which seroprevalence rates among women were found to be statistically higher than among men (i.e., 18 percent and 10 percent, respectively). Moreover, in all regions, seroprevalence rates among African Americans were higher than among whites; African Americans had a two- to sixfold increased seroprevalence compared with whites (rates ranged from 8 percent among African Americans compared with 3 percent among whites in the Midwest to 38 percent compared with 21 percent, respectively, in the Northeast). Hispanics in the Northeast were found to have prevalence rates similar to those of African Americans in the same region, but they were found to have higher rates than African Americans in the Midwest. These same surveillance data indicate that HIV seroprevalence has stabilized in most U.S. metropolitan areas. Although a moderate decline in HIV seroprevalence among the young (<30) white injection drug users was observed in high-seroprevalence areas (>10 percent), trends in annual seroprevalence were found to be stable among age and racial/ethnic subgroups. Des Jarlais et al. (1994) have also reported such declines in a cross-sectional survey among young injection drug users entering a detoxification unit in a high-seroprevalence area.
Similar patterns of seroprevalence by geographic areas have been reported by researchers at the National Institute on Drug Abuse (Battjes et al., 1991) from data collected as part of a series on nonblind point-prevalence surveys among injection drug users admitted to methadone treatment in seven areas (New York City; Trenton and Asbury Park, New Jersey; Baltimore, Maryland; Chicago, Illinois; San Antonio, Texas; and Los Angeles, California) over a 2-year period (from late 1987 through early 1989). These researchers reported significant variations in seroprevalence across geographical locations. The highest rates were observed in the Northeast: New York City and Asbury Park had rates ranging from 28.6 to 58.6 percent. Los Angeles (West) had low prevalence rates ranging from 0.9 to 3.4 percent over the course of this 2-year study. With the exception of Chicago, the multiple data points across time revealed stable seroprevalence rates within geographic location. In a more recent look (i.e., 1987 through 1991) at the seroprevalence rates in five of those original cities (i.e., New York City, Asbury Park, Trenton, Baltimore, and Chicago), Battjes et al. (1994) reported similar seroprevalence rates by location.
This reported stabilization of seroprevalence rates within geographical areas is comparable to the results of mathematical modeling studies that indicate that HIV incidence among injection drug users has shown a slight to moderate decline since the mid-1980s (Brookmeyer, 1991). Moreover, Drucker and Vermund (1989) have provided a mathematical model that allows the estimation of prevalence rates for population subgroups and overcomes some of the limitations that are associated with large-scale national cross-sectional surveys. National seroprevalence surveys are expensive,