Reporting Delays and the Incidence of AIDS

Abstract

It can take several months, and often years, for case reports of acquired immunodeficiency syndrome (AIDS) to be received by the Centers for Disease Control (CDC). As a result, the cumulative number of AIDS cases reported by the CDC at a given date may fall considerably below the actual number thus far diagnosed. Methods are described for estimating both the probability distribution of reporting delays and the actual incidence of AIDS. An estimated 62% of AIDS cases are reported more than 2 months after diagnosis, and 17% are reported with a delay of 3 years or more. An estimated 130,000 AIDS cases were actually diagnosed through March 1989, compared to about 91,000 cases reported by that time. There has been an increase in reporting delays during 1982–1989, as well as significant geographic variation in reporting delays. The actual incidence of AIDS is found to be rising most rapidly in nonurban regions and metropolitan areas with less than one million population. A model of AIDS incidence based upon the incubation of human immunodeficiency virus (HIV) is applied to data on reported AIDS cases among non-drug-using homosexual men. In this group, the estimated incidence of HIV infection peaked at about 8,900 cases per month in early 1983, and the resulting incidence of AIDS will peak at an estimated 3,200 cases per month in early 1993. The latter estimates were very sensitive to the specification of the incubation density for HIV infection.

Key Words:

• Truncated data
• Generalized linear models
• Human immunodeficiency virus
• Incubation period
• Epidemic models