are preferable, but data of this sort are often not available because, in many individuals, HIV is undetected or unreported. The HIV/AIDS epidemic, therefore, is often viewed through the imperfect lens of AIDS cases. AIDS is a clinically observable syndrome that is clearly defined and required to be reported in every state. Many individuals with new HIV infections have not progressed in their disease to the point that AIDS can be diagnosed, so AIDS statistics reflect HIV infections that occurred up to a decade or more in the past. Thus, five different type of epidemiological data can shed light on trends and patterns in perinatal transmission of HIV: (1) reported cases of AIDS in women, (2) reported cases of AIDS in children, (3) reported HIV infections in women (where available), (4) estimates of the prevalence of HIV in childbearing women based on the Survey of Childbearing Women (SCBW) (see below), and (5) reported HIV infections in newborn children. Data on HIV infections in children are typically available only in states with mandatory HIV reporting. These states monitor perinatally exposed children to see if they convert from HIV-positive due to maternal antibodies to HIV infection and AIDS status. Each type of data has its strengths and weaknesses, but taken together it is possible to construct a relatively complete picture of the perinatal HIV epidemic in the United States.
AIDS case reports, the source of the first two data series mentioned above, are gathered by state, territorial, and local health departments and reported to the Centers for Disease Control and Prevention (CDC) to form a national AIDS surveillance system. Standard CDC records for each case include information on age at diagnosis, sex, race and ethnicity, state of residence (and metropolitan area, if relevant), mode of exposure to HIV (including maternal risk for pediatric cases), month of AIDS diagnosis, date reported to CDC, and other information. The national data are made available in terms of biannual tabular reports (CDC, 1997c), an AIDS Public Information Data Set (APIDS) (CDC, 1997a), and other reports from CDC (see, for example, CDC, 1996a, 1997e; Wortley and Fleming, 1997). Many states also routinely produce HIV/AIDS surveillance reports.
Even though nearly all current pediatric AIDS cases are the result of perinatal transmission, information on reported AIDS in children provides only limited insight into the problem. First, many children infected with HIV perinatally do not develop AIDS until they are substantially older. There seems to be a bimodal distribution; approximately 48% of HIV-infected children develop AIDS by three years of age, and thereafter less than 3% per year develop AIDS (Pliner et al., 1998). Diagnosed AIDS cases thus reflect perinatal transmissions in births years earlier. Second, once diagnosed, AIDS data are subject to reporting delays. Overall, only 55% of cases are reported to CDC within three months of diagnosis, but 20% are reported more than one year after diagnosis (CDC, 1997a). Reporting delays are longer for pediatric cases (an average of six months) than for adult cases (which average three months). Published data are sometimes adjusted for