[Index event bias: why causal factors appear not to apply to disease recurrence]

In etiological epidemiological studies, bias is of constant concern. The peculiar phenomenon that causes of common disease do not predict the risk of recurrent disease may be explained by a specific form of bias that is not commonly recognised: index event bias. The principle of index event bias is that, in the presence of a specific cause of disease, the causal complement consisting of other disease causes is smaller than in the absence of this specific cause. Since the causal complement consists of both known and unknown disease causes, multivariable adjustment can only partially overcome this type of bias. Hence, in all forms of etiological studies addressing recurrent disease risk, index event bias is a source of potential underestimation of causal associations.