The data in this report provide estimates of HIV prevalence for specific subsets of the population of the United States from 1993–1997. Because national population-based estimates of HIV prevalence have not been available since the Survey of Childbearing Women was discontinued in 1995, we can only estimate HIV infection in the general population. MSM attending public STD clinics are not representative of all MSM, IDUs in treatment are not representative of all IDUs, and Job Corps entrants are not representative of all youth. In addition, military applicants and blood donors are not representative of the general population because of self-exclusions for both groups and the further exclusions, based on risk behaviors, of potential blood donors. However, trend data are less subject to bias, and although there may be limitations to extrapolations from one group to another, the general declines in seroprevalence in nearly all groups suggest that prevention efforts have met with some success.

Changes in the composition of the survey populations complicate interpretations of HIV prevalence trends. Although prevalence rates were adjusted to the 1993 populations to control for demographic characteristics in the selected populations that may have changed over time, other changes in the composition of the population under study may have influenced the observed trends. For example, changes in referral and access patterns may affect estimates of prevalence, regardless of the underlying prevalence among all persons served by a particular venue. Also, because of the nonrandom selection of STD clinics and DTCs, results cannot be generalized to other populations not served by these venues. However, the consistency of the collective trends suggests patterns and temporal changes in HIV infection.

The overall patterns of HIV seroprevalence data from 1993–1997 suggest that HIV prevalence continues to decrease or to remain stable in the populations included in this report. Overall prevalence did not increase in any sentinel population. Despite these encouraging findings, prevalence remained high among MSM and IDUs and was substantially higher among blacks than among whites in all populations included in this report. In addition, prevalence increased among certain subgroups, particularly female Job Corps entrants. Considering the young ages of this population, prevalence was high, especially among black women.

Patterns in the HIV epidemic among adults in the United States are primarily influenced by three interrelated subepidemics among groups at risk for HIV infection: MSM, IDUs, and heterosexual persons. STD clinic surveys provide information about HIV infection among MSM. Overall HIV prevalence was higher among MSM at STD clinics (26%) than among any of the other survey populations. Prevalence for these MSM was high in all metropolitan areas, ranging from 8% in Seattle to 39% in Atlanta. However, overall prevalence among MSM decreased steadily from 1993–1997. Although prevalence was lower among younger MSM, the 1997 standardized prevalence rate of 11% among those in the youngest age group (< 25 years) indicates that new infections continue to occur. This is consistent with the results of other recent studies showing high numbers of new infections among young MSM.

Because the STD clinic-based HIV surveys were conducted among persons at high risk for other STDs, it is difficult to generalize results from these surveys to other MSM and heterosexual persons for two reasons. First, the observed HIV prevalence among MSM and heterosexual patients at STD clinics was probably higher than the overall prevalence among similar groups in the general population. Also, the observed decreases in HIV prevalence could have underestimated true decreases among all MSM and heterosexual patients because persons who had changed their behaviors to lower their risk for STDs, including HIV, would have been less likely to attend STD clinics. Overall seroprevalence among IDUs entering treatment programs (19%) was the second highest of the populations in the surveys. In contrast to prevalence among MSM, prevalence among these IDUs differed greatly by region, ranging from 2% or less in Denver and Los Angeles to 37% in New York City. During the survey period, prevalence decreased or remained relatively stable among most subgroups of IDUs. A notable exception to the pattern of decreasing prevalence was white IDUs in the South, among whom prevalence increased more than threefold, from 3.2% in 1993 to 11.1% in 1997.

Because IDUs in treatment programs are typically older than those not in treatment and therefore have had more time to acquire HIV infection, the prevalence rates obtained from these surveys could be overestimates of HIV infection in the entire population of IDUs. On the other hand, if IDUs who are not in treatment programs engage in high-risk behaviors more frequently than those who enter treatment, as some studies suggest, the HIV prevalence rates in this report may underestimate HIV infection among all IDUs. However, recent interview-based incidence studies at many of the participating DTCs have shown continued high rates of sharing of needles and other drug injection equipment as well as high rates of risky sexual behaviors.

The HIV epidemic among heterosexual men and women is influenced by persons who inject drugs or who have heterosexual contact with IDUs, MSM, or other persons at high risk of HIV infection. HIV prevalence among heterosexual men and women at the STD clinics was substantially lower than that among MSM and IDUs at the STD clinics and IDUs entering drug treatment programs. Among high-risk heterosexual patients at STD clinics who reported no male same-sex contact and no injection drug use, overall prevalence for the 5-year period was 2.3%.

Overall prevalence rates among youth attending adolescent medicine clinics and those entering the Job Corps remained low (0.4% or less). However, considering the young ages of these populations, prevalence was high, especially among young disadvantaged women entering the Job Corps. These women, particularly those who were black, were infected with HIV at younger ages and at higher rates than men. The high prevalence rates among these women could be due to several factors, including lack of knowledge or lack of skills in negotiating condom use and early sexual intercourse with older men who are more likely than young men to be infected with HIV.

HIV prevalence was higher among men than among women in the lowest-risk populations military applicants and blood donors. However, prevalence in both populations remained very low, probably because of the self-exclusion of persons with known risk for HIV infection. HIV prevalence among female first-time blood donors in 1997 was only 0.010%. Because of stricter exclusion criteria for potential blood donors and better availability and awareness of HIV counseling and testing, the declining HIV prevalence trends among first-time blood donors may not be representative of a downward trend in the general population.

It is difficult to make inferences from the serosurveillance data about prevalence among heterosexuals in the general population. First, because HIV prevalence among MSM and IDUs is high, the misclassification of only a few of these persons could greatly increase the measured prevalence among persons classified as heterosexual patients who did not inject drugs. Second, a large percentage of the population is at risk for HIV infection through heterosexual contact although the level of risk differs greatly, depending on factors such as practiced sexual behaviors, social mixing patterns, choice of partners, and HIV prevalence in the community. Thus, it is difficult to generalize data from heterosexual men and women in the sentinel populations included in this report.

Strong geographic heterogeneity in HIV prevalence rates was observed among heterosexual patients at STD clinics, IDUs entering DTCs, Job Corps entrants, and military applicants. Although the reasons for this geographic diversity are not clear, HIV prevalence has been higher among all of these populations in the Northeast and the South since CDC began the serosurveillance system in 1988. These regional variations reflect the large differences in the effect of the HIV epidemic in different areas of the United States.

One of the most striking observations from these surveys is the marked race/ethnicity differences in HIV prevalence. In nearly all of the populations, prevalence was substantially higher among blacks than among whites. Although data from Hispanics were less consistent, prevalence among Hispanics was lower than among blacks and slightly higher than among whites in most populations. The exception was among IDUs in the Midwest and South, where prevalence was almost always higher for Hispanics than for blacks. The marked disparities by race and ethnicity suggest that social mixing patterns within racial and ethnic groups are important determinants of HIV transmission risk.

Even though the groups in the surveys are not representative of the general population of the United States, the consistency of findings of continuing overall decreasing or stable prevalence across surveys and geographic areas is encouraging, especially where the younger groups show decreases in prevalence. Stable HIV prevalence over time indicates that the number of HIV-positive persons entering a given population is approximately equal to the number of HIV-positive persons leaving the population. There are several possible explanations for stable or decreasing prevalence. The most optimistic is that HIV infection in the population and HIV incidence actually have decreased because of education and interventions for at-risk persons. A decrease in seroprevalence, however, does not exclude high rates of new infections. It is possible that as more persons were voluntarily tested for HIV and became aware of their serostatus, those with HIV infection have been less likely to have entered the populations under study. Also, changing referral or access patterns could have affected the number of HIV-positive persons who visited a given clinic.

HIV incidence estimates for specific risk groups are desirable so that prevention programs can be better targeted and evaluated. In the past, incidence was difficult and expensive to measure because multiple specimens from one person over time were needed to show a change from a seronegative test result to a seropositive test result. This "cohort" approach introduced significant biases related to a person’s decisions to be tested for HIV. For example, to be included in the cohort, persons must have volunteered to undergo HIV testing more than once during the survey period. Differential losses related to HIV status would result in bias associated with loss to follow-up. In an effort to better measure incidence, CDC has developed an accurate, inexpensive technology that can distinguish recent infections from longstanding infections by using a single specimen from each person; thus, estimates of HIV incidence that are not biased by testing decisions can be determined.

The HIV epidemic has slowed from a period of rapid growth (approximately 150,000 new infections per year) during the mid-1980s to late 1980s to a more stable epidemic (estimated at approximately 40,000 new infections per year) since the mid-1990s. A renewed national resolve is needed to further reduce the number of new infections and to ensure that people with HIV/AIDS have access to testing, care, and treatment so that they can live longer, healthier lives. A key component to this effort is surveillance. Although AIDS surveillance has given us valuable information in the past, the natural history of HIV infection has changed because of more effective treatments that can lengthen the time between HIV infection and the development of AIDS. Thus, surveillance of new HIV infection (incidence) is crucial for determining the current state of the epidemic and allowing agencies to mobilize and target populations where HIV transmission is still occurring. To reduce the number of new HIV cases per year, complacency must be abandoned, and new and innovative approaches must be explored.

Go to Appendix I