Estimating AIDS Incidence and Jack-knife Variance from a Continuous Delay Distribution and Incomplete Data

Usual methods for estimating AIDS incidences are based on the inflation of a discrete reporting delay distribution, which often results in very imprecise estimates of the incidence in the most recent past. In this paper, we propose an alternative approach to estimate the AIDS incidence by inflating a continuous reporting delay distribution for each reported case. Covariate effects on reporting delays are evaluated by a proportional hazards model for the reverse time hazard function. A jack-knife variance for the estimated AIDS incidence is given. Study results showed that precision of estimates is improved by using the continuous time model as compared with those estimates given by its discrete counterpart. This feature is useful in assessing current trends in AIDS incidence.