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. 👉