Causal inference: the case of hygiene and health.

A fundamental goal of applied epidemiology is to determine whether a relationship between 2 factors is causal. For example, the primary purpose of an outbreak investigation is to identify what factor(s) “caused” the problem, and the purpose of the Study of the Efficacy of Nosocomial Infection Control (SENIC Project) was to measure the effect of infection control and prevention programs on rates of nosocomial infections (ie, whether strong programs were associated with or “caused” a reduction in infections).1 There have been a number of historic efforts to formulate methods for valid causal inference.2-5 To the layperson, association between 2 variables is often assumed to be causal, but from the epidemiologic point of view, this is not the case. For example, in studies concerning health and hygiene, associations have been found between hand hygiene interventions and a decline in diarrhea and between laundering practices and prevalence of infectious disease in the home.6-9 Such associations, however, must properly reflect an underlying causal mechanism and must have been investigated with use of rigorous methodologic procedures.10-12 Our purpose is to discuss 1 of the historic methods devised for inferring causation and the epidemiologic concepts that relate to causal inference, with the example of health and hygiene.