Pseudoepidemics of heroin addiction.

Smooth unimodal skewed distributions, sometimes referred to as "epidemic-like" curves, are often used as evidence of time to time changes in incidence. Data on year of onset from groups of treated heroin addicts are generally alleged to reflect incidence, and this distribution is used to evaluate effectiveness and predict future needs. This paper describes the development of simple demographic models which demonstrate the production of epidemic-like curves without changes in incidence, ascertainment or duration of disorder. These graphic models are relevant to other non-infectious disorders where there is continuing exposure to the etiologic agents and where intervals between onset and ascertainment are not uniformly distributed.