Interval-censored survival data with informative examination times: Parametric models and approximate inference

We develop parametric methods for analysing interval-censored data when exa mination and survival times are not independent. The hazard function is mod elled by introducing individual frailties related to the frequency of exami nations. Model parameters may be obtained by direct maximization of the mar ginal log-likelihood. We develop a simpler approximate method in which the frailties are estimated by empirical Bayes. The two approaches are equivale nt asymptotically as the number of examinations on each individual increase s. Simulations suggest that the approximate method is adequate for estimati ng regression parameters even when the number of examinations on each indiv idual is small. The methods are used to estimate age and period effects on HIV incidence in a cohort of repeat attenders at genito-urinary clinics in London, U.K. Copyright (C) 1999 John Wiley & Sons, Ltd.