Estimating the morbidity risk of illness from survey data

A method is proposed for using survey data to estimate lower and upper bounds for the lifetime risk of an illness (morbidity risk). The mathematical model used, which is based on a three-state Markov process, assumes that the illness is irreversible, but allows differential mortality. The data required include information on age at onset collected from prevalent cases, and an estimate of the magnitude of differential mortality, which ordinarily must come from published research. The method is illustrated using data from a community survey of psychiatric illness conducted in Edmonton, Alberta, Canada.