Prevalence

Prevalence counts who has the condition right now, and it is the simplest fraction in epidemiology: cases over population. It is a stock rather than a flow — a photograph of a population at one moment, not a record of what happened to it — and the distinction from incidence is the first thing to fix, because the two behave completely differently.

Prevalence rises when new cases appear and falls when cases either recover or die, so it is roughly incidence multiplied by average duration. That product has a consequence people find backwards: a treatment that keeps patients alive without curing them raises prevalence. Antiretroviral therapy raised HIV prevalence in every country that adopted it, and that rise was the intervention working. Falling prevalence is not automatically good news and rising prevalence is not automatically bad; it depends entirely on which of the two terms moved.

Three variants get quoted under the same word and are not interchangeable. Point prevalence is the share affected on a given day. Period prevalence is the share affected at any time during a stated window. Lifetime prevalence is the share who have ever had it, and for a condition like depression it can be several times the point figure. A prevalence quoted without its variant is not a complete number.

The reason this page sits beside the diagnostic ones is that prevalence is the pre-test probability. It is what turns a test's sensitivity and specificity into what a result actually means, it is the term the predictive-value pages depend on, and it is the term most often left out of the conversation entirely. Where it is measured by a screening programme rather than known independently, there is a circularity to watch: the measured prevalence is itself a product of the test's accuracy, and an imperfectly specific test in a low-prevalence population will report a prevalence several times the truth.