repayable. Recovery sleep restores alertness quickly and restores some metabolic measures more slowly, and studies of weekend catch-up sleep find it does not fully reverse the metabolic effects of weekday restriction.

What sleep deprivation actually does

That last line is why drowsy driving is a substantial road safety problem and why regulated industries limit shift lengths. Shift work deserves specific mention: long-term night shift work is associated with higher rates of cardiovascular disease, type 2 diabetes, and some cancers, and the International Agency for Research on Cancer classifies shift work involving circadian disruption as probably carcinogenic.

The body clock

Nearly every cell in your body has a molecular clock, coordinated by a master clock in the hypothalamus. Its dominant input is light, detected by specialised retinal cells that respond most strongly to blue wavelengths and connect directly to the clock rather than to vision.

Practical consequences:

• Morning light anchors the clock. Getting outside within an hour or two of waking is the strongest available signal, and outdoor light is 10 to 100 times brighter than indoor lighting even on an overcast day.
• Evening light delays it. Screens matter less than total evening light exposure and less than what you do with the screen; the arousal from the content is often the bigger effect.
• Melatonin is a timing signal, not a sedative. It tells the body it is night. Small doses taken at the right time shift the clock, which is why it works for jet lag and shift work and is a weak sleeping tablet.
• Metabolism follows the clock. Insulin sensitivity is higher in the morning and lower late at night, which is one reason very late eating handles glucose worse.

What actually improves sleep

In short: Cognitive behavioural therapy for insomnia outperforms sleeping tablets and is the recommended first-line treatment nearly everywhere.

Sleeping tablets (benzodiazepines and "Z-drugs") produce modest objective improvement, cause dependence and rebound insomnia, impair memory, and increase falls and fractures in older adults. They have a place for short-term crisis use and are a poor long-term answer.

Stress: a system built for minutes

In short: Two responses, one fast and one slow, both adaptive in an emergency and both damaging when they never switch off.

The two arms

The fast one (seconds). The sympathetic nervous system fires, and adrenaline floods the circulation: heart rate and blood pressure rise, airways dilate, pupils widen, glucose is released, blood is redirected to muscle, digestion stops, and clotting is enhanced. This is the fight-or-flight response, and every element is useful if you are about to be injured, including the clotting.

The slow one (minutes to hours). The HPA axis: the hypothalamus signals the pituitary, which signals the adrenal glands, which release cortisol. Cortisol raises blood glucose, suppresses inflammation and immunity, sharpens attention, and mobilises fuel. It is designed to sustain you through a challenge and then switch off, via negative feedback onto the levels above.

What happens when it does not switch off

Chronic activation is a different physiological state, and the concept of allostatic load describes its accumulated cost.

The takotsubo case makes it concrete. Severe acute emotional stress can produce takotsubo cardiomyopathy, in which the left ventricle balloons and stops pumping properly, mimicking a heart attack with clear coronary arteries. It is usually reversible, it is predominantly seen in postmenopausal women, and it is a documented physiological demonstration that an emotional event can stop a heart from working.

What actually reduces the damage

The evidence is more modest than the wellness industry implies, and some of it is genuinely good:

Don't be confused: "stress causes disease" is too strong, and "stress is just in your head" is too weak. Chronic stress is not a proven direct cause of most chronic disease. It is a well-documented contributor that acts through measurable physiological changes and through behaviour, and its largest measured effects are on cardiovascular outcomes and on mental health. The honest framing is a real contributing factor of moderate size, not the root of all illness.

Mood: what depression does to the body

In short: Depression is a systemic illness with physical consequences, and the excess deaths it causes are mostly not suicide.

Depression is covered clinically in Chapter 41. What belongs here is its effect on the rest of the body, which is routinely underestimated.

That final row is the most important and the least known. The physical health of people with serious mental illness is systematically under-treated, a pattern with a name, diagnostic overshadowing: new physical symptoms get attributed to the psychiatric diagnosis and not investigated (Chapter 42).

Loneliness and connection

In short: One of the largest and least discussed effects in this book, comparable in size to established physical risk factors.

A meta-analysis of 148 studies covering over 300,000 people found that stronger social relationships were associated with a 50 percent increased likelihood of survival over the follow-up period. The authors compared the effect size to smoking 15 cigarettes a day, and found it exceeded the effects of obesity and physical inactivity.

The proposed mechanisms are the ones already described: chronic stress activation, worse sleep, higher inflammatory markers, less health-seeking behaviour, and no one to notice when something is wrong.

This matters practically in two places. In older adults, isolation is both common and addressable, and it is a modifiable dementia risk factor (Chapter 37). In illness, having someone involved measurably improves adherence, follow-up attendance, and outcomes, which is why so many chapters in this book end with a recommendation to tell someone.

Putting the three together

In short: Sleep, stress, and mood are a single interconnected loop, and intervening anywhere in it moves the others.

These three are usually discussed separately and behave as one system:

• Poor sleep raises cortisol, worsens mood, increases appetite, and impairs glucose handling.
• Chronic stress delays sleep onset, fragments sleep, and raises depression risk.
• Depression disrupts sleep architecture, raises inflammation, and reduces the activity and social contact that would improve both.

Which means the useful entry points are shared. Exercise improves all three. Sleep improves all three. Social contact improves all three. Treating one properly tends to improve the others, which is why treating insomnia improves depression outcomes and why treating depression improves diabetes control.

It also means the reverse: leaving one untreated undermines interventions aimed at the others, which is the practical reason this book keeps returning to sleep and mood in chapters ostensibly about hearts, kidneys, and blood sugar.

Sources and notes

Sleep architecture and function: standard sleep medicine references (Kryger, Roth, and Dement, Principles and Practice of Sleep Medicine). Insulin sensitivity after sleep restriction: Spiegel, Leproult, and Van Cauter, The Lancet, 1999, and subsequent replications. Appetite hormones: Spiegel et al., Annals of Internal Medicine, 2004. Testosterone after sleep restriction: Leproult and Van Cauter, JAMA, 2011. Glymphatic clearance during sleep: Xie et al., Science, 2013; amyloid-beta increase after one night of deprivation: Shokri-Kojori et al., PNAS, 2018. Sleep and susceptibility to the common cold: Prather et al., Sleep, 2015. Sleep deprivation equivalent to alcohol impairment: Williamson and Feyer, Occupational and Environmental Medicine, 2000. Shift work classification: IARC Monographs. CBT-I as first-line: American College of Physicians and NICE guidance. Allostatic load: McEwen, NEJM, 1998. Takotsubo cardiomyopathy: Templin et al., NEJM, 2015. Depression and coronary heart disease risk: multiple meta-analyses, including Nicholson, Kuper, and Hemingway, European Heart Journal, 2006. Life expectancy gap in severe mental illness: Hjorthøj et al., Lancet Psychiatry, 2017. Social relationships and mortality: Holt-Lunstad et al., PLoS Medicine, 2010 (148 studies, 308,849 participants).

Open questions. Whether treating insomnia prevents the cardiovascular and metabolic outcomes associated with poor sleep has not been demonstrated in long-term trials. The causal direction between depression and inflammation is only partly resolved. How much of the loneliness-mortality association is causal remains debated, though intervention trials are beginning.

Next: what happens to all of this over decades. 👉

How the Body Ages

TL;DR. Ageing is the gradual loss of the reserve capacity described in Chapter 1. Most organ systems lose roughly 1 percent of function a year from around the age of 30, which is invisible for decades because the reserve is so large, and then becomes suddenly visible when a system drops below the threshold daily life requires. That is why so much of ageing feels like it happens abruptly: nothing changed suddenly except which side of a line you were on. The rates of decline differ enormously between people, and a substantial share of the difference is not fixed biology but accumulated disuse, disease, and exposure. Peak capacity in early adulthood and rate of decline afterwards are both partly under your control, and between them they determine how many years you spend independent.

Key takeaways

• Reserve, not function, is what you lose first. A 70-year-old's kidneys work fine at rest and cannot respond to a challenge the way a 30-year-old's can.
• Most systems decline at roughly 1 percent a year after 30, but aerobic capacity falls faster (about 10 percent a decade) and muscle power faster still.
• Frailty is a distinct clinical state, not just old age, and it is partly reversible with resistance training and protein.
• Healthspan is not tracking lifespan. People are living longer with more years of disability, which makes compressing that period the actual goal.
• The biggest single reversible contributor to functional decline is disuse. Masters athlete data suggest a substantial fraction of what is attributed to ageing is deconditioning.
• Ageing research is genuinely progressing and is far ahead of its evidence in commercial form. Nothing sold today is demonstrated to extend healthy human lifespan.

What ageing actually is

In short: An accumulation of molecular damage that outpaces repair, producing loss of reserve rather than sudden failure.

Ageing is not a programme that runs on a schedule. It is what happens when damage accumulates faster than maintenance clears it, across many independent mechanisms at once. Biologists have catalogued these into a set of hallmarks of ageing:

None of these is the cause of ageing. They interact, and no single one has been shown to be the master switch, which is why interventions targeting one at a time have so far produced modest results.

The rate of decline, system by system

In short: Roughly 1 percent a year from 30, with wide variation between systems and between people.

This table is the practical heart of the chapter. All figures are averages for people who are not deliberately training, and the variation between individuals is enormous.

Three observations worth drawing out.

Not everything declines. Vocabulary, accumulated expertise, emotional regulation, and reported life satisfaction (which follows a U-shape, rising again after midlife in many populations) hold up or improve. The stereotype of uniform decline is wrong.

The systems that decline fastest are the ones most responsive to training. Aerobic capacity and muscle are at the top of the decline table and are also the two most modifiable items on it. That is not a coincidence; both are maintained on demand.

The threshold effect explains the suddenness. A person loses kidney function invisibly for forty years, then a routine infection plus an anti-inflammatory tips them into acute kidney injury. Nothing changed quickly except the margin.

Frailty: the state that actually matters

In short: A distinct, measurable, partly reversible condition of low reserve, which predicts outcomes better than age does.

Frailty is not a synonym for old. It is a state in which reserve across multiple systems is so depleted that a minor stressor, a urinary infection, a new drug, a small fall, produces a disproportionate decline. Two 85-year-olds can be entirely different: one runs a household, the other is destabilised by a chest infection.

A widely used definition requires three or more of five features:

1. Unintentional weight loss (roughly 5 kg or more in a year)
2. Self-reported exhaustion
3. Weakness (measured by grip strength)
4. Slow walking speed
5. Low physical activity

Frailty predicts falls, hospital admission, disability, surgical complications, and death better than chronological age does, which is why surgeons and oncologists increasingly assess it before deciding on treatment. It is a better answer than age to the question "will this person tolerate this?"

And it is partly reversible. Resistance training plus adequate protein improves strength, walking speed, and frailty scores in trials, including in nursing home residents in their nineties (Chapter 7).

Healthspan and the compression of morbidity

In short: Life expectancy rose faster than healthy life expectancy, so the goal is shortening the gap rather than extending the total.

Global life expectancy is roughly 73 years. Healthy life expectancy, meaning years lived without significant disability, is roughly 10 years shorter, and that gap has been widening in several countries.

James Fries proposed in 1980 that the goal should be compression of morbidity: pushing the onset of disability closer to the end of life, so that the period of dependency is shorter rather than longer. This is the framework almost all serious ageing research now uses, and it is a better goal than maximum lifespan for two reasons: it is what people actually want, and it is far more achievable.

The evidence that compression is possible is reasonable. Cohort studies of people with few risk factors, particularly those who exercise regularly, find not only longer life but a shorter period of disability at the end, meaning they are healthy for longer and decline faster when they do. That is exactly the desired shape.

What accelerates ageing

In short: The same list as everything else in this book, which is the strongest evidence that these mechanisms are shared.

What actually slows functional decline

In short: Four things have real evidence, and none of them is sold in a bottle.

One severely underrated item on that list is treating sensory loss. Hearing aids slowed cognitive decline in a randomised trial among higher-risk older adults, and cataract surgery is associated with lower dementia incidence in cohort studies. Both are common, both are fixable, and both are routinely left for years.

The science that is genuinely promising

In short: Real progress in the laboratory, no demonstrated human healthspan extension, and a commercial market far ahead of both.

The honest summary: the biology is real and progressing, the animal results are genuine, and nothing currently available has been shown to extend healthy human lifespan. Anyone selling you something on that basis is ahead of the evidence.

The Blue Zones, handled carefully

In short: A popular idea with genuine components and serious data problems, and worth knowing about both.

The "Blue Zones" (Okinawa, Sardinia, Ikaria, Nicoya, Loma Linda) were identified as regions with unusual concentrations of very long-lived people, and the lifestyle features described are plausible and overlap with everything else in this book: largely plant-based diets, daily physical activity built into life, strong social ties, and a sense of purpose.

The data problems are real and were not widely reported. Work by Saul Newman, awarded an Ig Nobel Prize in 2024, found that regions with the highest reported rates of extreme longevity tend to correlate with poor birth-record keeping, high poverty, and pension fraud, and that reported supercentenarian rates in several places fell sharply once birth registration improved. Okinawa's longevity advantage has also declined markedly in recent decades.

The reasonable position: the lifestyle recommendations derived from these populations are consistent with far better evidence from elsewhere, so following them is sensible. The specific claim that these regions hold a validated secret to extreme longevity does not survive scrutiny of the underlying records, and should not be the reason you do anything.

What ageing well actually looks like, by decade

In short: The interventions change, and the earlier ones are about building the peak while the later ones are about protecting the reserve.

Sources and notes

Hallmarks of ageing: López-Otín et al., Cell, 2013, updated 2023. Organ decline rates are drawn from the chapters cited in each row and from standard geriatric physiology; individual variation is large and these are population averages. VO2max decline: Fleg et al., Circulation, 2005. Sarcopenia rates: standard geriatrics literature. Kidney function decline: Denic et al., NEJM, 2017. Maximum heart rate: Tanaka, Monahan, and Seals, JACC, 2001. Frailty phenotype: Fried et al., Journals of Gerontology, 2001. Compression of morbidity: Fries, NEJM, 1980, with subsequent cohort evidence including Chakravarty et al., Archives of Internal Medicine, 2008. Healthy life expectancy figures: WHO Global Health Observatory. Hearing aids and cognition: ACHIEVE trial, The Lancet, 2023. Rapamycin in mice: Harrison et al., Nature, 2009. Epigenetic clocks: Horvath, Genome Biology, 2013, and successors. Partial reprogramming: Ocampo et al., Cell, 2016. Blue Zones data quality critique: Newman, preprint and subsequent publications, awarded the 2024 Ig Nobel Prize in demography; Okinawan longevity trends from Japanese national statistics.

Open questions. Whether any intervention can slow human ageing itself, as opposed to preventing specific diseases, is unproven. Whether epigenetic clocks measure a causal process or a consequence is unresolved. How much of observed functional decline is intrinsic biology versus accumulated disuse and disease is genuinely debated, and masters athlete data suggest the second component is larger than long assumed.

That completes the foundations. From here the book turns to what goes wrong, starting with what a disease actually is. 👉

What Is a Disease?

TL;DR. A disease is a failure of the body's regulation: some process that normally holds a value steady, repairs damage, or defends a border stops working, and the failure produces a recognisable pattern. That is the honest definition, and it is loose enough to admit an uncomfortable fact: for many of the biggest diseases in this book, the line between "healthy" and "sick" is a threshold that a committee chose on a continuous measurement. Blood pressure, blood sugar, bone density, and cholesterol have no natural cliff edge. Understanding that thresholds are decisions (well-evidenced decisions, but decisions) is the difference between reading medicine as revealed truth and reading it as what it is: a working model, revised often.

Key takeaways

• Almost every body system works by homeostasis, holding a value inside a narrow range using a sensor, a controller, and an effector. Most chronic disease is that loop failing at one of those three points.
• Disease, illness, and sickness are three different things: the biological process, the experience of it, and the social role it puts you in. A person can have any one without the others.
• Diagnostic thresholds move. The number that defines high blood pressure, diabetes, and high cholesterol has been lowered repeatedly, each time reclassifying millions of people overnight without anything changing in their bodies.
• Diseases sort into roughly eight mechanisms: infection, genetic, degenerative, metabolic, autoimmune, neoplastic (cancer), deficiency, and environmental. Most famous diseases are a mix.
• A syndrome is a pattern of findings that travel together before anyone knows the cause. Many diseases start life as a syndrome and get renamed once the mechanism is found. AIDS was a syndrome in 1981 and a viral infection by 1984.
• The single most useful question about any disease is not "what is it called" but "what is broken, and is the damage reversible?"

The body as a set of thermostats

In short: Almost every chronic disease is a control loop breaking at its sensor, its controller, or its effector.

Start with the normal case, because disease is only definable against it.

Your body holds dozens of quantities inside narrow ranges: core temperature near 37 degrees Celsius, blood pH between 7.35 and 7.45, blood glucose roughly between 70 and 140 mg/dL, blood sodium, blood calcium, blood oxygen, blood pressure, body water. Wander far outside any of those ranges and cells stop working, because the proteins that do the work of a cell are folded shapes that only hold their shape in specific conditions. This whole business of holding values steady is called homeostasis.

Every homeostatic system has the same three parts, exactly like a house thermostat:

Now you can define most chronic disease precisely: the loop breaks at one of those three points.

• The sensor/controller fails: the pancreas stops making insulin. That is type 1 diabetes.
• The effector stops listening: muscle and fat cells still receive the insulin signal but respond weakly. That is type 2 diabetes, and it is a completely different failure of the same loop.
• The effector is destroyed: nephrons in the kidney are lost, so the loop that controls blood pressure, fluid, and acid loses its actuator. That is chronic kidney disease.

This is why Chapter 18 spends its first pages on what insulin normally does. You cannot see a broken loop without knowing what the working loop looked like.

The other large family of failure is not a broken loop but a broken border. Your body maintains barriers: skin, gut lining, airway lining, blood-brain barrier, and the immune system that patrols behind them. Infection is a border breach. Allergy and autoimmunity are the guards attacking the wrong target. Cancer is a cell inside the walls that has stopped obeying the rules about when to divide.

Disease, illness, sickness: three different things

In short: The biological process, the felt experience, and the social role come apart routinely, which is why silent diseases go untreated.

Medical anthropology draws a distinction that clinical practice has quietly adopted, and it explains a lot of frustration on both sides of a consultation.

• Disease is the biological process: the blocked artery, the mutated gene, the replicating virus. It exists whether or not anyone notices.
• Illness is the lived experience: the pain, fatigue, fear, and disruption. It is what the patient actually brings to the appointment.
• Sickness is the social role: the sick note, the insurance code, the family's adjustment, the stigma.

These come apart routinely, and the mismatches are clinically important.

A patient who feels fine and is told to take a pill daily for the rest of their life is being asked to treat a disease while having no illness. Roughly half of people prescribed long-term preventive medication stop taking it within a year, and this mismatch is a large part of why. It is also why the hypertension chapter spends so long on the word "silent."

Don't be confused: a disease you cannot feel is not a mild disease. The correlation between how bad a disease feels and how much damage it is doing is weak, and for the biggest killers it is close to zero. High blood pressure, early type 2 diabetes, hepatitis C, HIV, and most early cancers are silent precisely during the years when treating them works best. Pain is a good alarm for injury and a terrible alarm for chronic disease.

Where the line is drawn, and who drew it

In short: Blood pressure, blood sugar, and bone density have no natural cliff edge, so the diagnostic threshold is a treatment decision, and it has moved repeatedly.

Here is the part that surprises people. For the largest chronic diseases, there is no biological cliff. Blood pressure risk rises smoothly and continuously from about 115/75 upward. Cardiovascular risk rises smoothly with LDL cholesterol. The complications of high blood sugar rise smoothly with average glucose. Nature supplies a gradient; medicine needs a line, because a doctor has to decide whether to treat.

So committees draw lines, using outcome data: at what value does the benefit of treating exceed the harm and cost of treating? That is a reasonable question with a defensible answer, and the answer changes as evidence and drugs improve.

None of this means the categories are fake. People below and above those lines really do have different futures on average, and lowering the hypertension threshold was backed by trials showing benefit from treating lower. But two things follow.

First, prevalence statistics are partly definitional. When you read that diabetes prevalence tripled, part of that is real (it mostly is) and part is the definition moving.

Second, a diagnosis near a threshold is a probability statement, not a verdict. A blood pressure of 132/82 does not mean a different body from 128/78. It means you have crossed into a group where, on average, action pays.

Eight mechanisms, and why most diseases are mixtures

In short: The textbook categories leak badly, so the useful question is not what kind of disease it is but where the chain of events can be interrupted.

Textbooks sort diseases by mechanism. The categories are useful as long as you hold them loosely.

Now watch how badly the boxes leak. Cervical cancer is neoplastic and infectious (caused by HPV, and therefore vaccine-preventable). Type 2 diabetes is metabolic, strongly genetic, and heavily environmental. Rheumatic heart disease is infectious and autoimmune (a strep infection triggers antibodies that attack heart valves). COPD is environmental, genetic in a minority (alpha-1 antitrypsin deficiency), and inflammatory. Cirrhosis can be viral, toxic, autoimmune, metabolic, or genetic, and looks nearly identical at the end regardless.

The lesson is that "what kind of disease is it" is less useful than "what is the chain of events, and where can we interrupt it?" Every treatment in this book is an interruption somewhere in a chain, and knowing where tells you what the treatment can and cannot do.

Syndrome, disorder, disease: a hierarchy of ignorance

In short: Those three words record how much is known about a cause, not how serious the condition is.

These three words track how much is known, not how serious it is.

• A syndrome is a cluster of findings that reliably occur together, cause unknown or unspecified. Named for the pattern.
• A disorder is a functional disturbance, often used where no structural lesion is visible, which is why most psychiatric conditions are officially "disorders."
• A disease implies a known mechanism, or at least a known lesion.

Things graduate. AIDS (Acquired Immune Deficiency Syndrome) was named in 1982 for a pattern of collapse in young men with no known cause; HIV was identified in 1983 to 1984, and the syndrome became the end stage of a viral infection. Down syndrome was described by its pattern in 1866 and explained by an extra copy of chromosome 21 in 1959. Peptic ulcer disease was a stress disorder until 1982, when Helicobacter pylori turned most of it into an infection curable with antibiotics.

There are still major conditions sitting on that ladder today. Migraine, irritable bowel syndrome, myalgic encephalomyelitis/chronic fatigue syndrome, long COVID, and most psychiatric diagnoses are defined by pattern rather than by a measurable lesion. That does not make them imaginary. It makes them unfinished.

Two more things that shape every chapter

Diseases rarely arrive alone. Multimorbidity, having two or more chronic conditions, is the normal state of older patients, and it is the rule rather than the exception past about age 65 in high-income countries. This matters because guidelines are written one disease at a time. A patient with diabetes, hypertension, kidney disease, and arthritis can end up on a dozen medications, several of which pull against each other. Chapter 16 covers what that does.

The same disease is not the same in two people. Age, sex, genetics, other diseases, and how early it was caught all change the course. This is why every survival number in this book describes a population. It is also why medicine has been drifting for two decades toward subtyping: not "breast cancer" but "hormone-receptor-positive, HER2-negative breast cancer," which is a different disease with different drugs and a different outlook.

Sources and notes

Homeostasis, the sensor/controller/effector framing, and the mechanism categories are standard physiology and pathology, drawn from general texts (Guyton and Hall, Textbook of Medical Physiology; Robbins, Basic Pathology). The disease/illness/sickness distinction comes from medical anthropology, notably Arthur Kleinman's work in the 1970s and 1980s. Threshold histories: the US hypertension threshold change is the 2017 ACC/AHA guideline; the diabetes fasting-glucose threshold moved from 140 to 126 mg/dL with the 1997 American Diabetes Association expert committee report; the osteoporosis T-score definition is the 1994 WHO working group. The H. pylori story is Marshall and Warren's 1983 to 1984 work, awarded the 2005 Nobel Prize in Physiology or Medicine. Prevalence percentages for reclassified hypertension are the widely cited estimates published alongside the 2017 guideline and should be read as approximate.

Open questions. Where thresholds should sit is genuinely contested, especially for prediabetes (whose value as a category is disputed) and for treating mildly elevated blood pressure in the very old, where the balance of benefit and harm is narrower than in middle age.

Next: the system that decides whether an infection becomes a disease at all. 👉

The Immune System, From Scratch

TL;DR. Your immune system is two systems bolted together. The innate half is fast, dumb, and identical in everyone: it recognises broad patterns that mean "not human," attacks within minutes, and produces the swelling, redness, and fever you actually feel. The adaptive half is slow, specific, and unique to your life history: it takes days to build a response against one exact molecular shape, and then remembers that shape for decades. Nearly everything in this book touches this machinery. Infections are the immune system winning or losing. Vaccines are the memory half, trained without the disease. Allergy is the system attacking something harmless. Autoimmunity is it attacking you. Cancer partly succeeds by hiding from it, and modern immunotherapy works by removing the hiding place.

Key takeaways

• Symptoms of infection are mostly the immune response, not the pathogen. Fever, aches, swelling, and mucus are your side of the fight.
• Innate immunity responds in minutes to generic "not-self" patterns. Adaptive immunity takes 7 to 14 days on first exposure and hours to days on every exposure afterwards, which is the entire point of vaccination.
• Antibodies are made by B cells and work outside cells (blocking and tagging). Killer T cells work on infected cells from the inside out. A virus already inside a cell can only be cleared by the second mechanism.
• Immunological memory is why you get chickenpox once and colds forever: memory is specific to a molecular shape, and some viruses change shape constantly.
• The system's central problem is tolerance, learning not to attack you. Failures of tolerance produce type 1 diabetes, rheumatoid arthritis, lupus, and multiple sclerosis.
• Chronic low-grade inflammation, unlike the acute kind, causes damage without fighting anything. It links obesity to diabetes, heart disease, and some cancers.

Layer 0: the walls

In short: Skin, mucus, stomach acid, and resident bacteria do most of the defending, which is why a catheter or a burn is more dangerous than an exposure.

Before any immune cell is involved, you are defended by barriers, and they do most of the work.

Skin is a dry, acidic, constantly shedding surface with a dead outer layer that few organisms can cross intact. Your airways are lined with cells whose beating hairs (cilia) push a sheet of mucus upward at roughly 1 cm per minute, carrying trapped particles to the throat to be swallowed. Stomach acid sits near pH 1.5 to 3.5 and sterilises most of what you swallow. Tears and saliva contain lysozyme, an enzyme that dissolves bacterial cell walls. Urine flow flushes the urinary tract. And your gut, skin, and mouth are already occupied by trillions of resident bacteria (the microbiome) that compete for space and nutrients with anything new.

This is why breaking a barrier is so much more dangerous than being exposed to germs: a surgical wound, a burn, a ventilator tube, a urinary catheter, or an intravenous line is a hole in the wall. It is also why smoking matters so much for pneumonia. Cigarette smoke paralyses the cilia, and the escalator stops.

Layer 1: innate immunity, the fast dumb response

In short: Within minutes, generic not-human patterns trigger the fever, swelling, and pus that you actually feel during an infection.

If something crosses a barrier, resident cells detect it within minutes. They do not recognise the specific organism. They recognise patterns that no human cell has: the LPS molecule in gram-negative bacterial membranes, the flagellin of bacterial tails, double-stranded RNA (which human cells make only when a virus is copying itself), unmethylated bacterial DNA. Sensor proteins called toll-like receptors detect these patterns and trigger an alarm.

The alarm does four things.

It calls in cells. Neutrophils, the most common white blood cell, arrive within an hour and eat bacteria (phagocytosis), then die in place. Pus is mostly dead neutrophils. Macrophages ("big eaters") are longer-lived, eat debris, and coordinate what happens next.

It opens the blood vessels. Local vessels dilate and become leaky so that cells and fluid can get into tissue. This is exactly the classic four signs of inflammation, described by Celsus in the first century: redness and heat (more blood flow), swelling (leaked fluid), and pain (chemicals like bradykinin and prostaglandins irritating nerve endings). Everything that makes an infected cut look infected is your own response.

It runs the complement cascade. Complement is a set of about 30 blood proteins that activate each other in sequence. The end products punch holes in bacterial membranes, coat bacteria so phagocytes grab them more easily, and shout for more cells. It is ancient, always on standby, and works with no prior exposure.

It raises the thermostat. Signalling molecules called cytokines, especially interleukin-1 and interleukin-6 and TNF-alpha, travel to the hypothalamus in the brain and raise the body's temperature set point. You then feel cold and shiver, because your actual temperature is now below the new target. Fever mildly impairs many pathogens and speeds immune cells up. This is also why fever-reducing drugs work: they block the enzyme that makes the prostaglandin that resets the thermostat.

Don't be confused: inflammation is not infection. Inflammation is the response; infection is one of many things that can trigger it. A sprained ankle, a splinter, a sunburn, an autoimmune attack, and a cholesterol plaque in an artery all inflame with no microbe involved. When a doctor says a disease is "inflammatory," they mean this machinery is running, not that you have germs.

Layer 2: adaptive immunity, the slow specific response

In short: The body pre-builds around a billion random receptors and then amplifies whichever one happens to fit, which takes a week the first time and hours ever after.

Innate immunity buys time. If it cannot finish the job in a few days, the adaptive system takes over. It has two arms and one trick.

The trick is enormous prebuilt diversity. During development, each B and T cell randomly shuffles and splices gene segments to build one unique receptor. The result is a repertoire of something on the order of a hundred million to a billion different receptor shapes, generated before your body has met any of them. Nothing is designed in response to a pathogen; the matching receptor almost certainly already exists, sitting in a lymph node, waiting.

When a pathogen arrives, the few cells whose receptors happen to fit it are activated and told to divide, repeatedly. This is clonal selection: the immune system does not invent an answer, it amplifies the cell that already had it. That amplification takes about a week, which is why a genuinely new infection makes you sick for a week.

Arm 1: B cells and antibodies. An activated B cell becomes a plasma cell, a factory that pumps out thousands of antibodies per second. An antibody is a Y-shaped protein whose two arm tips grip one specific molecular shape (an antigen). Antibodies work outside cells, in blood and on wet surfaces, and they do three jobs: neutralise (physically block a virus from docking onto a cell), opsonise (coat a microbe so phagocytes grab it), and activate complement.

Arm 2: T cells. T cells only respond to fragments displayed on the surface of your own cells, on molecules called MHC. Every cell in your body constantly chops up a sample of its interior proteins and displays the pieces on its surface, effectively publishing a manifest of what is inside it.

• Cytotoxic T cells (CD8) patrol those manifests. If a cell is displaying viral protein, the T cell orders it to self-destruct. This is the only way to clear a virus that is already inside cells.
• Helper T cells (CD4) are the coordinators. They license B cells to make high-quality antibodies, activate macrophages to kill what they have eaten, and direct the whole response. HIV kills CD4 cells specifically, which is why Chapter 30 is a chapter about the collapse of everything else.

Memory. After the fight, most of the expanded clones die, but some persist as memory cells for years to decades. On re-exposure, the response is faster (hours to days) and larger, often clearing the pathogen before you notice symptoms. That is immunity. That is also the entire mechanism of vaccination: present the shape without the disease, get the memory anyway.

Why some diseases come back forever

Memory is specific to shape. Pathogens that change shape defeat it.

Influenza mutates its surface proteins continuously (antigenic drift) and occasionally swaps whole gene segments with animal flu viruses (antigenic shift), which is why the vaccine is reformulated yearly and why pandemics happen. The common cold is caused by over 160 rhinovirus types plus other virus families, so you can be immune to dozens and still catch one. HIV mutates within a single patient faster than antibodies can track. Malaria parasites systematically vary their surface proteins, which is why people in high-transmission areas gain partial immunity over years and never full immunity.

Compare measles, which is antigenically stable: catching it once, or being vaccinated, protects for life. That stability is why measles is eliminable and flu is not.

The three ways the system fails

In short: Too little is immunodeficiency, too much aimed outward is allergy, and too much aimed inward is autoimmunity.

Too little: immunodeficiency. Missing or broken components. Inherited forms are rare. Acquired forms are common and matter enormously: HIV destroying CD4 cells, chemotherapy wiping out neutrophils, steroids and biologic drugs suppressing inflammation on purpose, poorly controlled diabetes impairing neutrophil function, old age (immunosenescence), and malnutrition, which remains the world's most common cause of immune deficiency.

Too much, aimed outward: allergy. The system builds an IgE antibody response against something harmless (pollen, peanut protein, cat dander). On re-exposure, IgE on mast cells triggers a massive histamine release: swelling, itching, mucus, constricted airways, and in the extreme, anaphylaxis. See Chapter 46.

Too much, aimed inward: autoimmunity. During development, T and B cells that strongly recognise your own tissue are deleted or suppressed. This is tolerance, and it is imperfect. When it fails, the same machinery that clears bacteria dismantles your joints (rheumatoid arthritis), your insulin-producing cells (type 1 diabetes), your nerve insulation (multiple sclerosis), or many organs at once (lupus). Autoimmunity is strikingly sex-skewed: roughly 4 out of 5 people with an autoimmune disease are women, for reasons involving X-chromosome gene dosage and sex hormones that are still being worked out.

The slow burn: chronic inflammation

In short: The same machinery running quietly for years with nothing to kill is what links obesity to diabetes, heart disease, and several cancers.

Acute inflammation is a controlled emergency that resolves. Chronic inflammation is the same machinery running at low intensity for years with nothing to kill, and it is one of the connective threads of modern chronic disease.

Fat tissue, especially visceral fat around the organs, is not inert storage. It secretes inflammatory cytokines. Those cytokines interfere with insulin signalling (a direct link from obesity to type 2 diabetes), promote the immune activity inside artery walls that builds atherosclerotic plaque, and create a tissue environment that favours some cancers. Chronic infections (hepatitis B and C, H. pylori, HPV) cause cancer through the same route: decades of inflammation and cell turnover in one tissue. Even in the absence of disease, inflammatory markers drift upward with age, a phenomenon labelled inflammaging.

This is why a blood test for C-reactive protein, a nonspecific inflammation marker, predicts cardiovascular events, and why anti-inflammatory approaches keep appearing in chapters that seem to have nothing to do with infection.

Sources and notes

Standard immunology, drawn from Janeway's Immunobiology and Abbas, Cellular and Molecular Immunology. The four cardinal signs of inflammation are from Celsus, De Medicina, first century CE. Receptor diversity estimates vary by method and are given as an order of magnitude. Toll-like receptor discovery: Hoffmann, Beutler, and Steinman's work on innate immunity and dendritic cells received the 2011 Nobel Prize. The female predominance in autoimmune disease is a robust epidemiological finding across many conditions; the roughly 4-to-1 figure is an aggregate and varies widely by specific disease (about 9-to-1 in lupus, closer to 1-to-1 in type 1 diabetes). Rhinovirus type counts are from serotype and genotype surveys.

Open questions. Why tolerance fails in any individual is not known for any autoimmune disease. The role of the microbiome in shaping immune development is real but frequently overstated in popular coverage, and most causal claims about it in humans are not yet settled.

Next: what a pathogen has to accomplish to get from one person into another, and why that determines almost everything about how a disease behaves. 👉

How Infections Spread

TL;DR. Every infectious disease is a solution to one engineering problem: get from inside one host to inside the next. The route it uses (air, water, sex, blood, mosquito, touch) determines nearly everything else about it: who catches it, how fast it spreads, which season it peaks in, which countries it devastates, and which intervention stops it. Sanitation ended cholera, not antibiotics. Bed nets cut malaria more than any drug. Masks and ventilation matter for tuberculosis and not at all for hepatitis C. If you know the route, you can usually predict the control measure without knowing any microbiology at all.

Key takeaways

• Five kinds of pathogen (viruses, bacteria, fungi, parasites, prions) differ so much that "germ" is close to useless as a category. Antibiotics work on one of the five.
• The transmission route predicts the control measure. Water route means engineering; vector route means the vector; sexual route means behaviour, barriers, and treatment-as-prevention.
• R0 is the average number of people one case infects in a fully susceptible population. It sets the herd immunity threshold, the fraction that must be immune to stop sustained spread: $1 - 1/R_0$. Measles at R0 near 15 needs about 93 percent; that is why measles is the first disease to come back when vaccination slips.
• The gap between infectiousness and symptoms decides whether isolation can work. SARS was controllable because people were sickest when most infectious. COVID-19 was not, because they were not.
• Roughly 60 percent of human infectious diseases came from animals. New ones keep arriving by the same door.
• Killing the host fast is usually bad strategy for a pathogen, which is why extremely lethal outbreaks often burn out locally while mild ones circle the globe.

The five kinds of pathogen

In short: Antibiotics work on one of the five, which is the entire reason they do nothing for a cold.

Two consequences of that table drive the rest of the book. Antibiotics do nothing to viruses, so taking them for a cold gives you all the risk (side effects, gut damage, resistance selection) and none of the benefit. See Chapter 36. And viruses are hard to drug because they use your machinery, so there is little to attack that is uniquely theirs. That is why medicine leans so heavily on vaccines for viruses and drugs for bacteria.

The routes, and what stops each

In short: The transmission route predicts the control measure, which is why sewers ended cholera and bed nets cut malaria more than any drug did.

A pathogen must exit one host, survive transit, and enter the next in sufficient number (the infectious dose, which ranges from about 10 organisms for Shigella to hundreds of thousands for some others). The route it uses is the most informative single fact about it.

This table is why history reads the way it does. London's cholera did not end because someone cured cholera. It ended because John Snow traced the 1854 Broad Street outbreak to one water pump, and because London then spent decades building sewers. Cutting the route beats treating the disease, almost every time, and it beats it at lower cost.

Don't be confused: "airborne" and "droplet" used to be a hard distinction, and that distinction was wrong. The old teaching held that large droplets fell within about two metres while only a few special diseases (measles, TB, chickenpox) travelled as true aerosols. Aerosol physics established during COVID-19 that exhaled particles form a continuum, that small ones stay suspended and accumulate in poorly ventilated rooms, and that the two-metre rule is a rough guide rather than a boundary. This is why ventilation became a serious public health topic for the first time in a century.

Timing: the numbers that decide whether control is possible

In short: If people become infectious before they feel ill, isolating the visibly sick cannot work, which is the whole difference between SARS and COVID-19.

Four intervals define an infection's behaviour, and their relative order matters more than their absolute lengths.

• Incubation period: infection to first symptom.
• Latent period: infection to becoming infectious.
• Infectious period: the window during which you can transmit.
• Serial interval: the average gap between one case's symptom onset and the next case's.

Now the crucial comparison. If you become infectious after you feel ill, symptoms are a usable alarm: people isolate before they spread. If you become infectious before symptoms, or without ever having them, isolation based on symptoms leaks.

Rabies is the extraordinary case: the incubation is so long that vaccinating after the bite still beats the virus to the brain. Almost no other infection allows this.

R0, and why measles is the canary

In short: Measles needs roughly 95 percent immunity to stop spreading, so it is always the first disease to return when vaccination coverage slips.

R0 (pronounced "R-nought") is the average number of secondary cases produced by one infectious case in a completely susceptible population. It is not a fixed property of a microbe; it depends on biology, behaviour, and setting. The same pathogen has a higher R0 in a crowded city than in a village.

If R0 is above 1, cases grow. Below 1, the outbreak dies out. Rt, the effective reproduction number, is the same quantity in the real population at time t, with immunity and interventions included. Every control measure is an attempt to push Rt below 1.

The herd immunity threshold is the immune fraction needed to hold Rt below 1:

$$ H = 1 - \frac{1}{R_0} $$

Two lessons fall out. Measles is the most contagious common human disease, so it needs the highest coverage, so it is always the first outbreak to return when vaccination rates dip a few points. It is an early-warning indicator for a whole immunisation programme. And herd immunity is a threshold, not a gradient of personal safety: below it, outbreaks are possible; the protection it offers to people who cannot be vaccinated (infants, the immunosuppressed) only exists once the threshold is cleared.

R0 also explains why smallpox could be eradicated (moderate R0, no animal reservoir, obvious rash, stable virus, effective vaccine) while flu cannot be (animal reservoirs in birds and pigs, constant mutation, presymptomatic spread).

Why the deadliest outbreaks are often the smallest

In short: Pathogens evolve toward maximum transmission rather than toward mildness, so lethality is selected against only when it interferes with spreading.

There is a widespread intuition that pathogens evolve toward mildness. The truth is more specific: pathogens evolve toward maximum transmission, and lethality is selected against only when it interferes with transmission.

Ebola kills a large share of the people it infects and requires close contact with body fluids. Patients become too ill to move around while at their most infectious, so outbreaks tend to burn out locally unless they reach a city. A rhinovirus that gives you a runny nose keeps you at work sneezing on people, and it circulates worldwide forever. The 1918 influenza pandemic is the awkward counterexample: trench and troop-transport conditions moved even severely ill patients efficiently, which removed the usual penalty on virulence.

Note the corollary: a pathogen with a long asymptomatic infectious period can be both lethal and extremely successful. Untreated HIV kills nearly everyone it infects and still spread to tens of millions, because it spent years being transmissible and invisible.

Where new diseases come from

Roughly 60 percent of known human infectious diseases and about three-quarters of recent emerging ones originated in animals. The pattern of spillover is consistent: sustained close contact between people and animals, an amplifying setting, then a virus that manages the jump.

HIV crossed from primates in Central Africa in the early twentieth century. Influenza pandemics assemble in pigs and birds. Ebola persists in bats. SARS reached humans through live-animal markets in 2002, MERS through camels, and SARS-CoV-2 most probably through an animal intermediary at a market in Wuhan, a question still formally open.

The drivers of spillover are not mysterious: expanding agriculture into wildlife habitat, dense livestock farming, wildlife trade, and air travel that converts a local outbreak into a global one within a week. This is why surveillance networks watch animal populations, and why influenza vaccine strain selection is an annual international process.

Sources and notes

Transmission categories, incubation periods, and R0 ranges are drawn from standard references (Heymann, Control of Communicable Diseases Manual; CDC and WHO fact sheets) and from the epidemiological literature; R0 values vary substantially between studies and settings and are given as ranges deliberately. The measles R0 range of 12 to 18 comes from classic pre-vaccine data and has been argued upward and downward since. The John Snow Broad Street investigation is 1854. The zoonotic-origin proportions come from Taylor, Latham, and Woolhouse (2001) and Jones et al. (2008). Aerosol transmission physics: the reassessment during 2020 to 2021 by Wang, Marr, Morawska, and colleagues.

Open questions. The origin of SARS-CoV-2 remains under investigation and is not settled. Herd immunity thresholds calculated from a single R0 assume uniform mixing, which no real population has, so the true figures are approximations.

Next: the other source of disease, the one you are born with. 👉

Genes, Heredity, and Disease

TL;DR. A gene is a recipe for a protein, and proteins do everything: they carry oxygen, digest food, build structure, and pass signals. A disease-causing mutation is a typo in one recipe, and its consequences follow from what that protein was for. Two very different things get called "genetic." A single-gene disease (sickle cell, cystic fibrosis, Huntington's) is one broken recipe with a predictable inheritance pattern. A polygenic disease (type 2 diabetes, heart disease, depression, most of what people actually get) is the summed nudge of hundreds or thousands of common variants, each nearly meaningless alone, none of them a verdict. Confusing the two is the single most common error in how people think about family history.

Key takeaways

• DNA to RNA to protein. A mutation changes a protein: makes it absent, misshapen, overactive, or made in the wrong amount. Everything downstream follows from that.
• Dominant means one bad copy is enough. Recessive means you need two, and people with one copy are carriers who are usually healthy and can be unaware for life.
• Penetrance is the fraction of people carrying a variant who actually develop the disease. Huntington's is essentially 100 percent. A BRCA1 mutation is high but not certain. Most risk variants are far below 10 percent.
• Germline mutations are inherited and in every cell. Somatic mutations happen during life in one cell line and are not passed to children. Almost all cancer is somatic.
• Heritability is a population statistic about variation, not a personal percentage. "80 percent heritable" does not mean 80 percent of your case was genes.
• Genes set a range; environment picks the point in it. PKU is the cleanest proof: a fully genetic disease, entirely prevented by diet.

The machinery in one page

In short: A gene is a recipe for a protein, and one wrong letter can change the protein's shape and therefore its job.

Your DNA is a four-letter code (A, C, G, T) roughly 3 billion letters long, packed into 23 pairs of chromosomes. One of each pair came from each parent. About 20,000 stretches of that code are genes, and each gene is a recipe for one protein (with variations that let one gene make several related versions).

Reading the recipe is a two-step process. The gene is copied into messenger RNA (transcription), and the RNA is read three letters at a time by a ribosome, which links the corresponding amino acids into a chain (translation). The chain folds into a specific three-dimensional shape, and the shape is the function. A protein works because its surface fits something else: a molecule to be carried, a reaction to be catalysed, a signal to be received.

That is why a single-letter typo can be catastrophic. Change one letter in the gene for haemoglobin's beta chain, and a glutamic acid becomes a valine. That swaps a water-loving amino acid for a water-fearing one on the protein's surface, so the molecules stick to each other when oxygen is low, and the whole red blood cell deforms into a rigid crescent. One letter, and you have sickle cell disease (Chapter 48).

Not every typo matters. Most of the genome is not gene, much of a gene's code is redundant (several triplets code for the same amino acid), and many amino acid swaps change nothing important. Everyone carries millions of variants and a handful of seriously damaged genes without consequence.

The four inheritance patterns

In short: Dominant, recessive, X-linked, and mitochondrial inheritance each leave a recognisable family pattern, and recessive disease often appears with no family history at all.

Because chromosomes come in pairs, a variant's effect depends on whether one copy or two is required, and on which chromosome it sits.

Autosomal dominant. One faulty copy causes disease. An affected parent passes it to each child with probability 1/2. Every generation is affected; the pattern is vertical in a family tree. Examples: Huntington's disease, familial hypercholesterolaemia, Marfan syndrome, most hereditary breast/ovarian cancer syndromes.

Autosomal recessive. Two faulty copies are required. Carriers with one copy are healthy. Two carrier parents have a 1/4 chance per child of an affected child, 1/2 of a carrier, 1/4 of neither. Disease often appears with no family history at all, which is why it feels like it comes from nowhere. Examples: cystic fibrosis, sickle cell disease, thalassemia, Tay-Sachs, phenylketonuria.

X-linked. The gene is on the X chromosome. Males have one X, so a single faulty copy is fully expressed; females have two and are usually carriers. The classic signature is a disease that skips through unaffected mothers to affected sons. Examples: haemophilia A and B, Duchenne muscular dystrophy, red-green colour blindness, G6PD deficiency.

Mitochondrial. Mitochondria (the cell's power plants) carry their own small genome, and you inherit all of yours from your mother. Affected mothers pass it to all children; affected fathers pass it to none. Examples: Leber's hereditary optic neuropathy, MELAS.

Don't be confused: "runs in the family" is not the same as "genetic." Families share far more than DNA. They share diet, smoking, income, language, neighbourhood, and air quality. Type 2 diabetes clusters in families partly through genes and partly because a family eats together. Conversely, a genetic disease can appear with no family history at all, either because it is recessive and both parents were silent carriers, or because the mutation is brand new in that child (de novo), which accounts for a large share of Duchenne muscular dystrophy and achondroplasia cases.

Penetrance: the difference between a gene and a fate

In short: Penetrance is the proportion of carriers who actually develop the disease, and for most variants it is low.

Penetrance is the probability that someone carrying a variant develops the disease. It is the number people most need and most rarely hear.

Expressivity is the separate question of how severely it shows. Two siblings with identical cystic fibrosis mutations can have markedly different lung disease, because other genes and environment modify the outcome.

Polygenic: how most common disease is actually inherited

In short: Common diseases are hundreds of small genetic nudges rather than one broken gene, which is why risk scores predict populations far better than people.

Sickle cell has one gene. Type 2 diabetes has hundreds of contributing variants, and so do coronary artery disease, obesity, depression, schizophrenia, asthma, and height.

Genome-wide association studies compare millions of common variants between people with and without a condition. They routinely find hundreds of hits, each shifting risk by a few percent. Sum them, weighted, and you get a polygenic risk score, which does have real predictive power at the extremes of its distribution: people in the top few percent of a coronary disease score can carry risk comparable to a single-gene familial cholesterol disorder.

Three honest limits on those scores. They predict populations far better than individuals. They were derived overwhelmingly from people of European ancestry and transfer poorly to others, which is an active fairness problem in genomics. And for most diseases they add modestly to what you already know from age, blood pressure, smoking, cholesterol, and family history.

Heritability, the most misunderstood number in medicine

"Schizophrenia is about 80 percent heritable." Almost everyone reads that as "80 percent of the disease is genetic." It does not mean that.

Heritability is the proportion of the variation between people in a specific population at a specific time that is attributable to genetic variation. It is a property of a population, not of a person or a disease.

The consequences are counterintuitive. If everyone in a population smoked identically, the variation in lung cancer between them would be almost entirely genetic, and lung cancer would look highly heritable, even though smoking causes most of it. Improve the environment for everyone and heritability goes up, because you have removed environmental variation. Height in well-fed countries is about 80 percent heritable and was much less so when childhood nutrition varied enormously.

So heritability tells you nothing about whether an intervention will work, and nothing about any individual's case.

Genes and environment, together

In short: PKU is entirely genetic and entirely preventable by diet, which is the cleanest proof that a genotype is a range rather than a verdict.

The cleanest demonstration in medicine is phenylketonuria (PKU). A recessive mutation disables the enzyme that breaks down the amino acid phenylalanine. It builds up and causes severe, permanent intellectual disability. This is 100 percent genetic and 100 percent preventable: put the infant on a low-phenylalanine diet from birth and development is normal. That is why every newborn in most of the world gets a heel-prick blood test in the first days of life, a programme running since the 1960s.

Other clean examples:

• Lactase persistence. Most adult humans worldwide lose the ability to digest milk sugar. Variants keeping the enzyme switched on arose independently in Northern Europe and in East African and Middle Eastern pastoralist populations. Whether that matters at all depends entirely on whether your diet contains milk.
• ALDH2 deficiency. About a third of people of East Asian descent carry a variant that slows breakdown of acetaldehyde, alcohol's toxic intermediate. It causes flushing, and it substantially raises oesophageal cancer risk if the person drinks. Never drink, and it is invisible.
• G6PD deficiency. Common around the historic malaria belt. Harmless until certain drugs (including some antimalarials) or fava beans trigger red cell breakdown.

The pattern is general: a genotype is a range of possible outcomes, and the environment selects within it.

Germline vs somatic, and why cancer sits in this chapter

Germline mutations are inherited, present in every cell, and passed to children. Somatic mutations arise during life in one cell and are copied only into that cell's descendants.

Cancer is fundamentally a somatic genetic disease: a cell accumulates mutations in the genes that control division and death until it escapes the rules (Chapter 24). Only about 5 to 10 percent of cancers begin with an inherited high-risk variant. When they do (BRCA1/2, Lynch syndrome, familial adenomatous polyposis), the inherited copy is a head start: every cell already carries one of the several hits required, so cancers appear earlier, in multiple sites, and across generations.

What is actually testable today

In short: The gap between what can be found and what can be done about it is the ethical core of genetic testing.

The gap between what can be found and what can be done about it is the ethical core of this field. Testing a healthy 25-year-old for Huntington's returns certain, unalterable information about their fifties. Roughly 10 to 25 percent of at-risk people choose to be tested, and that choice is a legitimate one either way.

Sources and notes

Molecular genetics and inheritance patterns are standard (Alberts, Molecular Biology of the Cell; Nussbaum, Thompson & Thompson Genetics in Medicine). BRCA1 penetrance figures are from Kuchenbaecker et al. (2017, JAMA), a large prospective cohort. APOE4 effect sizes vary by ancestry and are given as commonly cited ranges. The hereditary share of cancer (5 to 10 percent) is the standard estimate from cancer genetics literature. PKU newborn screening began with Robert Guthrie's test in 1963. Lactase persistence convergent evolution: Tishkoff et al. (2007). Polygenic score portability across ancestries: Martin et al. (2019, Nature Genetics).

Open questions. How much of the "missing heritability" for common diseases lies in rare variants, gene interactions, or measurement noise is unresolved. The clinical utility of polygenic risk scores outside research settings is actively debated.

Next: what happens after you swallow the pill. 👉

How Medicines Actually Work

TL;DR. A drug is a molecule shaped to stick to one particular protein and change what it does. That is the whole idea. Everything else (why you take it twice a day, why it takes six weeks to work, why it upsets your stomach, why grapefruit is a problem, why the dose is lower for your grandmother) follows from two questions: what does the drug do to the body, and what does the body do to the drug. Side effects are not accidents or impurities. They are the same mechanism acting where you did not want it, and understanding that turns a scary leaflet into a predictable list.

Key takeaways

• Most drugs work at one of four target types: receptors, enzymes, ion channels, or transporters. Anti-infectives are the exception, aiming at structures the pathogen has and you do not.
• Half-life determines dosing frequency. It takes about 4 to 5 half-lives to reach steady state, which is why some drugs work in an hour and others take weeks.
• Side effects have exactly three sources: the same target in the wrong tissue, binding to an unintended target, and downstream consequences of the intended effect.
• The therapeutic index (the gap between the effective dose and the toxic dose) is why some drugs need blood monitoring and most do not.
• The liver's CYP450 enzyme family metabolises most drugs, and it is where the majority of drug interactions happen, grapefruit included.
• Biologics (the drugs ending in -mab) are engineered antibodies. They are large, must be injected, cost far more than pills, and can hit targets small molecules cannot.

What a drug is

In short: Nearly every drug binds one of four things: a receptor, an enzyme, an ion channel, or a transporter.

A drug is a small molecule (usually 200 to 600 daltons, small enough to cross membranes and be swallowed) whose shape fits into a pocket on a protein. When it binds, it either mimics the natural signal, blocks it, or changes the protein's activity. Almost every drug in this book is one of the following.

That last row is the entire basis of selective toxicity, the reason antibiotics are safe and chemotherapy is not. Bacteria differ from us enormously, so bacterial-only targets are easy to find. Cancer cells are our own cells with a few changes, so classical chemotherapy attacks anything dividing fast and hits hair, gut lining, and bone marrow as collateral. Fungi are eukaryotes like us, which is why antifungals are comparatively harsh and few.

What the body does to the drug: ADME

In short: Absorption, distribution, metabolism, and excretion decide how much drug arrives and for how long, and half-life is what sets the dosing schedule.

Four processes decide how much drug reaches the target and for how long.

Absorption. A swallowed tablet must dissolve, survive stomach acid, and cross the gut wall. Then, crucially, all blood from the gut goes to the liver first, which may destroy much of the dose before it ever reaches the circulation. This is first-pass metabolism, and it is why some drugs are useless as tablets and must be injected, inhaled, or placed under the tongue. Nitroglycerin for angina goes under the tongue for exactly this reason. Bioavailability is the fraction of a dose that makes it into the bloodstream: 100 percent by definition for intravenous, often 20 to 80 percent by mouth.

Distribution. The drug spreads into tissues. Fat-soluble drugs accumulate in fat and linger; water-soluble ones stay in blood and fluid. The blood-brain barrier excludes most large or water-loving molecules, which is why some antihistamines make you drowsy (they cross) and newer ones do not (they do not), and why treating brain infections and brain tumours is so much harder than treating them anywhere else.

Metabolism. Mostly the liver, mostly a family of enzymes called cytochrome P450 (CYP). Their job is to make foreign molecules more water-soluble so kidneys can excrete them. Two consequences follow. First, drugs that induce or inhibit a CYP enzyme change the blood levels of every other drug using that enzyme, which is where most serious interactions come from. Grapefruit juice inhibits CYP3A4 in the gut wall, which can raise levels of certain statins and immunosuppressants enough to matter. Second, people carry different CYP variants, so identical doses give different blood levels, which is the practical basis of pharmacogenomics.

Excretion. Mostly kidneys, some in bile and stool. This is why kidney function governs the dosing of a huge range of drugs, and why an older person with half the kidney function of a 30-year-old needs a smaller dose of the same medicine to reach the same blood level.

Half-life, and why the schedule is what it is

Half-life is the time for blood concentration to fall by half. It determines two things.

Dosing interval: a drug with a 4-hour half-life dosed once daily would spend most of the day below the useful level, so it is dosed three or four times a day or reformulated as a slow-release tablet.

Time to steady state: with regular dosing, levels rise and plateau after about 4 to 5 half-lives. That plateau is when the full effect appears.

This is the honest explanation for something patients are often told without reason: antidepressants "take a few weeks to work." Part of that is pharmacokinetics and part is slower biological adaptation, but the practical rule stands. Judging an antidepressant after four days is judging it before it has arrived.

What the drug does to the body: dose and response

Plot effect against dose and you get an S-shaped curve. Below a threshold, nothing. Then a steep region where small changes matter a lot. Then a plateau where more drug adds effect only in side effects. Doubling a dose on the plateau buys toxicity, not benefit, which is the reason maximum doses exist.

Plot the toxicity curve alongside it and the gap between the two is the therapeutic index. A wide gap (penicillin, most modern drugs) means dosing can be approximate. A narrow one means small errors are dangerous, and those drugs get blood level monitoring: warfarin, lithium, digoxin, phenytoin, gentamicin, and many chemotherapies.

Why side effects exist

In short: There are only three causes: the intended target in the wrong tissue, an unintended target, or a logical consequence of the effect you wanted.

There are exactly three reasons, and every leaflet entry is one of them.

1. The intended target, in an unintended tissue. Receptors are not confined to the organ you care about. Beta blockers slow the heart via beta-1 receptors, and also hit beta-2 receptors in the airways, so they can worsen asthma. Antihistamines block histamine in the nose and also in the brain, so old ones cause sedation. Anticholinergics used for bladder or Parkinson's symptoms dry the mouth, blur vision, and cloud thinking, because the same receptor type runs all of it.

2. Off-target binding. No molecule is perfectly selective. Many antipsychotics bind histamine and muscarinic receptors as well as the dopamine receptors they are aimed at, producing sedation and weight gain that have nothing to do with the therapeutic effect.

3. Downstream consequences of the intended effect. This is the most logical category and the easiest to predict. Aspirin blocks COX-1, which reduces pain and clot formation, and also reduces the prostaglandins that protect the stomach lining, hence ulcers and bleeding. SGLT2 inhibitors push glucose into the urine to lower blood sugar, and sugary urine promotes genital yeast infections. Immunosuppressants for rheumatoid arthritis quiet the immune attack on your joints and also on tuberculosis.

Don't be confused: a side effect and an allergy are not the same thing. Nausea from an antibiotic is a side effect: predictable, dose-related, usually manageable. A true drug allergy is an immune reaction (hives, swelling, wheeze, anaphylaxis), is not dose-related, and can be worse on re-exposure. The distinction is not pedantry. Being wrongly labelled "penicillin allergic" pushes patients toward broader, more toxic, less effective antibiotics for life, and most such labels turn out to be wrong when properly tested.

The other kinds of medicine

In short: Beyond small molecules there are engineered antibodies, peptides, vaccines, gene therapies, and living cells, and the -mab ending marks an antibody.

Not everything is a small molecule.

The -mab suffix is worth learning because it appears in nearly every modern treatment chapter: adalimumab for autoimmune disease, trastuzumab for HER2 breast cancer, pembrolizumab for cancer immunotherapy, omalizumab for severe asthma. Each is an antibody engineered to grip one human protein.

Tolerance, dependence, and stopping

Tolerance is needing more drug for the same effect, because the body adapts (for example by reducing receptor numbers). Physical dependence is the body having adjusted so that abrupt withdrawal causes symptoms. Neither is the same as addiction, which is compulsive use despite harm (Chapter 43). People taking opioids or benzodiazepines correctly for months develop dependence without addiction, and taper slowly for that reason.

Several drug classes are dangerous to stop abruptly for reasons that are not addiction at all. Beta blockers can cause a rebound surge in heart rate and blood pressure. Long term corticosteroids suppress the body's own cortisol production, so stopping suddenly can precipitate an adrenal crisis. Antiepileptics stopped abruptly can trigger seizures. The general rule is that if a drug has been taken daily for months, stopping is a plan, not a decision.

Placebo, nocebo, and why trials need controls

People improve after taking an inert pill: natural recovery, regression to the mean, expectation, and the ritual of care all contribute. Placebo response is largest for subjective outcomes (pain, mood, nausea, fatigue) and negligible for objective ones (tumour size, blood glucose, mortality). Nocebo is the mirror image: told about possible side effects, people report them, including in placebo groups. In statin trials, muscle aches are reported at nearly identical rates on statin and on placebo, which does not mean any individual's pain is imaginary, but does explain why the drug gets blamed more often than it deserves.

This is the entire reason Chapter 17 exists: with effects this large and this ordinary, no one can tell whether a treatment works by trying it on a few people and watching.

Two practical problems that dominate real prescribing

Adherence. Roughly half of patients on long-term medication do not take it as prescribed, and the rate is worst exactly where the disease is silent: blood pressure, cholesterol, early diabetes. Adherence beats drug choice as a determinant of outcome for these conditions. The fixes are unglamorous: fewer pills, once-daily dosing, combination tablets, no-cost supply, and an explanation of what the pill is for.

Polypharmacy. Treating five diseases by five guidelines gives an older patient ten or more medicines, with interactions no single guideline anticipated. The countermove is deprescribing: periodically reviewing whether each drug still has a purpose, and stopping what does not. Some of the clearest benefit in geriatric medicine comes from subtraction.

Sources and notes

Standard pharmacology (Rang and Dale, Pharmacology; Katzung, Basic and Clinical Pharmacology; Goodman & Gilman). Half-life values are typical adult figures and vary with age, kidney and liver function, and formulation. The 4 to 5 half-lives rule to steady state is a mathematical property of first-order elimination. Nocebo and statin muscle symptoms: the SAMSON n-of-1 trial (Howard et al., 2021, JACC) and blinded re-challenge studies. Penicillin allergy delabelling: multiple studies find that around 90 percent of patients labelled penicillin-allergic tolerate it on formal testing. Adherence estimates are WHO figures for chronic therapy in developed countries.

Open questions. How much of the placebo response is genuine physiology versus measurement artefact remains debated, as does the clinical value of routine pharmacogenomic testing outside specific drugs.

Next: how anyone knows whether any of this works. 👉

How We Know Any of This

TL;DR. Medicine's hardest problem is not finding treatments, it is telling whether one worked. People get better on their own, get worse on their own, and differ from each other in a thousand ways that also predict their outcome. The randomised controlled trial exists to solve exactly one problem: making the treated group and the untreated group identical in every respect except the treatment. Every weaker form of evidence (observational studies, expert opinion, "I tried it and felt better") is vulnerable to confounding, and the history of medicine is a long list of treatments that looked obviously effective and were not. Some of them killed people at scale.

Key takeaways

• Correlation is not causation is a slogan; confounding is the mechanism. If the people who take a treatment differ from those who do not, the comparison is broken before it starts.
• Randomisation is the only method that balances the confounders you never thought of. That is the whole reason it is the gold standard.
• Relative risk reduction ("cuts risk 50 percent") is nearly meaningless without the baseline. Ask for the absolute reduction and the number needed to treat.
• A surrogate endpoint (a lab value) is not an outcome (a death, a stroke, a fracture). Drugs that improved surrogates and killed patients are the reason this distinction is written in blood.
• Screening statistics are systematically distorted by lead-time bias, length bias, and overdiagnosis. Five-year survival is the wrong measure for screening and is quoted for it constantly.
• A test's usefulness depends on how common the disease is. The same 99 percent accurate test is excellent for a common condition and misleading for a rare one.

The hierarchy of evidence, and what each level can and cannot do

In short: Case reports raise alarms, cohort studies show associations, and only randomisation can demonstrate cause.

The reason case reports still matter is that they are how the system notices. The AIDS epidemic was first visible as a June 1981 CDC report of five cases of an unusual pneumonia in previously healthy young men in Los Angeles. It proved nothing and it started everything.

Confounding, in one example

In short: Moderate drinkers looked healthier largely because the comparison group included people who had stopped drinking because they were already ill.

Observational studies found for years that people who drank moderately had lower mortality than people who did not drink at all. The obvious conclusion is that a glass of wine is protective. The problem is the comparison group. "Non-drinkers" includes former heavy drinkers who quit because they were ill, people too sick to drink, and people avoiding alcohol for religious or health reasons that correlate with everything else. Once studies separated lifelong abstainers and adjusted properly, and once genetic (Mendelian randomisation) analyses used alcohol-metabolism variants as a natural experiment, most of the protective effect disappeared.

That is confounding: a third factor that causes both the exposure and the outcome. It is not an occasional nuisance in observational research. It is the default condition.

Statistical adjustment helps and cannot finish the job, because you can only adjust for what you measured. Randomisation balances everything, measured and unmeasured, by construction. That is its entire and sufficient justification.

The most expensive lesson came from hormone replacement therapy. Observational studies through the 1990s consistently showed lower heart disease in women taking HRT, and it was prescribed widely on that basis. The Women's Health Initiative randomised trial, reported in 2002, found the combined therapy did not prevent heart disease and increased breast cancer and stroke. The observational studies had been comparing healthier, wealthier, more health-engaged women to everyone else. (The full picture is more nuanced than the 2002 headlines, with age at initiation mattering a great deal, which is its own lesson about how trial results get flattened in transmission.)

What makes a trial trustworthy

• Randomised, ideally with concealed allocation so no one can steer patients into arms.
• Blinded: single (patient), double (patient and clinician), or triple (plus analysts). Blinding matters most for subjective outcomes, and is impossible for some interventions, which is why surgical and behavioural trials are hard.
• Controlled against placebo, or against current best treatment where withholding it would be unethical.
• Analysed by intention to treat: everyone counted in the group they were assigned to, even if they stopped the drug. Analysing only completers reintroduces the confounding randomisation removed, because the people who quit are different.
• Preregistered: outcomes declared before the data are seen. Without this, a negative trial can be rewritten around whatever subgroup happened to look good, a practice common enough to have a name (HARKing).
• Powered and long enough to measure the outcome that matters.

Absolute vs relative: the number that gets hidden

In short: A 50 percent reduction can mean one percentage point, so ask for the absolute change and the number needed to treat.

A drug reduces heart attacks from 2 percent to 1 percent over five years.

• Relative risk reduction: 50 percent. True, and the number in the press release.
• Absolute risk reduction: 1 percentage point.
• Number needed to treat (NNT): 100. One hundred people take it for five years for one to avoid a heart attack. The other 99 get no benefit and all of the side effects.

Whether 100 is a good deal depends on the drug's cost, its harms, and the seriousness of the event prevented. For a cheap, well-tolerated pill preventing a heart attack, it is a good deal. For an expensive drug with serious toxicity preventing a mild symptom, it is not. None of that is decidable from the 50 percent figure alone.

Number needed to harm (NNH) is the same arithmetic for adverse effects, and a sensible treatment discussion is a comparison of the two.

This is the single most useful piece of numeracy a patient can carry into a consultation: "What is my absolute risk now, and what is it with treatment?"

Surrogates and hard endpoints

In short: Drugs that improved a lab value while killing patients are why a surrogate endpoint is a hypothesis until an outcome trial confirms it.

A surrogate endpoint is a measurable stand-in assumed to track the outcome you care about: blood pressure for stroke, LDL cholesterol for heart attack, HbA1c for diabetic complications, bone density for fractures, tumour shrinkage for survival. They make trials faster and cheaper, and they are sometimes wrong in ways that kill people.

The CAST trial (1989 to 1991) is the case every medical student learns. Extra heartbeats after a heart attack predict sudden death, and antiarrhythmic drugs suppressed them beautifully. The surrogate improved. When the drugs were finally tested against actual mortality, the treated patients died at roughly three times the rate of placebo. The trial was stopped early. Estimates of the excess deaths from the years of use beforehand run into the tens of thousands in the United States alone.

Other examples of the same failure mode: rosiglitazone lowered blood sugar and raised cardiovascular risk; torcetrapib raised "good" HDL cholesterol and increased mortality; high-dose oxygen, tight glucose control in critical illness, and several cancer drugs approved on tumour response later showed no survival benefit.

The rule that follows: a surrogate is a hypothesis until an outcome trial confirms it. When a chapter in this book says a treatment reduces deaths or strokes, that is a stronger claim than saying it improves a number.

Screening: why it needs its own statistics

In short: Lead-time bias, length bias, and overdiagnosis all inflate survival without saving anyone, so five-year survival is the wrong measure for screening.

Screening tests healthy people for early disease. It is intuitively obvious that finding cancer early must be good. It is often true and it is not automatically true, and three biases make screening look better than it is.

Lead-time bias. Diagnose a cancer three years earlier without changing the date of death, and survival time from diagnosis rises by three years. The patient gained nothing but three more years of knowing.

Length bias. Screening at intervals preferentially catches slow-growing tumours, because fast ones appear and cause symptoms between screens. So screen-detected cancers look more survivable, partly because they are a gentler selection of cancers.

Overdiagnosis. The extreme of length bias: finding disease that would never have caused symptoms in the person's lifetime. Every such case is counted as a life saved and is actually a patient given a diagnosis, treatment, and its harms for no benefit. Neuroblastoma screening in infants in Japan and Germany found many more tumours without reducing deaths and was stopped. Thyroid cancer diagnoses in South Korea rose roughly fifteen-fold after widespread ultrasound screening while mortality stayed flat.

The consequence: five-year survival is the wrong measure for screening, because all three biases inflate it. The right measure is disease-specific mortality in a randomised comparison, and by that standard some screening programmes are clearly worthwhile (cervical, colorectal), some are worthwhile with real trade-offs (mammography, lung CT in heavy smokers), and some have been abandoned.

Test accuracy and the base rate

In short: The same 99 percent accurate test is useful for a common disease and misleading for a rare one, because the population changed rather than the test.

Two properties describe a test. Sensitivity is the proportion of people with the disease it catches. Specificity is the proportion without it that it correctly clears. Neither answers the question a patient actually has, which is "I tested positive, do I have it?" That is the positive predictive value, and it depends on how common the disease is.

Take a test with 99 percent sensitivity and 99 percent specificity, applied to 10,000 people where 1 percent have the disease:

• 100 have it; 99 test positive.
• 9,900 do not; 99 test positive anyway (1 percent of 9,900).
• Total positives: 198. Half are wrong. PPV = 50 percent.

Now apply the same test where only 1 in 10,000 has the disease: 1 true positive and about 100 false positives. PPV under 1 percent. The test did not change. The population did.

This is why screening rare conditions in the general population produces mostly false alarms, why confirmatory testing exists, and why "the test is 99 percent accurate" is not an answer to anything on its own.

When randomising is impossible

Nobody will randomise people to smoke for thirty years. So how was smoking established as a cause of lung cancer? By meeting, across many observational studies, criteria articulated by Austin Bradford Hill in 1965:

Strength (smokers had many times the risk, not a few percent more), consistency (the same result in different countries, designs, and decades), temporality (smoking came first), biological gradient (more cigarettes, more cancer), plausibility and coherence (carcinogens in smoke, changes visible in airway tissue), experiment (quitting lowers risk; animal exposure causes tumours), and specificity where applicable.

No single criterion proves cause. Together, and with a dose-response relationship this steep, they are conclusive. The same framework now applies to air pollution, asbestos, alcohol and cancer, and lead and cognitive development.

Reading a health headline without being fooled

In short: Eight questions dispose of most health headlines, starting with whether the study was in humans and whether it was randomised.

A short checklist that resolves most of them:

1. Humans? Mice and cell cultures are not people, and most mouse results do not replicate in humans.
2. Randomised or observational? If observational, assume confounding until shown otherwise.
3. How many people, for how long? A 40-person 6-week study is a hypothesis.
4. Absolute numbers? If only percentages are given, the absolute effect is probably small.
5. Real outcome or surrogate?
6. Compared to what? Placebo, nothing, or the current best treatment (which is the only comparison that matters clinically).
7. Who funded it, and was it preregistered? Industry funding does not invalidate a trial, and it does predict which questions get asked and which results get published.
8. Has anyone replicated it? A single positive study is where science starts.

Don't be confused: "no evidence of benefit" is not "evidence of no benefit." A small trial that fails to find an effect may simply have been too small to see one. The two statements are routinely swapped in both directions, by enthusiasts and by sceptics.

Sources and notes

Evidence hierarchy and trial design: standard clinical epidemiology (Sackett et al., Evidence-Based Medicine; Guyatt et al., Users' Guides to the Medical Literature). CAST: the Cardiac Arrhythmia Suppression Trial, New England Journal of Medicine, 1989 and 1991. Women's Health Initiative: JAMA, 2002. The first AIDS report: MMWR, 5 June 1981. Bradford Hill criteria: Proceedings of the Royal Society of Medicine, 1965. South Korean thyroid cancer overdiagnosis: Ahn, Kim, and Welch, NEJM, 2014. Neuroblastoma screening: Schilling et al. and Woods et al., 2002. Alcohol and Mendelian randomisation: Millwood et al., The Lancet, 2019, among others. NNT figures here are illustrative round numbers, not guideline values.

Open questions. How much observational "real world evidence" can substitute for trials is a live methodological argument, sharpened by the availability of very large health datasets. The balance of benefit and harm in several screening programmes, prostate and breast especially, remains genuinely contested among reasonable experts.

That is the toolkit. Now the diseases, starting with the one that best rewards understanding it. 👉

Diabetes

TL;DR. Diabetes is a failure of the system that keeps blood sugar in a narrow range. In type 1, the immune system destroys the cells that make insulin, so there is no key for the lock; it is an autoimmune disease, usually starts young, and is fatal within weeks without injected insulin. In type 2, which is about 19 in every 20 cases, insulin is present but the body responds to it poorly and the pancreas eventually cannot keep up; it develops over years, silently. The high sugar itself rarely causes symptoms at first. What it does is slowly damage blood vessels and nerves, and that damage is where blindness, kidney failure, amputations, heart attacks, and strokes come from. It is not contagious, it is not caused by eating a dessert, and type 2 is now sometimes reversible.

Key takeaways

• About 589 million adults live with diabetes, roughly 1 in 9 adults worldwide, and about 252 million of them do not know it. It caused around 3.4 million deaths in 2024, roughly one every nine seconds.
• Type 1 and type 2 share a symptom and share almost nothing else. Treating them as one disease is the most common misunderstanding about diabetes and it hurts patients in both directions.
• High blood sugar damages the body mainly through small blood vessels (eyes, kidneys, nerves) and large ones (heart, brain, legs). Most of the harm is vascular.
• The complications are largely preventable, and the evidence for that is unusually strong: two landmark trials (DCCT in type 1, UKPDS in type 2) showed that lowering average glucose lowers complications, with benefits still visible decades later.
• Type 2 diabetes can go into remission. In the DiRECT trial, a structured weight-loss programme put 46 percent of participants into remission at one year, and remission tracked almost entirely with how much weight was lost.
• It is not contagious. There is no way to catch diabetes from another person.

What it is

In short: Chronically high blood glucose, diagnosed on a threshold, with HbA1c giving a three-month running average that cannot be improved by behaving well the week before an appointment.

Every cell in your body runs on glucose. Blood carries it, and the concentration has to stay inside a narrow band: too little and the brain fails within minutes, too much and the sugar slowly damages proteins throughout the body. A healthy adult holds blood glucose roughly between 70 and 140 mg/dL (3.9 to 7.8 mmol/L) all day, through meals, fasting, and exercise.

Diabetes mellitus is the state in which that control has failed and blood glucose runs chronically high (hyperglycaemia). It is diagnosed by any of the following, confirmed on a second occasion unless symptoms are obvious:

HbA1c deserves a moment, because it appears in every later section. Glucose sticks irreversibly to haemoglobin inside red blood cells, and the more glucose there has been, the more of it sticks. Red cells live about three months, so the percentage of glycated haemoglobin is a running average of blood sugar over roughly 8 to 12 weeks. It cannot be faked by behaving well the week before an appointment, which is exactly why it became the standard measure.

The types

Don't be confused: type 1 is not "the childhood one" and type 2 is not "the one you gave yourself." Roughly half of type 1 diagnoses occur in adults. Type 2 now appears in teenagers. And while excess weight is the dominant risk factor for type 2, the genetic contribution is large (larger than for type 1 by twin studies), which is why plenty of heavy people never develop it and plenty of slim people do. Telling someone their diabetes is their fault is both cruel and factually shaky.

The history

In short: Described 3,500 years ago, uniformly fatal until 1922, when insulin was isolated in Toronto and the patent sold for one dollar.

Diabetes is one of the oldest recorded diseases and one of the most recently made survivable.

An Egyptian medical text, the Ebers papyrus (about 1550 BCE), describes a condition of excessive urination. Around the second century CE, the Greek physician Aretaeus of Cappadocia named it diabetes, meaning "siphon" or "to pass through," describing patients as melting flesh into urine. Physicians in India and China independently noted that such urine attracted ants and flies. In 1675 Thomas Willis added mellitus, Latin for honey-sweet, having tasted it. In 1776 Matthew Dobson demonstrated that the sweetness came from actual sugar.

For the next 150 years there was no treatment. Type 1 diabetes was a death sentence, usually within a year of diagnosis, and the only intervention that bought time was near-starvation diets that left children skeletal.

The mechanism arrived in 1889, when Oskar Minkowski and Joseph von Mering removed a dog's pancreas and observed that it developed diabetes. Something in the pancreas controlled blood sugar.

In the summer of 1921 in Toronto, Frederick Banting, Charles Best, J.J.R. Macleod, and the biochemist James Collip extracted and purified that something. On 11 January 1922 they injected an extract into Leonard Thompson, a 14-year-old dying of diabetic ketoacidosis. The first, impure batch produced only an abscess. Collip's refined preparation, given twelve days later, drove his blood sugar down and revived him. He lived another 13 years. The discoverers sold the patent to the University of Toronto for one dollar each, on the principle that insulin belonged to the world, an intention the modern insulin market has not honoured.

The dates that follow matter because they explain the treatments in this chapter:

What actually goes wrong

In short: Type 1 destroys the insulin factory; type 2 is cells ignoring the signal until the factory wears out compensating.

The normal system

After a meal, glucose enters the blood from the gut. Beta cells in the pancreas sense the rise and release insulin, a small protein hormone. Insulin binds receptors on muscle, fat, and liver cells and does four things:

1. Muscle and fat: it triggers transporter proteins called GLUT4 to move to the cell surface, opening doors for glucose to enter. Without insulin, those doors are mostly closed.
2. Liver: it tells the liver to stop producing glucose and to store it as glycogen.
3. Fat tissue: it stops the breakdown of fat.
4. Overall, it signals "fed state: store, do not release."

Between meals, insulin falls and glucagon (from pancreatic alpha cells) rises, telling the liver to release stored glucose so the brain keeps running. The pair behaves like a thermostat with a heater and a cooler.

Type 1: the factory is destroyed

In type 1 diabetes, the immune system's T cells attack and destroy beta cells, over months to years, until insulin production essentially stops. Autoantibodies against beta cell proteins (GAD65, IA-2, insulin itself, ZnT8) are detectable in blood years before symptoms, which is what makes early screening and delay strategies possible. Risk is strongly linked to particular HLA immune genes, and something environmental (enteroviral infection is the leading suspect) appears to trigger it, though no trigger is proven.

Symptoms appear abruptly, usually once about 80 to 90 percent of beta cells are gone. With no insulin, muscle and fat cannot take glucose in, so blood glucose climbs while the cells starve. The liver, reading the absence of insulin as fasting, pours out more glucose, which makes it worse. Fat is broken down for fuel, producing acidic ketones. Blood turns acidic: diabetic ketoacidosis (DKA), which without treatment is fatal.

Type 2: the locks stop responding, then the factory wears out

Type 2 is a two-part failure over years.

Insulin resistance comes first. Muscle, liver, and fat cells respond weakly to insulin. The dominant driver is fat stored where it does not belong: inside liver and muscle cells, and in the abdomen around the organs (visceral fat), rather than under the skin. Fat inside these cells interferes with the insulin signalling cascade, and visceral fat secretes inflammatory cytokines that interfere further (the chronic inflammation of Chapter 13). Physical inactivity worsens it; muscle is the largest glucose sink in the body, and unused muscle takes up less.

The pancreas compensates, for a long time. Beta cells increase output, so blood glucose stays normal while insulin levels run two or three times high. This phase can last a decade or more with no symptoms and no abnormal glucose test.

Then compensation fails. Beta cells lose function and mass, from a mix of exhaustion, fat accumulation in the pancreas itself, and the toxic effect of high glucose and fatty acids on the cells themselves. Glucose rises into the prediabetic range, then the diabetic range. By the time type 2 diabetes is diagnosed, a substantial share of beta cell function is already lost, which is why it is often described as progressive.

The twin cycle model, developed by Roy Taylor's group, ties this together: excess fat in the liver drives excess fat export to the pancreas, and removing that fat, which weight loss does, can restart beta cell function. That model predicted, and the DiRECT trial confirmed, that type 2 diabetes can be reversed in many people.

What it does to the body

In short: High glucose damages small vessels in the eyes, kidneys, and nerves and large ones supplying heart, brain, and legs, so almost all the harm is vascular.

Persistent high glucose damages tissue through several parallel routes. Glucose sticks non-enzymatically to proteins, forming advanced glycation end products that stiffen collagen and blood vessel walls. Excess glucose entering the polyol pathway in cells that do not need insulin to take glucose up (nerve, retina, kidney, lens) consumes antioxidant capacity and produces osmotic stress. High glucose promotes inflammation and injures the lining of blood vessels.

The result is damage sorted into two families.

Microvascular (small vessel) damage is specific to diabetes:

Macrovascular (large vessel) damage is accelerated atherosclerosis (Chapter 21). Diabetes roughly doubles the risk of heart attack and stroke, and cardiovascular disease, not high sugar itself, is what most people with type 2 diabetes eventually die of.

The diabetic foot is where several of these meet, and it is worth understanding because it is the most preventable disaster in the disease. Neuropathy removes pain, so a blister or a stone in a shoe goes unnoticed. Poor circulation slows healing. High glucose impairs neutrophil function, so infection takes hold. An unnoticed ulcer becomes an infected one, then bone infection, then amputation. Diabetes remains the leading cause of non-traumatic lower limb amputation in most countries, and the majority of those amputations begin with an ulcer that a daily foot check would have caught.

Other effects: gastroparesis (damaged nerves to the stomach cause bloating and unpredictable digestion), erectile dysfunction (vascular and neural, often an early sign), frequent infections, and slow wound healing.

The two acute emergencies

Diabetic ketoacidosis (DKA), mostly type 1: high glucose, ketones, acidic blood, vomiting, deep rapid breathing, drowsiness, coma. Develops over hours to a day. Requires emergency intravenous fluids, insulin, and potassium.

Hyperosmolar hyperglycaemic state (HHS), mostly type 2 in older people: extremely high glucose (often above 600 mg/dL, 33 mmol/L) with severe dehydration and confusion, developing over days, with a higher death rate than DKA.

And the emergency created by treatment: hypoglycaemia, blood sugar too low, caused by insulin or sulfonylureas. Sweating, shaking, hunger, confusion, and at the extreme seizure and unconsciousness. It is treated with fast sugar, and it is the main reason tight control is not pushed to the limit in frail or elderly patients.

Is it deadly?

In short: About 3.4 million deaths a year, almost all through complications, and the trials show those complications are largely preventable.

Yes, though usually slowly and usually through its complications.

• About 3.4 million deaths in 2024 were attributed to diabetes, and that undercounts, since many people with diabetes die of heart disease or kidney failure and are recorded that way.
• Type 2 diabetes is associated with roughly 6 years of reduced life expectancy on average when diagnosed in middle age, and considerably more when diagnosed young.
• Type 1 diabetes, before insulin, was fatal within months to a couple of years. Today, with modern care, life expectancy is within several years of the general population and closing.
• Untreated DKA is rapidly fatal. With treatment, mortality in well-resourced settings is under 1 percent; in places without reliable insulin and emergency care, it remains a major killer of children.

The strongest counterpoint in the whole chapter: complications are not inevitable. The DCCT trial in type 1 showed that intensive glucose control cut retinopathy by 76 percent and nephropathy by around 50 percent, and its follow-up study found the treated group still had less heart disease decades later, a phenomenon named metabolic memory. UKPDS showed comparable benefits in type 2, and showed that controlling blood pressure mattered as much as controlling glucose.

Is it contagious?

No. Diabetes cannot be transmitted between people by any route: not by contact, saliva, blood, sex, sharing food, or living together. It is not caused by a microbe.

The confusion is worth addressing directly because it causes real harm. Families cluster for diabetes because they share genes and because they share meals, kitchens, and neighbourhoods, not because it spreads. Children with type 1 diabetes have been excluded from schools and camps by adults who believed otherwise, and adults with type 2 have been treated as contaminated. Neither has any basis.

The one genuine link between infection and diabetes runs the other way: high blood glucose impairs immune function, so people with poorly controlled diabetes catch infections more easily and recover from them more slowly, including tuberculosis, severe COVID-19, and skin infections.

Who gets it

In short: 589 million adults, over 250 million of them undiagnosed, with disease arriving at lower body weights in South Asian, East Asian, African, and Pacific populations.

Scale. 589 million adults aged 20 to 79, about 1 in 9. Projected to reach 853 million by 2050. About 252 million are undiagnosed, and in some regions more than half of all cases are undiagnosed. Global spending exceeded 1 trillion US dollars in 2024, around 12 percent of all health expenditure.

Geography. The largest absolute numbers are in China and India. The highest prevalence rates are in Pacific Island nations and in the Gulf states, where several countries exceed 20 percent of adults. Sub-Saharan Africa has the highest proportion of undiagnosed cases.

Ethnicity and body type. People of South Asian, East Asian, African, Hispanic, and Pacific Islander descent develop type 2 diabetes at lower body weights than people of European descent, at younger ages, and with less visible obesity. South Asian populations in particular carry more visceral and liver fat at a given BMI. This is why several countries use lower BMI screening thresholds for these groups. It is a real biological difference in fat distribution and beta cell reserve, not a difference in discipline.

Genes. Type 2 is highly polygenic, with over 500 identified risk loci (TCF7L2 being the strongest common one), and identical twin concordance is high. Type 1 is strongly linked to HLA genes but has lower twin concordance, meaning environment matters more for its onset than most people assume. Having a parent with type 2 diabetes roughly doubles to triples risk.

Sex and age. Type 2 rises steeply with age. Men develop it at slightly lower BMI than women. Gestational diabetes affects a substantial minority of pregnancies and identifies women at high subsequent risk of type 2.

Income. The gradient flipped over a century. Diabetes was once a disease of the wealthy; today, within high-income countries, it is concentrated among poorer people, tracking food environment, working conditions, stress, and access to care. Between countries, more than three-quarters of people with diabetes now live in low- and middle-income countries.

Treatment, and how each treatment works

In short: Type 1 requires insulin replacement; type 2 has a ladder of drug classes, two of which now prevent heart and kidney disease beyond anything they do to blood sugar.

Type 1: replace what is missing

There is no alternative to insulin, and the goal is to imitate a working pancreas.

Basal-bolus insulin uses two kinds: a long-acting insulin (glargine, detemir, degludec) once or twice daily to cover the liver's background glucose output, and a rapid-acting insulin (lispro, aspart, glulisine) with each meal, dosed to the carbohydrate eaten. The analogues are human insulin with small amino acid changes that alter how quickly the molecules clump and dissolve, and therefore how quickly they act.

Insulin pumps deliver rapid-acting insulin continuously through a small cannula, with programmable rates and mealtime boluses.

Continuous glucose monitors (CGM) measure glucose in the fluid under the skin every few minutes, showing not just a value but a direction, with alarms.

Closed-loop systems (the "artificial pancreas") connect the two: an algorithm reads the CGM and adjusts pump delivery automatically every few minutes. These are now standard care in many health systems and represent the largest practical improvement in type 1 management since the insulin pen.

Teplizumab is an antibody against the CD3 protein on T cells. Given to people in the early autoimmune stage before symptoms, it blunts the attack on beta cells and delays clinical type 1 diabetes by roughly two years on average. It does not prevent it.

Type 2: a ladder of mechanisms

The GLP-1 class gets a chapter of its own. Semaglutide (sold as Ozempic and Wegovy) and tirzepatide (Mounjaro and Zepbound) are the most prescribed and least understood drugs in modern medicine, and the row above compresses a great deal. What GLP-1 is, why it is called that, how these drugs were engineered from a lizard peptide, their full safety profile, and the crucial fact that they cannot replace insulin are covered in Chapter 66.

The order matters. Historically the sequence was metformin, then sulfonylurea, then insulin, chosen by glucose lowering alone. It has changed, because outcome trials showed that SGLT2 inhibitors and GLP-1 agonists prevent heart attacks, heart failure, kidney failure, and death to a degree not explained by their glucose lowering. In someone with diabetes plus heart or kidney disease, those two classes are now recommended regardless of how good the HbA1c already is, which is a genuine shift from treating a number to treating an outcome.

Metabolic surgery

Gastric bypass and sleeve gastrectomy produce diabetes remission in a large share of patients, often within days, before meaningful weight loss has occurred, through changes in gut hormone signalling as well as through later weight loss. Randomised trials consistently show better glucose outcomes than medical therapy in people with obesity and type 2 diabetes.

Treating the rest of the risk

Because most people with type 2 diabetes die of vascular disease, complete treatment is not just glucose:

• Blood pressure control, usually with an ACE inhibitor or ARB, which also protects the kidneys.
• A statin, for nearly everyone with diabetes over 40, because diabetes raises cardiovascular risk into the range where statins clearly pay.
• Annual retinal photography, which catches treatable retinopathy before vision changes.
• Annual kidney testing (urine albumin and blood creatinine).
• Annual foot examination, plus daily self-inspection.

What treatment costs

In short: Each class has a characteristic price, from metformin's diarrhoea to insulin's hypoglycaemia, and most are managed by dose, timing, or switching.

What the person can do

In short: Weight loss can put type 2 diabetes into remission, and exercise moves glucose into muscle without needing insulin at all.

This section is where diabetes differs most from other chapters, because the daily management belongs to the patient, and because several of these interventions are as powerful as the drugs.

Weight loss, for type 2, is the closest thing to a cure. In the DiRECT trial, a structured programme of total diet replacement followed by food reintroduction and maintenance produced remission (HbA1c below 6.5 percent off all diabetes medication) in 46 percent at one year and 36 percent at two years. The relationship with weight was almost linear: about 86 percent of those who lost 15 kg or more were in remission, versus about 7 percent of those who lost less than 5 kg. Remission is most achievable within the first few years after diagnosis, while beta cells can still recover.

Exercise works through a mechanism worth knowing. Contracting muscle moves GLUT4 transporters to the cell surface without needing insulin. That is why a walk after a meal lowers blood glucose even in someone with severe insulin resistance, and why exercise both lowers glucose acutely and improves sensitivity for a day or two afterwards. Resistance training adds muscle, which enlarges the tank glucose can go into. Guideline targets (about 150 minutes a week of moderate activity plus two resistance sessions) are useful, but the largest single gain is from doing something rather than nothing.

Diet. No single named diet has proven superior for everyone. What consistently helps: fewer refined carbohydrates and sugary drinks (they produce the sharpest glucose spikes), more fibre, and whatever pattern the person can sustain to produce an energy deficit if weight loss is the goal. Low-carbohydrate approaches reliably lower post-meal glucose and HbA1c in the short to medium term; Mediterranean-style patterns have the strongest cardiovascular outcome evidence. Anyone on insulin or sulfonylureas who changes their carbohydrate intake substantially needs their doses reviewed, because the same dose becomes an overdose.

Sleep and stress. Short sleep and untreated sleep apnoea worsen insulin resistance measurably. Sleep apnoea is common in type 2 diabetes and often undiagnosed.

Stop smoking. Smoking multiplies the vascular damage of diabetes; the two risks compound rather than add.

Daily foot check. Look at both feet, including between the toes and the soles, every day, especially if sensation is reduced. Shoes that fit. Never walk barefoot. This single habit prevents a large share of amputations.

Screening attendance. Eyes annually, kidneys annually, feet annually. All three detect damage during the window when it is still treatable and before it can be felt.

Vaccination. Diabetes raises the risk of severe outcomes from influenza, pneumococcal disease, and COVID-19, and these vaccines are recommended accordingly.

For anyone on insulin or sulfonylureas: carry fast-acting sugar, know the early symptoms of hypoglycaemia, and make sure someone at home and at work knows what to do.

Living with it

In short: The treatment is delegated to the patient, dozens of decisions a day, and the resulting burnout has a name and predicts worse control.

Diabetes is unusual in how much of the treatment is delegated to the patient. Someone with type 1 makes dozens of decisions a day about food, dose, activity, and correction, each with immediate consequences. The term diabetes distress describes the resulting burnout, which is common, distinct from clinical depression, and predicts worse glucose control. Depression itself is roughly twice as common in people with diabetes as without, in both directions of causation.

The financial burden is severe where insulin is not affordable. Insulin rationing has been documented in high-income countries with weak price controls, and it kills people; the resulting DKA is exactly the death the 1922 discovery was meant to end.

Two social points worth stating plainly. Type 1 and type 2 being conflated means people with type 1, an autoimmune disease no behaviour caused, absorb blame aimed at type 2. And blame aimed at type 2 is itself poorly founded and counterproductive, since shame is not a documented driver of sustained behaviour change while access, cost, and support are.

What's next

• Stem-cell derived islet transplantation. Lab-grown insulin-producing cells transplanted into people with type 1 have restored insulin production and eliminated insulin injections in early trials. The remaining obstacle is immune rejection, which currently requires immunosuppression; encapsulation and gene-edited hypoimmune cells are the two routes being pursued.
• Once-weekly insulin, reducing basal injections from 365 to 52 a year.
• Oral GLP-1 drugs, moving the most effective class from injection to tablet.
• Better closed-loop systems, including fully automatic ones that need no meal announcement, plus dual-hormone pumps delivering glucagon as well as insulin.
• Earlier interception of type 1, combining autoantibody screening of children with immune therapies like teplizumab.
• AI retinal screening, already deployed, where a camera and an algorithm perform the annual eye screen without a specialist, which matters most where specialists are scarce.

Sources and notes

Prevalence, undiagnosed proportion, mortality, and expenditure figures are from the International Diabetes Federation, IDF Diabetes Atlas, 11th edition (2025), reporting 2024 estimates: 589 million adults aged 20 to 79 (1 in 9), 252 million undiagnosed, 3.4 million deaths, over 1 trillion USD, and a projection of 853 million by 2050. Diagnostic thresholds are the American Diabetes Association Standards of Care, consistent with WHO criteria. DCCT: NEJM 1993, with EDIC follow-up. UKPDS: The Lancet 1998. DiRECT remission figures: Lean et al., The Lancet 2018 and Lancet Diabetes & Endocrinology 2019 (46 percent at 12 months, 36 percent at 24 months; remission by weight-loss band as cited). Twin cycle hypothesis: Taylor et al. Cardiovascular and renal outcome data for SGLT2 inhibitors and GLP-1 agonists come from EMPA-REG OUTCOME, LEADER, SUSTAIN-6, DAPA-HF, CREDENCE, and related trials. Teplizumab delay: Herold et al., NEJM 2019. Insulin discovery history: Bliss, The Discovery of Insulin. Life expectancy reduction estimates vary by cohort and age at diagnosis and are given as approximations.

Open questions. How durable type 2 remission is beyond five years, whether very early intensive treatment changes the disease trajectory permanently, and whether the muscle loss accompanying rapid GLP-1-induced weight loss has long-term consequences are all unresolved. The environmental trigger for type 1 remains unidentified.

Next: the condition that sits upstream of most type 2 diabetes, and is the most argued-about diagnosis in medicine. 👉

Obesity and Metabolic Syndrome

TL;DR. Obesity is excess body fat in an amount that damages health. The arithmetic is energy in versus energy out, and that is where most explanations stop and go wrong, because body weight is not a bank balance, it is a regulated system. The brain defends a weight the way it defends body temperature, using hormones from fat, gut, and pancreas, and it defends the highest weight reached far more vigorously than it resists gaining. That biology explains almost everything that confuses people about obesity: why deliberate weight loss usually stalls and reverses, why willpower is a poor predictor of outcome, and why drugs that act on the brain's appetite circuits have suddenly achieved what decades of advice could not.

Key takeaways

• About 1 in 8 people worldwide live with obesity, and roughly 2.5 billion adults are overweight or obese. Prevalence has more than doubled since 1990.
• BMI is a screening tool for populations, not a diagnosis for individuals. It cannot distinguish muscle from fat and it misclassifies athletes, older adults, and several ancestries.
• Where fat sits matters more than how much there is. Visceral fat around the organs and fat inside the liver and muscle drive metabolic damage; fat under the skin on hips and thighs is comparatively benign.
• After weight loss, the body reduces energy expenditure below what its new size predicts and raises hunger hormones. This metabolic adaptation persists for years and is the main mechanical reason diets fail.
• Obesity raises the risk of type 2 diabetes, cardiovascular disease, at least 13 cancers, fatty liver disease, sleep apnoea, and osteoarthritis. Roughly 5 million deaths a year are attributed to high BMI.
• GLP-1 based drugs changed the field. Semaglutide achieves about 15 percent average weight loss and tirzepatide about 20 percent, versus 5 to 8 percent for intensive lifestyle programmes and 25 to 30 percent for surgery.

What it is

In short: BMI is a population screening tool used as an individual diagnosis, and where the fat sits matters more than how much of it there is.

Obesity is excess adiposity sufficient to impair health. Measuring it is where the trouble starts.

Body mass index (BMI) is weight in kilograms divided by height in metres squared. It was devised in the 1830s by the Belgian statistician Adolphe Quetelet to describe populations, not to diagnose patients, and that is still what it is good for.

Its failures are systematic, not random. It counts muscle as excess weight, so athletes are misclassified. It misses fat gained as muscle is lost with age, so an older person with a "normal" BMI can carry a high fat percentage. And it needs different cut-offs by ancestry: people of South and East Asian descent develop metabolic disease at lower BMI, so many countries use 23 for overweight and 27.5 for obesity in these groups.

Better single measures for individual risk are waist circumference and waist-to-height ratio (keeping waist under half your height is a serviceable rule), because they capture where the fat is. A 2025 Lancet Commission proposed splitting the diagnosis into preclinical obesity (excess fat, organs still working normally) and clinical obesity (excess fat that is already causing organ dysfunction), precisely to stop treating a number as a disease.

Metabolic syndrome is the clustering of the damaging features, diagnosed when three of these five are present: large waist circumference, raised triglycerides, low HDL cholesterol, raised blood pressure, and raised fasting glucose. The cluster is not a coincidence. All five follow from insulin resistance driven by fat in the wrong places.

Don't be confused: obesity and being unfit are not the same thing, and neither is a moral category. Cardiorespiratory fitness independently predicts mortality, and a fit person with obesity has substantially lower risk than an unfit person of the same weight. Weight is one input to health, not a summary of it.

The history

In short: Human genetics did not change; the food supply, portion sizes, and the physical activity built into ordinary life did.

For nearly all of human existence, food scarcity was the design problem, and the metabolic system was tuned to store energy whenever it was available. Obesity was rare and, in many societies, a marker of wealth.

The shift is recent and fast. Global obesity prevalence has roughly doubled to quadrupled since 1975, depending on the age group, with childhood obesity rising fastest. Nothing about human genetics changed in that window. What changed was the environment: cheap refined carbohydrate and vegetable oil, industrially formulated foods engineered for palatability, portion sizes, sugary drinks, cars, desk work, screens, and the near-elimination of incidental physical activity.

Medicine's classification followed slowly. The American Medical Association formally recognised obesity as a disease in 2013, a contested decision: supporters argued it unlocked treatment and insurance coverage, critics argued it medicalised a body size and that many people with high BMI are metabolically healthy. Both points have merit, which is what the 2025 preclinical/clinical split is trying to resolve.

What actually goes wrong

In short: Body weight is defended by a hormonal control system that resists loss far more vigorously than it resists gain.

Weight is regulated, not chosen

Fat tissue secretes leptin in proportion to its mass. Leptin acts on the hypothalamus to reduce hunger and permit energy expenditure. The stomach secretes ghrelin, which rises before meals and drives hunger. The gut releases GLP-1, PYY, and CCK after eating, which signal fullness. The pancreas releases insulin and amylin. The hypothalamus integrates all of it and adjusts appetite and energy use to defend a weight.

The system is asymmetric. It defends strongly against loss and weakly against gain, which makes evolutionary sense in a world of periodic famine and is a catastrophe in a world of continuous abundance.

Leptin's discovery in 1994 created enormous excitement: mice lacking leptin were grossly obese, and giving them leptin normalised them. Giving leptin to humans with common obesity did almost nothing, because people with obesity already have high leptin levels and have become leptin resistant. The brain reads the signal as absent. That failure was the field's most instructive dead end and it reframed obesity as a disorder of signalling rather than of storage.

Metabolic adaptation, the reason diets fail

Lose weight and three things happen. Resting energy expenditure falls, by more than the smaller body accounts for. Ghrelin rises and satiety hormones fall, so hunger increases. And muscle becomes more efficient, burning fewer calories for the same movement.

The clearest demonstration came from following contestants of the television programme The Biggest Loser. Six years after extreme weight loss, most had regained much of the weight, and their resting metabolic rates remained suppressed by hundreds of calories a day relative to predictions for their body size. The defence did not fade with time.

This is why "eat less, move more" is technically correct and practically insufficient. The behaviour is being fought by a control system that treats the loss as a threat.

What the environment contributes

A tightly controlled inpatient randomised trial by Kevin Hall's group at the US National Institutes of Health fed participants either ultra-processed or minimally processed diets, matched for calories, sugar, fat, fibre, and macronutrients, with unlimited eating. On the ultra-processed diet, people ate about 500 more calories a day and gained weight; on the minimally processed diet they lost it. Same person, same nutrients on paper, opposite outcome. Something about the form of the food (energy density, texture, eating rate, and how little it satisfies) drives intake independently of its nutrient label.

Other contributors with real evidence: short sleep (raises ghrelin, lowers leptin), chronic stress and cortisol, several drug classes (antipsychotics, some antidepressants, corticosteroids, insulin, sulfonylureas), and endocrine conditions such as hypothyroidism and Cushing's syndrome, which are rare but treatable.

Genes

Obesity is 40 to 70 percent heritable by twin studies, which does not mean fated (Chapter 15 explains why heritability is not a personal percentage). Most of it is polygenic: hundreds of variants, each small, mostly acting on brain appetite circuits rather than on metabolism. The FTO variant, the best known, shifts average weight by only a couple of kilograms.

Rare single-gene forms exist and are informative: leptin deficiency, leptin receptor mutations, and MC4R mutations, the last being the commonest monogenic cause at roughly 1 to 5 percent of severe early-onset obesity. Children with these have relentless hunger from infancy. Setmelanotide, a drug that activates the MC4R pathway downstream of the defect, works dramatically in specific genetic subtypes and not at all in common obesity.

The honest summary: genes largely determine who gains weight in an obesogenic environment, and the environment determines how many people do.

What it does to the body

In short: The damage is metabolic, mechanical, and neoplastic, with at least 13 cancers linked to excess body fat.

Fat tissue is an active endocrine organ, not padding. Visceral fat drains directly to the liver and secretes inflammatory cytokines. The consequences fall into three categories.

Metabolic

• Insulin resistance and type 2 diabetes. Obesity is the dominant modifiable risk factor; the mechanism is in Chapter 18.
• Metabolic dysfunction-associated steatotic liver disease (MASLD), formerly called non-alcoholic fatty liver disease. Fat accumulates in liver cells, in some people progressing to inflammation (MASH), then fibrosis, cirrhosis, and liver cancer. It is now among the leading causes of liver transplantation.
• Dyslipidaemia: high triglycerides, low HDL, and small dense LDL particles.

Mechanical

• Obstructive sleep apnoea: fat in the neck and tongue collapses the airway during sleep, causing repeated oxygen drops, unrefreshing sleep, daytime sleepiness, and raised blood pressure. It is common and frequently undiagnosed.
• Osteoarthritis, mostly of knees and hips, from load plus inflammatory signalling.
• Gastro-oesophageal reflux, and hernias.

Vascular and neoplastic

• Hypertension, coronary disease, heart failure, atrial fibrillation, and stroke.
• Cancer. The International Agency for Research on Cancer identifies sufficient evidence linking excess body fat to at least 13 cancers, including oesophageal adenocarcinoma, colorectal, postmenopausal breast, endometrial, kidney, liver, pancreatic, gallbladder, ovarian, thyroid, gastric cardia, meningioma, and multiple myeloma. Mechanisms include insulin and IGF-1 signalling, oestrogen produced by fat tissue, and chronic inflammation.

Plus reduced fertility in both sexes, pregnancy complications, and higher rates of depression, with causation running in both directions and stigma contributing to the mental health burden independently of the physiology.

Is it deadly?

Yes, indirectly and at scale. High BMI is estimated to contribute to roughly 5 million deaths a year, overwhelmingly through cardiovascular disease, diabetes, and cancer. Severe obesity (BMI above 40) is associated with a reduction in life expectancy on the order of 8 to 10 years, comparable to lifelong smoking.

Two caveats keep this honest. The relationship with mortality is J-shaped, and the lowest-risk BMI in older adults is higher than in the young. And a subset of people with obesity show no metabolic abnormality for decades, though the risk of transitioning into metabolic disease over time remains higher than for people without obesity.

Is it contagious?

No. Obesity is not an infection and cannot be transmitted.

There is one genuinely interesting finding that is often misreported as contagion. In the Framingham social network analysis (Christakis and Fowler, 2007), obesity appeared to cluster through social ties, with a person's chance of becoming obese rising when a close friend did. The proposed mechanism is social: shared norms about portion size, activity, and acceptable body weight. The study has been criticised on methodological grounds (shared environments and friend selection are hard to separate from influence). It is a story about culture, not transmission.

The gut microbiome is likewise sometimes described as making obesity "transmissible" because transferring gut bacteria between mice transfers some weight phenotype. In humans the effect is far smaller and not established as causal.

Who gets it

In short: One person in eight worldwide, rising fastest in low- and middle-income countries, and socially patterned in opposite directions depending on national income.

• Scale. In 2022, about 2.5 billion adults were overweight, of whom about 890 million had obesity: 1 in 8 people alive. Over 390 million children and adolescents aged 5 to 19 were overweight, including 160 million with obesity.
• Fastest growth is now in low- and middle-income countries, particularly in urban areas, producing the double burden of undernutrition and obesity coexisting in the same country and sometimes the same household.
• Highest prevalence is in Pacific Island nations (exceeding 50 percent of adults in several), the Gulf states, and the United States (around 40 percent of adults).
• Income gradient. In high-income countries, obesity is more common among poorer people, tracking the cost of calories per dollar, food retail environments, working hours, and neighbourhood walkability. In low-income countries the gradient is often reversed, though it flips as countries get richer.
• Sex and age. Prevalence is somewhat higher in women globally, with wide regional variation. It rises through middle age and declines in the very old.

Treatment, and how it works

In short: Lifestyle programmes achieve 5 to 8 percent, GLP-1 based drugs 15 to 22 percent, and surgery 25 to 30 percent sustained over a decade.

Treatment intensity is matched to risk, not to appearance. The goal is health outcomes, and the amount of weight loss needed differs by outcome: about 3 to 5 percent improves glucose and triglycerides, 5 to 10 percent improves blood pressure and sleep apnoea, and 10 to 15 percent or more is where diabetes remission and MASH improvement appear.

Lifestyle programmes. Structured, supported programmes with regular contact produce average losses of 5 to 8 percent at a year, with substantial regain thereafter unless support continues. That average conceals wide variation: some people do far better. The key evidence point is that intensive programmes work considerably better than advice alone, and that maintenance contact is what preserves the result.

Medications

These drugs are covered in full in Chapter 66, including how they were engineered, all five of their mechanisms, and the complete safety picture: the vomiting and dehydration pathway, the deaths linked to compounded products, the anaesthesia risk, and why they are not a substitute for insulin.

The GLP-1 class matters beyond the numbers for two reasons. It works on the regulatory system itself, lowering the defended weight rather than fighting it, which is why people describe hunger and "food noise" simply receding. And the SELECT trial showed semaglutide reduced major cardiovascular events by about 20 percent in people with obesity and established cardiovascular disease but without diabetes, moving these drugs from cosmetic framing to outcome-modifying therapy.

Metabolic (bariatric) surgery. Sleeve gastrectomy and Roux-en-Y gastric bypass produce 25 to 30 percent weight loss sustained over a decade or more, with high rates of diabetes remission and, in long-term cohort studies, reduced mortality. The mechanism is only partly restriction; changed gut hormone signalling (GLP-1 and PYY rise sharply after bypass) and altered bile acid handling do much of the work, which is why glucose improves within days of surgery.

What treatment costs

• GLP-1 and dual agonists: nausea, vomiting, constipation, diarrhoea, all dose-related and usually improving. Gallstones with rapid loss. Rare pancreatitis. Loss of lean mass alongside fat, which matters most in older patients and is an active research target. Weight regain on stopping is substantial and rapid, because the drug was lowering the defended weight rather than curing anything, which makes these long-term therapies rather than courses. Cost and supply remain major access barriers.
• Orlistat: oily stools and urgency if fat intake is not reduced, plus reduced absorption of fat-soluble vitamins.
• Surgery: operative mortality is low in experienced centres (comparable to gallbladder surgery) but not zero. Long-term requirements include lifelong vitamin and mineral supplementation (B12, iron, calcium, vitamin D), risk of internal hernia and dumping syndrome, and a documented increase in alcohol use disorder after bypass.

What the person can do

In short: Protein, fibre, resistance training, sleep, and cutting liquid calories, undertaken with the expectation that the body will fight back.

Framed honestly, because this is a field saturated with confident bad advice.

• Protein and fibre first. Both increase satiety per calorie, and protein preserves muscle during weight loss, which protects resting metabolic rate.
• Cut liquid calories. Sugary drinks are the single most consistently implicated dietary item, because liquid calories produce little satiety.
• Reduce ultra-processed food where feasible. The Hall trial gives this a causal basis rather than a moralistic one.
• Resistance training, not only cardio. Exercise is a weak weight-loss tool on its own (it is easy to eat back the deficit) and a strong tool for preserving muscle, improving insulin sensitivity, and maintaining loss.
• Sleep 7 to 9 hours, and get snoring with daytime sleepiness assessed. Treating sleep apnoea improves energy, blood pressure, and often the ability to be active.
• Self-monitoring (weighing regularly, tracking intake) is among the most consistent predictors of maintenance in long-term registries.
• Expect defence, and plan for it. Weight loss plateaus are physiology, not failure. Maintenance is an active, ongoing task rather than a finish line.
• Ask about medication-driven gain. If weight rose sharply after starting a drug, alternatives often exist.

Living with it

Weight stigma is itself a health hazard. People with obesity report discrimination in healthcare, employment, and education, and experimental work shows clinicians spend less time with and give less information to patients with higher BMI. The measurable results are avoidance of medical care, delayed diagnoses, and worse mental health. Stigma does not produce weight loss; the evidence points the other way, toward increased eating and avoidance of exercise settings.

Weight cycling (repeated loss and regain) is common and demoralising, and the evidence on whether it independently harms health is mixed. What is clear is that repeated failure of a strategy that is fighting a regulatory system should update the strategy rather than the person's self-assessment.

What's next

• Oral GLP-1 drugs at doses matching injections, removing the needle barrier.
• Amylin analogues such as cagrilintide, alone and combined with semaglutide, in trials showing loss beyond current agents.
• Muscle-sparing combinations, pairing weight-loss drugs with myostatin or activin pathway agents to preserve lean mass.
• Better subtyping: identifying who has hunger-driven, satiety-driven, or emotional eating patterns, and matching the treatment.
• Policy: sugar-sweetened beverage taxes have measurably reduced purchases in several countries, and reformulation rules and marketing restrictions to children are where population-level change, as opposed to individual treatment, is likely to come from.

Sources and notes

Prevalence figures are WHO (2022 data, published 2024): 2.5 billion adults overweight, 890 million with obesity, 1 in 8 people, and the child and adolescent counts as given. Deaths attributable to high BMI are Global Burden of Disease estimates and vary by method. The Lancet Commission on Clinical Obesity (2025) proposed the preclinical/clinical distinction. Leptin: Zhang et al., Nature, 1994. Metabolic adaptation after major weight loss: Fothergill et al., Obesity, 2016 (Biggest Loser follow-up). Ultra-processed food trial: Hall et al., Cell Metabolism, 2019. Semaglutide weight loss: STEP trials (Wilding et al., NEJM, 2021). Tirzepatide: SURMOUNT-1 (Jastreboff et al., NEJM, 2022). Cardiovascular outcomes: SELECT (Lincoff et al., NEJM, 2023). Obesity-cancer links: IARC Handbooks of Cancer Prevention, volume 16. Social network clustering: Christakis and Fowler, NEJM, 2007, with subsequent methodological criticism. Asian BMI cut-offs: WHO expert consultation, 2004.

Open questions. Whether ultra-processed food harms through composition, texture, additives, or simply energy density is unresolved. Long-term (decade-plus) outcomes of GLP-1 therapy, including what happens to bone and muscle, are not yet known. Whether weight cycling independently harms health remains debated.

Next: the risk factor that kills more people than any other, and that almost nobody can feel. 👉

High Blood Pressure

TL;DR. Blood pressure is the force your blood exerts on artery walls. Run it too high for years and it damages the arteries themselves, then everything they supply: heart, brain, kidneys, eyes. It produces no symptoms until it has already caused harm, which is why it is called the silent killer and why roughly half the people who have it do not know. It is the largest single contributor to death worldwide, ahead of smoking. It is also cheap to detect, cheap to treat, and among the most thoroughly proven treatments in medicine: lowering it prevents strokes and heart attacks in direct proportion to how much you lower it.

Key takeaways

• About 1.28 billion adults aged 30 to 79 have hypertension. Roughly 46 percent do not know, and only about 1 in 5 have it under control.
• Two numbers: systolic (the peak, when the heart contracts) over diastolic (the trough, when it refills). Systolic matters more after about age 50.
• Risk rises smoothly from around 115/75 upward. The diagnostic line at 130/80 (US) or 140/90 (most other guidelines) is a treatment decision, not a biological edge.
• The damage is mechanical and cumulative: pressure injures the artery lining, thickens and stiffens vessel walls, forces the heart to work against resistance, and destroys the kidney's filtering units.
• Lowering systolic pressure by 10 mmHg cuts major cardiovascular events by roughly 20 percent, stroke by about 27 percent, and death by about 13 percent.
• Salt reduction, weight loss, potassium, exercise, and less alcohol each lower pressure measurably, and combined they rival a single drug.

What it is

In short: Two numbers, routinely measured badly in clinics, marking a risk that rises smoothly with no natural cut-off.

Each heartbeat pushes blood into the arteries, and the pressure inside them rises and falls with the cycle. Systolic pressure is the peak during contraction; diastolic is the resting level between beats. Both are quoted in millimetres of mercury (mmHg), a unit inherited from the mercury columns of early instruments.

Pressure is determined by two things: how much blood the heart pumps per minute (cardiac output) and how hard it is to push blood through the small arteries (peripheral resistance). Anything that raises either raises blood pressure.

Most of the world outside the United States still uses 140/90 as the diagnostic threshold, with treatment decisions modified by overall cardiovascular risk. Both approaches are defensible; the 2017 US change reclassified tens of millions of people without changing anyone's arteries, which is the clearest illustration of the threshold point made in Chapter 12.

How it is measured matters as much as the number. Blood pressure varies minute to minute with posture, talking, caffeine, bladder fullness, and anxiety. A single reading in a clinic proves little. Two artefacts are common enough to have names: white-coat hypertension (high in clinic, normal elsewhere, present in perhaps 15 to 30 percent of raised clinic readings) and masked hypertension (normal in clinic, high at home, which is the more dangerous error). This is why guidelines now recommend home or 24-hour ambulatory monitoring before committing someone to lifelong treatment.

Don't be confused: high blood pressure is not the same as feeling stressed, and it does not cause headaches at ordinary levels. Acute stress raises blood pressure temporarily; chronic hypertension is a sustained baseline elevation. And the belief that you can tell when it is high is not just wrong but dangerous, because people use it to decide when to take their tablets. Symptoms appear at crisis levels, and by then the disease has usually been present for years.

The history

In short: High blood pressure was thought to be a necessary compensation until trials in 1967 and 1970 were stopped early because the untreated group was having strokes.

Stephen Hales first measured arterial pressure in 1733 by inserting a brass tube into a horse's artery and watching blood rise nine feet in a glass column. The method was accurate and unrepeatable in patients.

The practical instrument came in stages: Scipione Riva-Rocci's inflatable arm cuff in 1896 gave systolic pressure, and in 1905 the Russian surgeon Nikolai Korotkov discovered that listening over the artery while deflating the cuff produced sounds whose appearance and disappearance marked systolic and diastolic pressure. The sphygmomanometer in every clinic is the direct descendant.

For the next fifty years, high blood pressure was widely considered essential, meaning necessary: physicians believed elevated pressure was a compensation required to push blood through stiffened arteries, and that lowering it would be harmful. The term "essential hypertension" survives as a fossil of that belief.

The turning point was political as well as medical. Franklin D. Roosevelt's blood pressure was recorded above 180/100 by 1937 and above 260/150 in 1945; he died of a cerebral haemorrhage that April, having received no effective treatment, because none existed and because his readings were regarded as unremarkable for his age.

Three things then changed everything. The Framingham Heart Study, started in 1948, followed a whole town for decades and established hypertension as a measurable predictor of stroke and heart disease, introducing the term "risk factor" itself. Effective drugs arrived from the 1950s onward. And the Veterans Administration Cooperative trials (1967 and 1970) randomised men with high blood pressure to treatment or placebo and were stopped early because the untreated group was having so many strokes and deaths. That settled it: treating high blood pressure saves lives.

What actually goes wrong

In short: Kidney salt handling, the renin-angiotensin cascade, sympathetic overdrive, and stiffening arteries, plus a specific curable cause in 5 to 10 percent of cases.

In roughly 90 to 95 percent of cases the cause is not a single identifiable lesion. This is primary (essential) hypertension, and it emerges from several interacting systems.

The kidney and salt handling. The kidney regulates long-term blood pressure by adjusting how much sodium and water it excretes. If the kidney requires a higher pressure to excrete a given salt load (the pressure-natriuresis relationship shifted rightward), pressure settles at a higher point. Sodium intake matters because of this relationship, and populations with very low sodium intake show almost no rise in blood pressure with age, unlike industrialised populations where a rise with age is so universal it was long assumed to be natural ageing.

The renin-angiotensin-aldosterone system (RAAS). When the kidney senses low perfusion, it releases renin, which converts a liver protein into angiotensin I, which the angiotensin-converting enzyme (ACE) converts to angiotensin II, a potent vessel constrictor that also triggers aldosterone release, causing the kidney to retain sodium and water. This cascade evolved to protect against blood loss and dehydration. In hypertension it runs inappropriately high, and it is the target of two of the most-used drug classes.

The sympathetic nervous system. The fight-or-flight system raises heart rate and constricts vessels. Chronic activation, driven by stress, obesity, and untreated sleep apnoea, keeps pressure elevated.

Arterial stiffening. With age, elastin in large arteries fragments and is replaced by stiffer collagen. A stiff aorta cannot cushion each heartbeat, so systolic pressure rises and diastolic often falls. That is why isolated systolic hypertension dominates after 60 and why a wide gap between the two numbers (pulse pressure) is itself a risk marker.

Secondary hypertension is the 5 to 10 percent with a specific cause, and finding it matters because it can be cured: primary aldosteronism (the commonest, and much underdiagnosed), kidney artery narrowing, chronic kidney disease, obstructive sleep apnoea, thyroid disease, phaeochromocytoma, coarctation of the aorta, and drugs including oral contraceptives, NSAIDs, decongestants, steroids, and stimulants. Clues are onset before 30 or after 55, very high pressure, sudden change, resistance to three drugs, or low blood potassium.

What it does to the body

In short: Pressure damages the arteries first and then everything they supply: heart, brain, kidneys, and eyes.

Sustained high pressure damages the arteries first, then the organs they feed. Every consequence below is a plumbing consequence.

Arteries. Pressure injures the endothelium (the single-cell lining), which accelerates atherosclerosis (Chapter 21). Vessel walls thicken and narrow. Weak spots balloon into aneurysms, most dangerously in the abdominal aorta and the brain. The aortic wall can split (aortic dissection), which is catastrophic and for which hypertension is the leading risk factor.

Heart. Pumping against higher resistance thickens the left ventricle (left ventricular hypertrophy). Thicker muscle needs more oxygen while its blood supply does not grow proportionally, and it becomes stiff, so the chamber fills poorly. The result is heart failure with preserved ejection fraction, plus a higher risk of atrial fibrillation and of heart attack.

Brain. Hypertension is the single largest risk factor for stroke, both ischaemic (a blocked artery) and haemorrhagic (a burst one). Deep small vessels in the brain are particularly vulnerable, and their gradual damage produces vascular dementia and contributes to mixed dementia (Chapter 37).

Kidneys. High pressure damages the glomeruli, the tiny filters. Damaged kidneys regulate pressure worse, which raises pressure, which damages more kidney. Hypertension is the second leading cause of end-stage kidney disease after diabetes (Chapter 23).

Eyes. Retinal arteries can be seen directly through the pupil, which makes them a window onto what is happening everywhere else. Hypertensive retinopathy progresses from narrowed arterioles to haemorrhages and, at crisis levels, swelling of the optic disc.

Hypertensive emergency is pressure above roughly 180/120 with evidence of acute organ damage: chest pain, breathlessness, neurological deficit, or visual loss. It requires immediate hospital treatment, with pressure lowered in a controlled way rather than abruptly.

Is it deadly?

In short: It is the largest single contributor to death worldwide, and each 10 mmHg reduction in systolic pressure cuts strokes by about a quarter.

It is the leading contributor to death on earth. High systolic blood pressure is consistently ranked by the Global Burden of Disease study as the single largest attributable risk factor for mortality, associated with on the order of 10 million deaths a year, mostly through heart disease and stroke.

The benefit of treatment is quantified better than almost anything in medicine. A meta-analysis of over 300,000 participants across many trials found that each 10 mmHg reduction in systolic pressure produced approximately:

The SPRINT trial went further, randomising higher-risk patients without diabetes to a systolic target below 120 rather than below 140, and found fewer cardiovascular events and deaths, at the cost of more low blood pressure, fainting, electrolyte abnormalities, and kidney function decline. That trade-off is the reason targets are individualised, tighter in a robust 60-year-old and looser in a frail 85-year-old at risk of falls.

Is it contagious?

No. Blood pressure is a property of your own circulation. Nothing about it can be transmitted.

Families share it for two reasons that both look like transmission: genetics (hypertension is polygenic and heritable) and a shared salt-heavy diet, since everyone eating from the same kitchen consumes the same sodium load.

Who gets it

In short: 1.28 billion adults, nearly half unaware they have it, in patterns that track salt, age, and living conditions far more than ancestry.

Age. Prevalence rises steeply: uncommon below 30, present in the majority of people over 65 in most countries. Arterial stiffening makes this partly a function of ageing, though not an inevitable one, as low-sodium populations demonstrate.

Sex. Higher in men until about age 55 to 60, then higher in women, whose pressure rises faster after menopause.

Ancestry and geography. Hypertension is more common, appears earlier, and causes more organ damage in people of African descent in the United States, the Caribbean, and the United Kingdom. The explanation is frequently oversimplified into genetics. The better supported picture combines salt sensitivity (higher on average, mechanism debated), socioeconomic conditions, food environment, healthcare access, and the measurable physiological effects of chronic exposure to racial discrimination, which raises blood pressure in longitudinal studies. Notably, blood pressure in West African populations is generally lower than in African-descended populations in the United States, which argues strongly against a purely genetic account.

Salt intake. Global average sodium consumption is roughly double the WHO recommended maximum of 2 grams of sodium (about 5 grams of salt) a day. Most of it, in industrialised diets, comes from processed and restaurant food rather than the salt shaker.

Other risk factors: obesity, alcohol, physical inactivity, low potassium intake, sleep apnoea, chronic kidney disease, and family history.

Control rates are the scandal. Of the 1.28 billion adults with hypertension, roughly 46 percent are unaware, about 42 percent are diagnosed and treated, and only around 21 percent are controlled. This is not a technology problem. The drugs are generic and cost pennies.

Treatment, and how each drug works

In short: Four cheap first-line classes, each blocking a different mechanism, and adding a second at low dose beats doubling the first.

Four first-line classes, each interrupting a different part of the pressure system.

Second-line and specific-indication drugs: beta blockers (slow the heart and reduce its output; no longer first line for uncomplicated hypertension but essential after a heart attack or in heart failure), spironolactone (blocks aldosterone; the single most effective add-on for resistant hypertension), alpha blockers, and hydralazine.

Three principles govern how they are used.

Combination beats escalation. Adding a second drug at a low dose lowers pressure roughly five times more than doubling the first, because you are blocking a second mechanism instead of pushing harder on one. This is why single-pill combinations are increasingly the starting point.

Match the drug to the patient. Someone with diabetes and protein in the urine gets an ACE inhibitor or ARB for kidney protection. Someone after a heart attack gets a beta blocker. Someone of African descent without those indications typically starts with a calcium channel blocker or thiazide, which work better on average in that group.

Resistant hypertension means pressure above target on three drugs including a diuretic. The first step is not a fourth drug: it is checking adherence, confirming with home readings, reviewing NSAIDs and other pressure-raising drugs, and screening for secondary causes, particularly primary aldosteronism and sleep apnoea.

Renal denervation, a catheter procedure burning sympathetic nerves in the renal artery walls, has a mixed trial history. Early enthusiasm collapsed when a properly sham-controlled trial found no benefit; better-designed later trials show modest reductions, and it is now approved for selected patients rather than as a general solution. It is a clean case study in why sham controls matter.

What treatment costs

In short: Side effects are mostly minor and class-specific, and are usually solved by switching drug rather than by stopping treatment.

The general point: first-line antihypertensives are cheap, generic, and well tolerated by most people, and side effects are usually solved by switching class rather than by abandoning treatment. The most common real-world problem is not toxicity, it is people stopping a drug that makes them feel no better because the disease it prevents is invisible.

What the person can do

In short: Diet, salt, potassium, weight, exercise, and alcohol each move blood pressure measurably, and combined they rival a drug.

Lifestyle change is not a token gesture here. The measured effects are large enough to delay or avoid medication in mild cases and to improve control in everyone.

Two practical additions. Measure at home, with a validated upper-arm cuff, seated, back supported, feet flat, arm at heart level, after five minutes of rest, twice in the morning and twice in the evening for a week. Home readings predict outcomes better than clinic readings and give you and your clinician something real to work with. Take the tablets, including on days you feel well, which is all of them.

Salt substitutes deserve a specific mention because the evidence is unusually strong. The SSaSS trial in rural China randomised nearly 21,000 people at high stroke risk to a potassium-enriched salt substitute (75 percent sodium chloride, 25 percent potassium chloride) instead of regular salt, and found significant reductions in stroke, cardiovascular events, and death. Swapping one product in the kitchen produced an outcome benefit. The caution is that people with advanced kidney disease or on potassium-raising drugs must not use them.

Living with it

The defining difficulty is that hypertension is a disease of numbers rather than sensations. Adherence to antihypertensives at one year runs at roughly 50 percent, and the reasons people give are consistent: no symptoms to relieve, side effects that are noticeable when the disease is not, cost, complexity, and no sense of progress. The countermeasures that work are equally consistent: single-pill combinations, once-daily dosing, home monitoring so the person can see their own numbers move, and an explanation of what is actually being prevented.

What's next

• Twice-yearly injectable therapy. Zilebesiran, an siRNA that silences the liver's production of angiotensinogen, the precursor of the whole RAAS cascade, has produced sustained pressure reduction from a single injection in phase 2 trials. If outcome trials confirm it, adherence stops being a daily problem.
• Polypills, combining a statin and two or three antihypertensives in one cheap tablet, have shown large event reductions in trials in low-resource settings.
• Population salt reduction, through mandatory reformulation targets, which is the intervention with the best cost-per-life-saved of anything in this chapter.
• Cuffless continuous monitoring in wearables, currently not accurate enough for diagnosis, but improving.

Sources and notes

Prevalence, awareness, and control figures are from WHO's Global report on hypertension (2023), which gives 1.28 billion adults aged 30 to 79 with hypertension, about 46 percent unaware, and around 21 percent controlled, based on NCD-RisC pooled analyses. Diagnostic categories are the 2017 ACC/AHA guideline; the 140/90 threshold is ESC/ESH and WHO. Treatment effect sizes per 10 mmHg: Ettehad et al., The Lancet, 2016. SPRINT: NEJM, 2015. Lifestyle effect sizes are from the 2017 ACC/AHA guideline's summary of trial evidence and are averages, not guarantees. SSaSS salt substitute trial: Neal et al., NEJM, 2021. Renal denervation: SYMPLICITY HTN-3 (NEJM, 2014) was the negative sham-controlled trial; SPYRAL HTN-OFF MED and RADIANCE trials were later positive. Zilebesiran: KARDIA-1 phase 2 results, 2024. Roosevelt's blood pressure history is documented in Bruenn's clinical notes and subsequent historical analyses. Blood pressure in West African versus African-American populations: the ICSHIB studies (Cooper et al.).

Open questions. The optimal systolic target in the very old and the frail is unsettled. Whether population-wide sodium reduction benefits everyone equally, given individual variation in salt sensitivity, is still argued, though the population-level case is strong.

Next: what that pressure, plus cholesterol, plus time, does to the arteries feeding the heart. 👉

Heart Disease

TL;DR. The heart is a pump with its own small dedicated blood supply, the coronary arteries. Over decades, cholesterol-carrying particles lodge in the walls of those arteries and provoke an inflammatory response that builds a plaque. A plaque that narrows the artery causes chest pain on exertion (angina). A plaque that cracks open triggers a clot that blocks the artery completely, and the heart muscle downstream begins dying within minutes: a heart attack. Muscle that dies becomes scar, and enough scar produces heart failure, a pump that can no longer keep up. This chain is the single largest cause of death in the world, and nine modifiable factors account for about 90 percent of the risk of a first heart attack.

Key takeaways

• Ischaemic heart disease is the world's leading cause of death, responsible for roughly 13 percent of all deaths.
• Atherosclerosis is not passive clogging. It is an inflammatory disease of the artery wall in which LDL particles get trapped, are engulfed by immune cells, and build a lipid core under a fibrous cap.
• Most heart attacks come from moderate-sized plaques that rupture, not from the tightest narrowings. That is why a "clean stress test" is not a guarantee and why drugs that stabilise plaques save lives.
• A heart attack and a cardiac arrest are different events. One is a plumbing problem, the other an electrical one. The treatments share nothing.
• Time is muscle. Opening a blocked artery within the first hour or two saves most of the territory; after about 12 hours, the muscle is largely gone.
• Nine factors (lipids, smoking, hypertension, diabetes, abdominal obesity, stress, diet, exercise, alcohol) accounted for about 90 percent of population-attributable risk of first heart attack across 52 countries in the INTERHEART study.

What it is

In short: One label covering stable angina, acute coronary syndromes, heart failure, and atrial fibrillation, which interlock but are not the same thing.

"Heart disease" covers several distinct conditions that interlock.

Coronary artery disease (CAD), also called ischaemic heart disease, is narrowing of the arteries that supply the heart muscle. Its manifestations:

• Stable angina: chest pressure, tightness, or heaviness on exertion, relieved by rest within minutes. The artery supplies enough blood at rest and not enough under load.
• Acute coronary syndrome: a plaque ruptures and a clot forms. This includes unstable angina (pain at rest, no muscle death yet), NSTEMI (partial blockage, some muscle death), and STEMI (complete blockage, a characteristic ECG pattern, the classic full-thickness heart attack).
• Myocardial infarction (heart attack) is diagnosed by a rise in blood troponin, a protein released by dying heart muscle, together with evidence of ischaemia.

Heart failure is the pump failing to meet the body's demands, whether from scar after infarction, long-standing high blood pressure, valve disease, or damaged muscle.

Arrhythmias are electrical disorders, most importantly atrial fibrillation.

Don't be confused: a heart attack is not a cardiac arrest. A heart attack is a blocked coronary artery starving muscle of oxygen; the person is usually conscious, in pain, and needs the artery opened. A cardiac arrest is the heart's electrical activity collapsing into a rhythm that pumps no blood; the person is unconscious, not breathing normally, and needs CPR and a defibrillator within minutes. A heart attack can cause a cardiac arrest, which is the main way heart attacks kill people before they reach hospital. Chest compressions are useless for a conscious person with chest pain and are the only thing that matters for someone in arrest.

The history

In short: The disease became understandable in 1912, survivable in the 1960s with coronary care units, and largely preventable once statins and risk factor control arrived.

Chest pain on exertion was described precisely by William Heberden in 1768, who named it angina pectoris without knowing its cause. The connection to blocked coronary arteries was suspected through the 1800s and established for clinical purposes by James Herrick in 1912, who described the syndrome of coronary thrombosis in living patients rather than at autopsy. For decades the treatment was bed rest for six weeks.

The modern picture assembled in the second half of the twentieth century:

Age-adjusted death rates from coronary disease in high-income countries have fallen by more than half since the 1970s. Roughly half of that fall is attributed to risk factor change (mostly smoking and cholesterol) and roughly half to treatments.

What actually goes wrong

In short: LDL particles get trapped in the artery wall and provoke an inflammatory plaque, and most heart attacks come from moderate plaques that rupture rather than from the tightest narrowings.

Atherosclerosis, step by step

1. Injury and entry. The artery lining (endothelium) is damaged by high blood pressure, smoke chemicals, high glucose, or turbulent flow at branch points. Injury makes it permeable and sticky.
2. Retention. LDL particles carrying cholesterol pass into the artery wall. The key event is that they are retained there, bound to the wall matrix. The more LDL particles in the blood, the more get trapped, which is why particle count (measured as apolipoprotein B, one per particle) predicts risk slightly better than cholesterol concentration.
3. Oxidation and alarm. Trapped LDL is chemically modified and the immune system treats it as damage.
4. Foam cells. Monocytes enter the wall, become macrophages, engulf the modified LDL, and become lipid-stuffed foam cells. Enough of them form a visible fatty streak, present in the arteries of many teenagers.
5. Plaque. Smooth muscle cells migrate over the lesion and lay down a fibrous cap over a soft core of lipid and dead cells. This is a mature atherosclerotic plaque.
6. Rupture. Inflammation inside the plaque thins the cap. When it tears, the thrombogenic core is exposed to flowing blood, platelets pile on, and a clot forms in minutes.

Step 6 explains the most counterintuitive fact in cardiology. Most heart attacks arise from plaques that were narrowing the artery by less than 70 percent, because soft-cored, thin-capped plaques are more rupture-prone than old, calcified, tight ones. That is why treating a tight narrowing with a stent relieves symptoms without necessarily preventing future heart attacks, and why drugs that lower LDL and calm inflammation across the whole arterial tree do prevent them.

Cholesterol, briefly and correctly

Cholesterol is essential: it is a component of every cell membrane and the precursor of steroid hormones and vitamin D. It is not water-soluble, so it travels in lipoprotein particles.

The causal role of LDL is one of the best-established facts in medicine, supported by randomised drug trials, by inherited conditions with very high LDL (familial hypercholesterolaemia, causing heart attacks in the thirties and forties), and by inherited variants that lower LDL for life and confer large reductions in heart disease.

Heart failure

When muscle dies or is chronically overloaded, the heart pumps less effectively. The body responds as though to blood loss: the sympathetic nervous system and the RAAS activate, retaining salt and water and constricting vessels. In the short term this maintains blood pressure. Sustained, it makes everything worse, driving fluid overload and further remodelling of the heart. This neurohormonal insight is why the drugs that improve survival in heart failure are ones that block those systems, not ones that stimulate the heart to beat harder.

Two forms:

• HFrEF (reduced ejection fraction): the ventricle is weak and dilated, ejecting under 40 percent of its volume per beat. Best treated, with four drug classes proven to extend life.
• HFpEF (preserved ejection fraction): the ventricle is stiff and fills poorly, often from hypertension, obesity, diabetes, and ageing. Roughly half of all heart failure, more common in women and older patients, and until recently without proven treatments.

Symptoms of both: breathlessness on exertion and lying flat, ankle swelling, fatigue, waking at night short of breath.

Atrial fibrillation

The atria (the heart's upper chambers) lose coordinated contraction and quiver at 300 to 600 impulses per minute, of which the ventricle receives an irregular subset. Two consequences. The pulse becomes irregular and often fast, which can cause palpitations, breathlessness, and, over time, a weakened heart. And blood pools in the left atrial appendage, forms a clot, and that clot can travel to the brain. Atrial fibrillation increases stroke risk roughly fivefold, and the strokes it causes are more severe than average.

What it does to the body

In short: Muscle starts dying within half an hour, becomes scar, and leaves both a weaker pump and an electrically unstable region.

• Angina limits activity, sometimes severely, without necessarily damaging muscle.
• A heart attack kills muscle at a rate that depends on collateral supply. Irreversible damage begins within about 20 to 30 minutes and is largely complete within 6 to 12 hours. The dead area becomes fibrous scar over weeks.
• Scar produces two long-term problems: a weaker pump (heart failure) and an electrically unstable region that can generate lethal arrhythmias.
• Sudden cardiac death is often the first presentation of coronary disease. A substantial share of people who die of a heart attack die before reaching hospital.
• The same disease elsewhere: the identical process in leg arteries causes peripheral artery disease (pain on walking, poor healing, gangrene in the extreme); in the neck and brain arteries it causes stroke (Chapter 22); in the kidney arteries it causes renal impairment. This is why a patient with one is screened for the others.

Is it deadly?

In short: The world's single leading cause of death, though survival after a heart attack has improved several-fold since the pre-coronary-care era.

Ischaemic heart disease is the leading single cause of death worldwide, responsible for around 13 percent of all deaths, roughly 9 million a year. Cardiovascular disease as a whole (heart disease plus stroke plus the rest) accounts for approximately 17 to 20 million deaths annually, about a third of all deaths.

The individual picture is better than those numbers suggest, and improving:

• In-hospital mortality after a STEMI has fallen from roughly 30 percent in the pre-coronary-care era to under 5 to 7 percent where primary angioplasty is available promptly.
• Out-of-hospital cardiac arrest survival to discharge remains poor, roughly 8 to 12 percent overall, and is several times higher when a bystander performs CPR and a defibrillator is used quickly.
• Heart failure carries a prognosis comparable to many cancers: roughly half of patients die within five years of diagnosis, though modern four-drug therapy has changed that trajectory substantially for HFrEF.

Is it contagious?

No. Atherosclerosis cannot be caught.

Three genuine connections to infection are worth knowing, because they get garbled. Rheumatic heart disease is caused by an infection: untreated streptococcal throat infection can trigger an autoimmune attack on heart valves, and it remains a major cause of valve disease in low-income countries, entirely preventable with antibiotics. Infective endocarditis is a bacterial infection of the heart valves themselves. And acute infections trigger heart attacks: influenza and COVID-19 measurably raise the risk of myocardial infarction in the weeks after infection, through inflammation and increased clotting. That last point is why influenza vaccination is recommended for people with heart disease and reduces cardiovascular events in trials.

Who gets it

In short: Nine modifiable factors account for roughly 90 percent of first heart attacks worldwide, and women are systematically diagnosed later than men.

Age and sex. Risk rises steeply with age. Men develop coronary disease roughly 7 to 10 years earlier than women on average, and the gap narrows after menopause. Women's heart disease is systematically underdiagnosed: symptoms more often include breathlessness, nausea, fatigue, and jaw or back discomfort rather than crushing central chest pain, and women are less likely to receive prompt angiography and guideline treatment. Heart disease kills more women than all cancers combined in most high-income countries.

The nine factors. The INTERHEART case-control study across 52 countries found that abnormal lipids, smoking, hypertension, diabetes, abdominal obesity, psychosocial stress, low fruit and vegetable intake, physical inactivity, and alcohol together accounted for about 90 percent of the population-attributable risk of a first heart attack, consistently across regions, ethnicities, and both sexes. The pattern is not a Western phenomenon.

Genetics. Family history of early heart disease is a real independent risk factor. Specific high-impact conditions: familial hypercholesterolaemia, affecting roughly 1 in 250 people, causing lifelong very high LDL and premature heart disease, badly underdiagnosed. Elevated lipoprotein(a), affecting about 1 in 5, genetically fixed, and only now becoming treatable.

Geography. Age-standardised cardiovascular mortality has fallen substantially in Western Europe, North America, Japan, and Australasia, and is high and in some places rising in Eastern Europe, Central Asia, South Asia, and parts of sub-Saharan Africa. South Asian populations develop coronary disease earlier and at lower BMI. Over three quarters of cardiovascular deaths now occur in low- and middle-income countries.

Treatment, and how it works

In short: Open the blocked artery as fast as possible, then lower LDL, blood pressure, and clotting for life, with four specific drug classes for heart failure.

The acute heart attack

The objective is to restore flow as fast as possible.

• Aspirin, chewed immediately, irreversibly blocks platelet COX-1 and reduces further clotting. ISIS-2 showed aspirin alone cut deaths by about 23 percent.
• A second antiplatelet drug (ticagrelor, clopidogrel, prasugrel) blocks a different platelet activation pathway.
• Primary percutaneous coronary intervention (PCI): a catheter is passed from the wrist or groin to the blocked artery, a balloon opens it, and a stent (a metal mesh tube, usually coated with a drug that prevents scar tissue regrowth) holds it open. This is the preferred treatment when it can be delivered quickly.
• Thrombolysis: clot-dissolving drugs, used where PCI cannot be reached in time. Effective, with a small risk of causing bleeding including in the brain.
• Oxygen only if oxygen levels are low, morphine for pain, and beta blockers and nitrates as appropriate.

The controlling variable is time. "Door-to-balloon" targets of under 90 minutes exist because the amount of muscle saved falls steeply with delay, and most of the delay in practice is the patient's, not the hospital's. This is why public campaigns focus on calling emergency services immediately rather than driving to hospital or waiting to see whether it passes.

Long-term prevention after a diagnosis

Revascularisation for stable disease

For stable angina, the ISCHEMIA trial randomised patients with moderate or severe ischaemia to an invasive strategy (angiography plus stenting or bypass) or optimal medical therapy alone, and found no reduction in death or heart attack from the invasive approach, though it did relieve angina better. The practical conclusion: stents relieve symptoms; drugs and risk factor control prevent events. Bypass surgery retains an advantage in specific anatomies, notably left main disease, multivessel disease with diabetes, and reduced pump function.

Heart failure

The four pillars for HFrEF, each shown to reduce mortality, and used together:

1. ARNI (sacubitril/valsartan) or an ACE inhibitor/ARB: blocks RAAS and, in the ARNI case, also boosts protective natriuretic peptides.
2. Beta blocker (bisoprolol, carvedilol, metoprolol succinate): blocks chronic sympathetic overdrive.
3. Mineralocorticoid receptor antagonist (spironolactone, eplerenone): blocks aldosterone-driven fibrosis and fluid retention.
4. SGLT2 inhibitor (dapagliflozin, empagliflozin): originally a diabetes drug, now proven in heart failure with and without diabetes, through mechanisms still being worked out.

Plus diuretics for symptoms (they relieve fluid overload without extending life), implantable defibrillators for those at high arrhythmic risk, cardiac resynchronisation pacing for specific conduction patterns, and in end-stage disease, left ventricular assist devices and transplantation.

For HFpEF, SGLT2 inhibitors are the first class with clear benefit, and treating the drivers (blood pressure, weight, atrial fibrillation, sleep apnoea) is central.

Atrial fibrillation

Three separate decisions. Anticoagulation to prevent stroke, guided by a risk score (CHA2DS2-VASc); direct oral anticoagulants (apixaban, rivaroxaban, edoxaban, dabigatran) have largely replaced warfarin, with less intracranial bleeding and no routine monitoring. Rate control with beta blockers or diltiazem. Rhythm control with antiarrhythmic drugs or catheter ablation, which burns or freezes the tissue around the pulmonary veins where the abnormal impulses usually originate; early rhythm control improves outcomes compared with delaying it.

Note the logic: anticoagulation is decided by stroke risk, not by how much the fibrillation bothers the patient. Someone whose AF is silent still needs the anticoagulant.

What treatment costs

In short: Every antithrombotic trades clots for bleeding, and most statin muscle symptoms turn out to be nocebo effects under blinded testing.

• Statins. Muscle aches are the common complaint; blinded n-of-1 trials (SAMSON) and randomised re-challenge studies find symptom rates on statin and placebo nearly identical, meaning most attributed symptoms are nocebo effects, though individual intolerance is real. Serious muscle injury (rhabdomyolysis) is rare, roughly 1 to 3 per 100,000 patient-years. A small increase in new type 2 diabetes diagnoses exists and is outweighed by cardiovascular benefit in those at risk.
• Antiplatelets and anticoagulants. All of them trade clot prevention for bleeding risk. That trade is favourable after a heart attack or in atrial fibrillation with meaningful stroke risk, and unfavourable in low-risk people, which is why routine aspirin for primary prevention has been withdrawn from most guidelines.
• Beta blockers. Fatigue, cold extremities, reduced peak exercise capacity.
• PCI. Bleeding at the access site, contrast-induced kidney injury, rare stent thrombosis, and the requirement for months of dual antiplatelet therapy afterwards.
• Bypass surgery. A major operation with a small mortality risk, stroke risk, and measurable cognitive effects in some patients, in exchange for durable revascularisation.

What the person can do

In short: Stopping smoking outperforms any single drug, and cardiac rehabilitation is a treatment rather than an optional extra.

The lifestyle evidence here is stronger than in almost any other chapter.

• Stop smoking. The single highest-yield action. Excess coronary risk falls sharply within the first year and approaches that of a never-smoker over 10 to 15 years. It is more effective than any single drug.
• Cardiac rehabilitation. A structured supervised exercise and education programme after a heart attack reduces cardiovascular mortality and hospitalisation. It is underused everywhere, with participation often under half of eligible patients.
• Exercise. 150 minutes a week of moderate activity, or 75 of vigorous, plus resistance work. Benefit appears well below those targets, and the largest gain is from sedentary to slightly active.
• Diet. The PREDIMED trial randomised people at high cardiovascular risk to a Mediterranean diet supplemented with extra virgin olive oil or nuts, or to a low-fat control, and found roughly 30 percent fewer major cardiovascular events in the Mediterranean groups. The practical core: more olive oil, nuts, vegetables, legumes, fish, and whole grains; less processed meat, refined carbohydrate, and sugary drinks.
• Control blood pressure and diabetes, per their chapters.
• Know the symptoms and act. Chest discomfort lasting more than a few minutes, with or without radiation to arm, jaw, or back, breathlessness, sweating, or nausea: call emergency services. Chew an aspirin unless told otherwise. Do not drive yourself.
• Learn CPR. Bystander CPR roughly doubles to triples survival from out-of-hospital cardiac arrest, and compression-only CPR is effective and easy to learn.

Living with it

After a heart attack, the psychological load is substantial: depression affects roughly a fifth of patients and independently predicts worse outcomes. Fear of exertion is common and usually the opposite of what is needed, which is one of the things supervised rehabilitation exists to correct. Return to work, driving, and sexual activity all have straightforward guidance that patients frequently do not receive unless they ask.

Heart failure imposes a different burden: daily weights, fluid and salt limits, multiple medications, and a fluctuating course punctuated by hospital admissions. Advance care planning has more value here than in most chronic diseases, because the trajectory is unpredictable and the last decline can be rapid.

What's next

• Lipoprotein(a) lowering. Pelacarsen, olpasiran, and related RNA therapies reduce Lp(a) by 80 percent or more; outcome trials are reporting through the mid-2020s and will determine whether this fixed genetic risk factor becomes modifiable.
• Gene editing. A single infusion base-editing the PCSK9 gene in the liver has entered early human trials, aiming at permanent LDL reduction from one treatment.
• Anti-inflammatory therapy. CANTOS showed that an antibody against interleukin-1beta cut cardiovascular events without changing lipids, proving inflammation is an independent target; cheaper agents such as colchicine now carry that idea into practice.
• AI on the ECG, detecting reduced pump function, valve disease, and future atrial fibrillation from an ordinary tracing.
• Polypills and population-level lipid and blood pressure control, which is where the largest global gains remain available.

Sources and notes

Global mortality figures are WHO Global Health Estimates and Global Burden of Disease data; ischaemic heart disease is consistently the leading cause of death at around 13 percent of the global total. INTERHEART: Yusuf et al., The Lancet, 2004. ISIS-2: The Lancet, 1988. Statin effect per 1 mmol/L: Cholesterol Treatment Trialists' Collaboration meta-analyses. ISCHEMIA: Maron et al., NEJM, 2020. PREDIMED: Estruch et al., NEJM, 2013 and 2018 (republished after correction of randomisation irregularities in some sites; the main conclusion was retained). CANTOS: Ridker et al., NEJM, 2017. SAMSON: Howard et al., JACC, 2021. Heart failure four-pillar therapy: PARADIGM-HF, DAPA-HF, EMPEROR-Reduced, RALES, and related trials. Rhabdomyolysis incidence and familial hypercholesterolaemia prevalence (about 1 in 250) are commonly cited literature estimates. Out-of-hospital cardiac arrest survival figures vary widely by registry and region.

Open questions. The optimal LDL target ("how low is too low") has moved steadily downward without a floor being found. Treatment of HFpEF remains far less effective than for HFrEF. Whether Lp(a) lowering translates into event reduction is not yet known.

Next: the same vascular disease, in the organ least able to tolerate it. 👉

Stroke

TL;DR. A stroke is a heart attack of the brain. Either an artery is blocked (ischaemic, about 85 percent of cases) or one bursts (haemorrhagic, the rest). Brain tissue deprived of blood starts dying within minutes, at a rate of roughly 1.9 million neurons per minute, so every part of stroke care is organised around speed. Which abilities are lost (speech, movement, vision, balance, personality) depends entirely on which artery is involved, which is why two strokes can look like completely different diseases. Stroke is the second or third leading cause of death worldwide and the leading cause of acquired adult disability, and about 90 percent of the risk is attributable to factors that can be modified.

Key takeaways

• Almost 12 million strokes occur each year and over 7 million people die of one. Roughly 1 in 4 adults will have a stroke in their remaining lifetime after age 25.
• Time is brain. Treatments that dissolve or remove the clot work in hours, not days, and the benefit shrinks steeply with every minute of delay.
• FAST: Face drooping, Arm weakness, Speech difficulty, Time to call emergency services. Add BE for Balance and Eyes to catch posterior strokes.
• A transient ischaemic attack (TIA) is a stroke whose symptoms resolve. It is not a false alarm; it is a warning, with the highest risk of a full stroke in the following days.
• High blood pressure is the dominant risk factor for both types. The INTERSTROKE study found ten factors account for about 90 percent of stroke risk worldwide.
• Organised stroke unit care saves more lives at population level than any single drug, simply by doing the ordinary things (swallowing assessment, early mobilisation, clot prevention) reliably.

What it is

In short: Either a blocked artery or a burst one, and telling them apart needs a scan because the treatments are opposites.

The brain is about 2 percent of body weight and consumes roughly 20 percent of the body's oxygen. It stores almost no fuel, so it depends on continuous flow. Interrupt that flow and function stops within seconds and cells begin dying within minutes.

Ischaemic stroke (about 85 percent): an artery supplying part of the brain is blocked by a clot. The clot may form locally on an atherosclerotic plaque (Chapter 21), travel from the heart (usually from atrial fibrillation), or arise from small vessel disease deep in the brain.

Haemorrhagic stroke: a vessel ruptures and blood damages tissue directly, while also raising pressure inside the skull. Two subtypes: intracerebral haemorrhage (bleeding into brain tissue, usually from chronic hypertension) and subarachnoid haemorrhage (bleeding into the space around the brain, usually from a ruptured aneurysm, classically presenting as a "thunderclap" headache, the worst of the person's life, reaching maximum intensity within seconds).

Transient ischaemic attack: identical symptoms that resolve completely, usually within an hour, because the blockage clears itself. The risk of a full stroke in the following 90 days is substantial and is concentrated in the first two days, which makes a TIA an emergency, not a reassurance.

Don't be confused: the two stroke types need opposite treatments. An ischaemic stroke is treated by breaking up or removing a clot, using drugs that promote bleeding. Give those to someone with a brain haemorrhage and you will kill them. The two are indistinguishable at the bedside, which is why an urgent CT scan comes before any treatment. This is the entire reason stroke care requires a hospital rather than a paramedic's judgement.

The history

In short: CT scanning made diagnosis possible, thrombolysis made treatment possible, and five trials reported in 2015 made clot retrieval standard.

Apoplexy, from the Greek for "struck down," described sudden collapse and paralysis from antiquity. In 1658 Johann Jakob Wepfer performed autopsies showing bleeding into the brain in some cases and blocked vessels in others, distinguishing the two mechanisms for the first time. Rudolf Virchow in the mid-1800s described thrombosis and embolism and the conditions favouring clot formation still known as Virchow's triad.

For a century after that, diagnosis remained guesswork and treatment was supportive. Three developments made modern stroke care possible.

CT scanning, from the 1970s, allowed bleeding and blockage to be distinguished within minutes in a living patient. Everything downstream depends on this.

Thrombolysis. The NINDS trial in 1995 showed that intravenous alteplase, a clot-dissolving drug, given within three hours of symptom onset improved outcomes, despite causing some brain haemorrhages. The window was later extended to 4.5 hours.

Thrombectomy. In 2015, five randomised trials reported within months of each other (MR CLEAN, ESCAPE, EXTEND-IA, SWIFT PRIME, REVASCAT), all showing that physically retrieving a clot from a large artery with a catheter produced dramatically better outcomes than drugs alone. The effect size was among the largest in modern medicine: the number needed to treat for one patient to be functionally independent was in the range of 3 to 7. Trials in 2018 (DAWN, DEFUSE-3) extended the window to 24 hours in carefully selected patients using perfusion imaging to identify tissue still salvageable.

What actually goes wrong

In short: A dead core surrounded by a still-salvageable penumbra, which is what every acute stroke treatment is aimed at.

The ischaemic cascade

When flow stops, neurons run out of ATP within minutes. Without ATP the pumps that maintain ion gradients fail, sodium and calcium flood in, and the cell releases its stores of the neurotransmitter glutamate. Glutamate overstimulates neighbouring neurons, driving more calcium in, which activates enzymes that digest the cell from inside. This is excitotoxicity, and it spreads outward from the dead zone.

Two regions therefore exist:

• The core: flow near zero, cells dead within minutes, unsalvageable.
• The penumbra: surrounding tissue kept marginally alive by collateral vessels, electrically silent but structurally intact, and salvageable if flow is restored.

Every acute stroke treatment is aimed at the penumbra. Its size and how long it survives vary enormously between people, depending on their collateral circulation, which is why some patients benefit at 20 hours and others have lost everything at 3.

Where the clots come from

Why vessels burst

Intracerebral haemorrhage is overwhelmingly a consequence of chronic high blood pressure, which damages small deep arteries until they rupture. In older people, cerebral amyloid angiopathy (amyloid protein deposited in vessel walls, related to the protein in Alzheimer's disease) causes bleeding nearer the brain surface. Anticoagulant drugs increase the risk and worsen the outcome of any bleed.

Subarachnoid haemorrhage usually comes from a berry aneurysm, a balloon at an arterial branch point. Aneurysms are present in perhaps 3 percent of adults, and most never rupture. Risk of rupture rises with size, location, smoking, and hypertension.

What it does to the body

In short: The deficit maps precisely onto the artery involved, which is why two strokes can look like completely different diseases.

The lost function maps onto the damaged territory, which is why a neurologist can predict the artery from the deficit.

Longer-term consequences that are underappreciated: dysphagia (unsafe swallowing, leading to aspiration pneumonia, the leading cause of death in the weeks after a stroke), spasticity and contractures, post-stroke depression (affecting roughly a third), post-stroke fatigue, emotional lability, epilepsy in about 5 to 10 percent, and vascular cognitive impairment, which can progress to dementia.

Is it deadly?

• Nearly 12 million strokes and over 7 million stroke deaths annually. Stroke is the second leading cause of death worldwide and the third leading cause of death and disability combined.
• Roughly 1 in 4 people aged 25 and over will have a stroke in their remaining lifetime.
• About a third of survivors are left permanently dependent on others for daily activities, which makes stroke the leading cause of acquired adult disability.
• Haemorrhagic strokes are deadlier than ischaemic ones: intracerebral haemorrhage carries roughly 30 to 40 percent mortality at one month. Subarachnoid haemorrhage kills a substantial fraction before they reach hospital.
• The global burden is rising in absolute terms (ageing populations) while age-standardised rates fall in high-income countries, and over 85 percent of stroke deaths now occur in low- and middle-income countries.

Is it contagious?

No. A stroke is a vascular event in one person's brain.

Two indirect infection links exist. Acute infections, including influenza and COVID-19, transiently increase stroke risk in the following weeks through inflammation and clotting. And in children, some strokes follow varicella (chickenpox) infection causing inflammation of cerebral arteries. Neither makes stroke itself transmissible.

Who gets it

In short: Ten factors account for about 90 percent of stroke risk, with high blood pressure far ahead of everything else.

INTERSTROKE, a case-control study across 32 countries, found ten factors accounting for about 90 percent of population-attributable stroke risk: hypertension (the largest by a wide margin), physical inactivity, abnormal lipids, poor diet, abdominal obesity, psychosocial factors, smoking, cardiac causes (chiefly atrial fibrillation), alcohol, and diabetes.

Age roughly doubles risk every decade after 55, though about 10 to 15 percent of strokes occur in adults under 50, where dissection, patent foramen ovale, drug use, and clotting disorders are more prominent causes.

Sex. Men have more strokes at younger ages; women have more strokes overall because they live longer, and worse outcomes on average. Pregnancy, pre-eclampsia, and some hormonal contraception modestly raise risk.

Ancestry and geography. Stroke rates are highest in East Asia, Central and Eastern Europe, and parts of sub-Saharan Africa. In the United States and the UK, people of African and Caribbean descent have roughly twice the stroke rate of white populations at younger ages, tracking earlier and more severe hypertension. Intracranial atherosclerosis is a more common stroke mechanism in East Asian populations, while carotid disease is more common in European ones.

Specific high-risk groups: atrial fibrillation (a fivefold increase, largely preventable with anticoagulation), sickle cell disease (a major cause of childhood stroke, preventable by transcranial Doppler screening and transfusion), and people who have already had a TIA or stroke.

Treatment, and how it works

In short: Speed, then a scan, then drugs or clot retrieval, and then a stroke unit, which is the one intervention that benefits every patient.

The acute phase

1. Recognise and call. Symptom onset time is the single most important piece of information, because it determines eligibility for treatment. If onset was unwitnessed (waking with symptoms), imaging can sometimes estimate it.
2. Immediate CT. To exclude haemorrhage. CT angiography identifies large vessel occlusion; CT perfusion estimates core and penumbra.
3. Thrombolysis with alteplase or, increasingly, tenecteplase (a single bolus rather than an infusion, which is faster and more practical) within 4.5 hours. It activates plasminogen, dissolving the fibrin holding the clot together.
4. Mechanical thrombectomy for large vessel occlusion: a catheter is threaded from the groin or wrist into the cerebral artery and a stent-retriever or aspiration device pulls the clot out. Standard within 6 hours, and up to 24 hours in patients whose imaging shows salvageable tissue.
5. Stroke unit admission. Care in a dedicated unit with a specialist multidisciplinary team reduces death and dependency regardless of stroke type or treatment given. It remains the intervention that benefits the largest number of patients, because everyone is eligible.
6. Aspirin within 48 hours for ischaemic stroke; anticoagulation is generally delayed.
7. Decompressive hemicraniectomy (removing part of the skull to let a swelling brain expand) in selected younger patients with large strokes, which reduces mortality substantially at the cost of surviving with significant disability.

For haemorrhagic stroke: rapid blood pressure lowering, urgent reversal of any anticoagulant, neurosurgery in selected cases, and for a ruptured aneurysm, securing it by endovascular coiling (packing it with platinum coils via catheter) or surgical clipping.

Preventing the next one

• Antiplatelet therapy (aspirin, clopidogrel, or short-term dual therapy after minor stroke or high-risk TIA) for non-cardioembolic strokes.
• Anticoagulation for atrial fibrillation, which reduces stroke risk by about two-thirds and is the highest-yield preventive intervention in the field.
• Blood pressure lowering, the largest single lever.
• High-intensity statin, which reduces recurrent stroke and cardiac events.
• Carotid endarterectomy or stenting for significant symptomatic carotid narrowing, best done within two weeks of the event.
• Closure of a patent foramen ovale in selected young patients with otherwise unexplained stroke.

Rehabilitation

Recovery happens through neuroplasticity: surviving brain regions take over functions of the damaged area, and connections reorganise. This process is use-dependent, which is why rehabilitation is a treatment rather than a comfort. What the evidence supports: starting early (though not aggressively within the first 24 hours), high repetition and intensity, task-specific practice, and specialist speech and language therapy for aphasia and swallowing. Constraint-induced movement therapy, which restrains the good arm to force use of the affected one, has good evidence in selected patients. Most recovery occurs in the first three to six months, but improvement well beyond that is documented and the old teaching that recovery stops at six months is wrong.

What treatment costs

• Thrombolysis causes symptomatic intracranial haemorrhage in roughly 2 to 7 percent of treated patients, some fatal. It is given anyway because the net benefit across eligible patients is clearly positive, and this is a good example of a treatment whose harms are real, visible, and outweighed.
• Thrombectomy: vessel perforation, dissection, clot fragments travelling to new territories, groin haematoma. Complication rates are low in high-volume centres, which is an argument for centralised stroke services even at the cost of longer transport.
• Anticoagulants: bleeding, including the rare but serious intracranial bleed. The trade-off strongly favours treatment when atrial fibrillation stroke risk is meaningful.
• Decompressive surgery: survival with major disability is a realistic outcome, and the decision requires an honest conversation about what the patient would accept.

What the person can do

In short: Learn FAST, note the exact time symptoms began, and call an ambulance rather than driving to hospital.

Before a stroke, prevention is the same list as Chapter 20 and Chapter 21, with blood pressure at the top. Two additions specific to stroke: get an irregular pulse checked, because undiagnosed atrial fibrillation is common and silent, and take anticoagulation seriously if it is prescribed, because it is one of the most effective preventive drugs in medicine and adherence to it is poor.

During a stroke, the highest-value action belongs to whoever is nearby. Learn FAST. Note the time symptoms began. Call emergency services rather than a family doctor or a taxi, because ambulances pre-alert stroke teams and can route to a thrombectomy-capable centre. Do not give food, drink, or aspirin before assessment, since swallowing may be unsafe and the stroke may be a bleed.

After a stroke: attend rehabilitation and do the home exercises, which are the active ingredient; treat depression, which is common and undertreated and which impairs rehabilitation; take secondary prevention medication indefinitely; and address driving, work, and mood explicitly rather than waiting to be asked.

Living with it

Stroke is the most disability-producing common disease in this book, and much of the burden is invisible from outside. Aphasia is particularly isolating: intelligence and personality are intact while the ability to express them is not, and people frequently speak to a person with aphasia as though they were cognitively impaired. Neglect after a right-hemisphere stroke can be misread as inattention or lack of effort. Post-stroke fatigue is disproportionate to physical deficit and poorly understood.

Informal caregivers, usually family, provide most long-term care and experience high rates of depression and financial strain themselves. Caregiver support is part of stroke treatment rather than an optional extra.

What's next

• Mobile stroke units, ambulances carrying a CT scanner, delivering thrombolysis at the roadside and cutting time to treatment substantially.
• Tenecteplase replacing alteplase as the standard thrombolytic on practical grounds.
• AI imaging triage, automatically detecting large vessel occlusion on CT and alerting the thrombectomy team while the patient is still in the scanner.
• Extended and imaging-selected windows, moving from clock-based to tissue-based eligibility.
• Neuroprotection, a field with a long record of failure in humans despite success in animals, now being retried in combination with reperfusion rather than instead of it.
• Wider access. The largest available gain worldwide is not a new drug; it is organised stroke units and basic hypertension control in the countries where 85 percent of stroke deaths now occur.

Sources and notes

Incidence, mortality, and lifetime risk figures are from the World Stroke Organization Global Stroke Fact Sheet 2025 and Global Burden of Disease analyses: nearly 12 million new strokes and over 7 million deaths annually, and approximately 1 in 4 lifetime risk after age 25. The 1.9 million neurons per minute estimate is from Saver, Stroke, 2006. INTERSTROKE: O'Donnell et al., The Lancet, 2010 and 2016. NINDS alteplase trial: NEJM, 1995. The 2015 thrombectomy trials: MR CLEAN, ESCAPE, EXTEND-IA, SWIFT PRIME, REVASCAT, with the HERMES pooled analysis. Extended window: DAWN and DEFUSE-3, NEJM, 2018. Stroke unit care: Stroke Unit Trialists' Collaboration, Cochrane reviews. Subtype proportions follow TOAST classification studies and vary by population.

Open questions. Whether thrombectomy benefits patients with large established cores is being actively redefined by recent trials. Optimal blood pressure targets in the first days after ischaemic stroke remain uncertain. No neuroprotective agent has yet succeeded in humans.

Next: the organ that quietly cleans everything up, and what happens when it stops. 👉

Kidney Disease

TL;DR. Your kidneys filter your entire blood volume roughly 30 times a day, producing about 180 litres of filtrate and reclaiming almost all of it, so that what leaves is about 1.5 litres of urine containing exactly the waste, salt, acid, and water you needed to lose. They also make the hormone that tells your marrow to produce red blood cells, activate vitamin D, and set your blood pressure. Chronic kidney disease is the slow loss of the million filtering units in each kidney, usually caused by diabetes or high blood pressure, and it is silent until roughly 80 to 90 percent of function is gone. Most people with it never reach dialysis, because they die of cardiovascular disease first, which the kidney disease itself accelerated.

Key takeaways

• More than 850 million people worldwide have some form of kidney disease, roughly one in ten adults, and most do not know.
• Diabetes and hypertension cause the majority of cases. Kidney disease is largely a complication of the two preceding chapters.
• It is graded by eGFR (how fast the kidneys filter) and albuminuria (how much protein leaks into urine). Both are needed; either alone misses people.
• Damage is a self-accelerating loop: lose nephrons, the survivors filter harder, overwork scars them too.
• Kidney disease is primarily a cardiovascular risk multiplier. A person with moderate CKD is far more likely to die of a heart attack than to reach dialysis.
• SGLT2 inhibitors changed the field, slowing progression substantially in people with and without diabetes, after two decades in which only blood pressure drugs helped.

What it is

In short: Two axes define it, filtration rate and protein leak, and using either alone misses people who are at high risk.

Each kidney contains about a million nephrons, and each nephron is a filter plus a long processing tube. Blood enters a tuft of capillaries called the glomerulus, where pressure forces water and small molecules through a three-layer barrier that holds back cells and proteins. The filtrate then travels down a tubule where the body reclaims what it wants: nearly all the water, all the glucose, most of the sodium, and adjusts acid and potassium precisely.

Beyond filtration, the kidney does four other jobs that explain most of the symptoms of kidney failure:

Chronic kidney disease (CKD) is abnormal kidney structure or function present for more than three months. It is staged on two axes.

Crossed with albuminuria: A1 (normal, under 30 mg albumin per gram creatinine), A2 (30 to 300), A3 (over 300). Someone at G1 A3 (normal filtration, heavy protein leak) can be at higher risk than someone at G3a A1, which is why both axes are needed.

Don't be confused: acute kidney injury and chronic kidney disease are different events. AKI is a sudden drop in function over hours to days, usually from dehydration, sepsis, severe blood loss, obstruction, or a nephrotoxic drug. It is often reversible if the cause is corrected fast. CKD is slow, structural, and mostly irreversible. They interact: an episode of AKI raises the long-term risk of CKD, and CKD makes AKI more likely and more damaging.

The history

In short: From Bright's 1827 description to a dialyser built from sausage casing and a washing machine, and the rationing committee that founded modern bioethics.

Richard Bright at Guy's Hospital in London established the field in 1827 by connecting three findings that had been seen separately: swelling of the body, protein in the urine (detected by heating it and watching it coagulate), and diseased kidneys at autopsy. Kidney disease was called Bright's disease for the next century.

Understanding did not translate into treatment until the twentieth century. In 1943, in occupied Holland, Willem Kolff built the first working dialysis machine from sausage casing, orange juice cans, and a washing machine drum. His first fifteen patients died; the sixteenth, in 1945, survived. Dialysis could only be used a few times per patient because each session required cutting into a new artery and vein, until Belding Scribner's Teflon shunt in 1960 made repeated access possible and turned dialysis into a long-term therapy.

That created the first modern rationing crisis. Seattle had far more eligible patients than machines, and an anonymous lay committee decided who received treatment, weighing occupation, family, and perceived social worth. A 1962 Life magazine article about the "God Committee" is widely regarded as a founding moment of modern bioethics.

The first successful kidney transplant was performed in 1954 by Joseph Murray between identical twins, which sidestepped rejection entirely. Effective immunosuppression, above all ciclosporin in the 1980s, made transplantation between unrelated people routine. Recombinant erythropoietin arrived in 1989 and largely ended the severe anaemia that had defined life on dialysis.

What actually goes wrong

In short: Diabetes and hypertension cause most of it, and the surviving filtering units then overwork themselves into scarring.

Causes, in rough global order:

1. Diabetes (about 30 to 40 percent of cases in most high-income countries). High glucose damages the glomerular capillaries, thickens the filtration barrier, and lets protein through.
2. Hypertension (about 25 percent). Pressure damages the small vessels feeding the nephrons.
3. Glomerulonephritis: a family of immune-mediated diseases attacking the glomerulus directly, including IgA nephropathy (the commonest worldwide), lupus nephritis, and membranous nephropathy.
4. Polycystic kidney disease: an inherited autosomal dominant condition in which fluid-filled cysts progressively replace working tissue. It affects roughly 1 in 1,000 people and is the commonest inherited cause of kidney failure.
5. Obstruction: prostate enlargement, stones, tumours.
6. Chronic kidney disease of unknown cause (CKDu): epidemics among agricultural workers in Central America, Sri Lanka, and India, discussed below.

The self-accelerating loop. Whatever the initial insult, once a meaningful number of nephrons are lost, the survivors increase their individual filtration rate to compensate (hyperfiltration). That maintains total function for a while, and it damages the overworked glomeruli, which scar (glomerulosclerosis), which increases the load on the remaining ones. Beyond a threshold, progression continues even if the original cause is removed.

Protein leak drives damage too. Albumin passing into the tubules is not an innocent marker; it provokes inflammation and scarring in the tubule cells. This is why reducing proteinuria, not just lowering blood pressure, is a treatment target in its own right, and why drugs that lower it slow progression.

What it does to the body

In short: Silent until late, then anaemia, bone disease, acidosis, fluid overload, and a large rise in cardiovascular risk.

Early CKD produces nothing at all. Symptoms appear late, and they are the accumulated failure of all five kidney jobs.

The most important single fact about CKD is one people rarely hear: it is a cardiovascular disease multiplier. Someone with stage 3 CKD is several times more likely to have a cardiovascular event than to progress to dialysis, and cardiovascular disease is the leading cause of death at every stage of CKD.

Is it deadly?

Yes, and increasingly so in global terms.

• More than 850 million people worldwide have some form of kidney disease, roughly double the number with diabetes.
• Around 4 million people are on dialysis or living with a transplant, while millions more who need kidney replacement therapy die without access to it, particularly in low-income countries where dialysis is unaffordable.
• CKD is projected to become the fifth leading cause of death globally by 2050, one of the largest projected rises of any condition.
• Mortality on dialysis is high: five-year survival is comparable to several common cancers, driven mostly by cardiovascular death. Transplantation roughly doubles life expectancy compared with remaining on dialysis.

Is it contagious?

No. Kidney disease itself does not spread.

Two related points. Some causes of kidney disease are infectious: hepatitis B and C and HIV can cause kidney damage, streptococcal infection can trigger post-infectious glomerulonephritis, malaria and schistosomiasis damage kidneys, and untreated urinary infections that reach the kidneys can scar them. Historically, hepatitis B and C spread within dialysis units through inadequate infection control, which is why dialysis units now have rigorous isolation and screening protocols.

Who gets it

In short: Mostly people with diabetes and high blood pressure, plus an ancestry-linked genetic risk and an occupational epidemic among heat-exposed farm workers.

Diabetes and hypertension account for most of it. Add age (filtration declines naturally with age, though how much is normal ageing is debated), obesity, smoking, cardiovascular disease, family history, recurrent kidney stones, prolonged NSAID use, and episodes of acute kidney injury.

Genetics: the APOL1 story. Two variants of the APOL1 gene, common in people of West African ancestry, substantially raise the risk of several kidney diseases, including hypertension-attributed kidney failure, HIV-associated nephropathy, and focal segmental glomerulosclerosis. Carrying two copies raises the lifetime risk of kidney failure several-fold. These variants persisted because they confer protection against the parasite causing African sleeping sickness, the same evolutionary bargain visible in sickle cell and malaria (Chapter 62). This explains a substantial part of the roughly threefold to fourfold higher rate of kidney failure among African Americans, though socioeconomic factors, access to care, and diabetes and hypertension prevalence contribute as well. It is one of the clearest examples in medicine of an ancestry-linked genetic risk that is specific, mechanistic, and now druggable, and it is worth contrasting with the many claims about "racial" differences that turn out to be about environment.

CKD of unknown cause (CKDu). Since the 1990s, epidemics of kidney failure have appeared among young agricultural workers, chiefly sugarcane cutters, in Central America (especially Nicaragua and El Salvador), Sri Lanka, and parts of India. These are men in their twenties to forties with no diabetes and no hypertension. The leading hypothesis is recurrent heat stress and dehydration during hard physical labour, possibly combined with agrochemical exposure. In some communities it has become the leading cause of death in working-age men. As global temperatures rise, this is one of the first clearly documented occupational disease epidemics attributable in part to heat.

Access is the other determinant. In high-income countries, kidney failure means dialysis or transplant. In much of the world it means death, because a year of dialysis costs more than most families earn. The gap between the number of people who need kidney replacement therapy and the number who receive it is one of the largest inequities in medicine.

Treatment, and how it works

In short: Nothing regrows nephrons, so treatment slows the loss, and SGLT2 inhibitors substantially changed how much can be slowed.

There is no way to regrow nephrons. Treatment slows loss, manages consequences, and eventually replaces function.

Slowing progression

The arrival of SGLT2 inhibitors is the most consequential change in nephrology in a generation. A drug developed to lower blood sugar turned out to slow kidney decline in people who do not have diabetes, and the effect is large.

Managing the consequences

Anaemia: erythropoiesis-stimulating agents plus iron, with a target haemoglobin deliberately below normal, because trials showed that correcting anaemia fully increased strokes and death. Bone and mineral disorder: phosphate binders taken with meals, vitamin D analogues. Acidosis: oral sodium bicarbonate. High potassium: dietary restriction and potassium-binding drugs, which increasingly allow patients to stay on the kidney-protective drugs that raise potassium. Fluid overload: loop diuretics and salt restriction.

Replacing function

Haemodialysis. Blood is pumped through a filter of thousands of hollow fibres, with dialysis fluid flowing the other way. Waste diffuses out down its concentration gradient and excess fluid is pulled off by pressure. Typically three sessions a week, four hours each, usually in a centre. It requires vascular access, ideally an arteriovenous fistula, a surgically joined artery and vein in the arm that enlarges over weeks into a vessel able to take large needles.

Peritoneal dialysis. Fluid is run into the abdominal cavity through a permanent catheter, and the peritoneal membrane acts as the filter. Done at home, often overnight by machine. It preserves independence and residual kidney function and carries a risk of peritonitis.

Transplantation is the best treatment for those eligible: better survival, better quality of life, and lower long-term cost than dialysis. A kidney from a living donor lasts longer on average (median around 15 to 20 years) than one from a deceased donor (around 10 to 15). Recipients take lifelong immunosuppression, which brings its own infection and cancer risks. The limiting factor everywhere is organ supply; waiting lists run to years, and many patients die waiting.

Conservative management is a legitimate choice, particularly for frail elderly patients in whom dialysis may extend life little while dominating what remains of it. Choosing not to dialyse, with good symptom control, is a decision an informed patient is entitled to make.

What treatment costs

• ACE inhibitors and ARBs cause an expected small rise in creatinine when started (up to about 30 percent is acceptable and reflects the intended change in glomerular pressure), plus raised potassium.
• SGLT2 inhibitors cause a similar initial dip in eGFR that reverses and predicts long-term benefit, plus genital yeast infections and rare euglycaemic ketoacidosis.
• Haemodialysis: fatigue and low blood pressure after sessions, cramps, access infections and clotting, and a very large time cost, roughly 12 to 15 hours a week before travel.
• Peritoneal dialysis: peritonitis, hernias, and eventual membrane failure after several years.
• Transplant immunosuppression: infections, higher rates of skin cancer and lymphoma, diabetes, kidney toxicity from the drugs themselves, and rejection if doses are missed.