A simple correction of AIDS surveillance data for reporting delays

Abstract

A simple noniterative method is presented for adjustment of AIDS surveillance data for reporting delays. The method estimates the distribution of reporting delays from a cross-classification of incident cases according to date of diagnosis vs. length of reporting delay. Using the delay distribution, the number of cases reported to have occurred within each period of calendar time is inflated to reflect the actual number of cases diagnosed but not yet reported. The method assumes that the delay distribution is independent of the calendar period of diagnosis, but can readily be modified to account for early departures from the current pattern. The method is illustrated using surveillance data from the northeastern United States.