Trends in Heterosexually Acquired AIDS in the United States, 1988 Through 1995

Cited Here... We used national AIDS surveillance data to characterize trends in the numbers and proportions of heterosexually acquired AIDS cases diagnosed from January 1988 through December 1995 among adults and adolescents. We adjusted for expansion of the 1993 AIDS surveillance case definition and for delays in reporting, and we redistributed cases initially reported without risk. The chi-square test for linear trend was used to analyze trends at the p< 0.01 level by half-year of diagnosis and by sex, age, race or ethnicity, geographic region of residence at diagnosis, and partner's HIV exposure risk. From 1988 through 1995, heterosexual contact accounted for 10% of all AIDS cases. Heterosexual contact increased the most rapidly of all HIV exposure modes, with increases found among men and women in all age groups; among blacks, whites, and Hispanics; and among persons living in all geographic regions of the country. Blacks and Hispanics accounted for 75% of all persons reported with AIDS attributed to heterosexual contact. Although heterosexual contact with an injection drug user (IDU) accounted for most cases until 1993, cases increased most rapidly among persons reporting heterosexual contact with an HIV-infected partner whose risk was not specified. Findings suggest continued growth of the heterosexual AIDS epidemic. Because of the disproportionate and increasing number of heterosexually acquired AIDS cases among blacks and Hispanics, black and Hispanic communities at risk for HIV infection should be considered a high priority for prevention and education programs specifically targeting heterosexually active adolescents and adults. Epidemiologic and behavioral research and prevention program evaluation are urgent public health priorities to better control and prevent the further spread of HIV among heterosexually active adults and adolescents.

Heterosexual intercourse is the primary mode of HIV transmission worldwide⁽¹⁾. Although male-to-male sex and injection drug use remain the predominant modes of HIV exposure in the United States^(2,3), transmission of HIV from persons infected through these routes to their heterosexual partners has resulted in an AIDS epidemic among heterosexuals. More importantly, the number of AIDS cases attributed to heterosexual contact (referred to here-after as heterosexually acquired AIDS) is increasing at a faster rate than are the numbers of those acquired through other routes^(2-4).

National AIDS surveillance data provide a means to track the HIV epidemic and characterize trends among persons infected by all modes of exposure so that populations at high risk can be targeted for education and prevention activities. In this article, we describe recent trends in heterosexually acquired AIDS in the United States.

METHODS

Study Population

For U.S. national AIDS surveillance, state and local health departments report demographic and geographic data, clinical characteristics, and HIV-exposure information for persons with conditions that meet the AIDS surveillance case definition⁽⁵⁾. Using surveillance data reported through September 1996, we analyzed trends in the numbers and proportions of heterosexually acquired AIDS cases diagnosed from January 1988 through December 1995 among persons 13 years of age and older by half-year of diagnosis and by age, race or ethnicity, sex, geographic region of residence (U.S. census regions)⁽⁶⁾, and partner's HIV exposure risk.

Definitions

For AIDS surveillance, risk information is obtained from the medical record and HIV exposure is classified hierarchically according to the most probable means of transmission. Persons reported with multiple modes of HIV exposure are classified in the category listed first in the hierarchy, except for men with histories of sexual contact with other men and injection drug use, who are classified in a separate category⁽²⁾. Heterosexual contact is considered to be the HIV exposure mode for persons whose only reported risk is heterosexual contact with a partner who is HIV infected or at increased risk for HIV infection, such as bisexual men, IDUs, and persons with coagulation disorders treated with blood or blood products. In the absence of specific behavioral risks for HIV exposure, AIDS cases among persons born in countries where heterosexual transmission of HIV predominates (formerly classified as pattern II countries by the World Health Organization) or among persons having had heterosexual contact with a person born in such countries are not classified as heterosexually acquired⁽⁷⁾.

We defined primary heterosexual transmission of HIV as transmission through heterosexual contact with a bisexual male, an IDU, a person who had received clotting factor concentrates, or a person with transfusion-associated HIV infection. Secondary heterosexual transmission was defined as transmission via heterosexual contact with an HIV-infected partner who has no reported or acknowledged history of male-to-male sex, injection drug use, or receipt of clotting factor concentrates or HIV-contaminated blood or blood products. Although this transmission category may include some cases acquired through heterosexual contact with an IDU or other primary-risk partner(i.e., through "primary" heterosexual transmission), it also represents cases acquired through heterosexual contact with a heterosexually infected partner of unknown risk and may represent the leading edge of another wave of the AIDS epidemic.

Adjustment for Expansion of AIDS Case Definition

Analyses based on the year of AIDS diagnosis, with adjustments for delays in reporting, are typically used to monitor temporal changes in AIDS trends⁽⁸⁾. An expanded AIDS surveillance case definition for adolescents and adults was implemented on January 1, 1993⁽⁹⁾. Reporting criteria added to the definition were HIV-related severe immunodeficiency(<200 CD4⁺ T-lymphocytes per µL or a CD4⁺ T-lymphocyte percentage of total lymphocytes <14) and HIV infection plus pulmonary tuberculosis, recurrent pneumonia, or invasive cervical cancer.

Because most HIV-infected persons develop severe immunosuppression before the onset of AIDS-defining opportunistic illnesses (AIDS-OIs), the inclusion of the CD4⁺ reporting criteria in the surveillance definition has required an additional adjustment to estimate when persons who were reported using the CD4⁺ criteria will develop an AIDS-OI. Estimated AIDS-OI incidence is the sum of the observed AIDS-OI incidence and the incidence based on estimated dates of AIDS-OI diagnosis for persons reported with AIDS based only on severe immunosuppression (as previously defined); both incidences are adjusted for reporting delays (estimated by a maximum likelihood statistical procedure [8] from data reported through September 1996) and anticipated reclassification of cases initially reported without an identified risk.

Trends in incidence were analyzed using the earliest date of diagnosis of any AIDS-OI. For cases with no reported AIDS-OI, this date was estimated using a time-to-AIDS-OI distribution appropriate for the reported CD4⁺ count. Estimated numbers of AIDS-OIs were rounded according to routine presentation of AIDS surveillance data⁽²⁾. For cases reported with a CD4⁺ percentage less than 14 but without a CD4⁺ count, the count was imputed from a regression model fit using an additivity and variance stabilizing transformation^(10,11). The data used to estimate the regression function consist of 5010 paired measurements of CD4⁺ counts and CD4⁺ percentages from HIV-positive patients, and they are a subset of the data used to establish CD4⁺ percent equivalences to CD4⁺ counts of 200 cells/µL and 500 cells/µL for the 1993 revised classification system for HIV infection^(5,10).

The probability of an AIDS diagnosis within a given number of months after a specified CD4⁺ count was estimated using data from the Adult and Adolescent Spectrum of HIV Disease Project, an ongoing survey of medical records of HIV-infected patients receiving medical care from more than 100 clinics, hospitals, and private medical practices in 10 cities in the United States and Puerto Rico⁽¹²⁾. Estimates are based on 9479 patients receiving care from January 1990 through August 1996 who had at least one CD4⁺ determination before the diagnosis of an AIDS-OI. Of these, 4915 developed such an illness; for the remaining patients, time to AIDS was censored at the time of last clinical visit or at death. Each patient was assigned to a CD4⁺ stratum of width 10 cells/µL (e.g., 0 through 9, 10 through 19) based on the first CD4⁺ count meeting the AIDS surveillance definition or the first CD4⁺ count after a CD4⁺ percentage meeting this definition. The Kaplan-Meier procedure was used to estimate the distribution of the time to AIDS-OI in each stratum⁽¹³⁾. Results from log rank and Wilcoxon tests were used to combine the original strata into seven strata defined by counts of 0 to 29, 30 to 69, 70 to 119, 120 to 169, 170 to 249, 250 to 399, and ≥400 cells/µL. (Strata including counts of ≥200 cells/µL are needed because some persons who have been reported with a CD4⁺ percentage of less than 14 have CD4⁺ counts higher than 200 cells/µL.) Time-to-AIDS-OI distributions were then estimated for these strata, again using the Kaplan-Meier procedure.

Redistribution of Cases Reported Without Risk Information

Some persons with AIDS are reported with no specified risk for HIV infection, and additional follow-up is conducted to ascertain risk. Follow-up assessment usually requires review of the patient's medical record and may include a patient interview. While reports are reviewed and until risk is ascertained, the exposure mode for these cases is classified as "no identified risk" (NIR). Follow-up is less likely to have been completed among the more recently reported NIR cases. To adjust for the disproportionate number of NIRs among the most recently reported AIDS cases, all cases with an undetermined exposure mode were redistributed according to historical patterns of reclassification⁽¹⁴⁾. This redistribution was based on the current exposure category distribution of cases, stratified by sex and race or ethnicity (e.g., non-Hispanic, white, other), initially reported as NIR from 1989 through 1993 but which have been reclassified into a defined exposure category or have undergone a complete investigation without having an exposure mode identified. The redistribution is carried out by apportioning current NIR cases, similarly stratified and excepting those for which a complete investigation failed to identify an exposure mode, among exposure categories according to these distributions.

Incidence rates were calculated on an annual basis per 100,000 population. Population denominators for computing rates are based on official post-census estimates from the U.S. Bureau of the Census. Each annual rate is the estimated number of AIDS cases diagnosed during that year, divided by that year's population and then multiplied by 100,000. The denominators for computing race-specific rates are based on extrapolations from the 1980 and 1990 censuses⁽¹⁵⁾. Trends were analyzed for significance at thep < 0.01 level using the chi-square test for linear trend.

RESULTS

All trends described are statistically significant at a level of p < 0.01 unless otherwise stated. Through September 1996, 447,506 persons had been reported as having an AIDS diagnosis from 1988 through 1995. Of these, 38,742 (9%) had been reported with heterosexual contact as the exposure category, and 31,354(7%) were initially reported without a risk for HIV exposure. After estimating the incidence of AIDS-OIs and adjusting for delays in reporting, we estimated that 394,900 persons had AIDS-OIs diagnosed from 1988 through 1995 (Table 1). After redistributing cases initially reported without risk, we estimated that 38,900 (10%) persons with AIDS-OIs diagnosed from 1988 through 1995 had acquired HIV through heterosexual contact.Cited Here... All data in the remainder of this article are derived from this data set, and "AIDS cases" refers to cases estimated to have been diagnosed with an AIDS-OI from 1988 through 1995.

Table 1

From 1988 through 1995, heterosexual contact was the most rapidly increasing mode of HIV exposure. The average annual percent increase in the number of AIDS cases was 28% for heterosexual contact, 12% for injection drug use, 6% for men who have sex with men, 5% for men who have sex with men and inject drugs, and 2% among persons with hemophilia or coagulation disorders. The average annual percent decrease in the number of AIDS cases among blood transfusion or blood product recipients was 2%.

Comparing numbers of AIDS cases diagnosed in 1995 with those diagnosed in 1988, the absolute increases in the number of cases diagnosed were largest among IDU (9500) and men who have sex with men(8900). The absolute increase in number of cases attributed to heterosexual contact (7300), however, was also substantial. While the proportion of AIDS cases attributed to heterosexual contact increased each year, from 5% in 1988 to 15% in 1995, the proportions attributed to all other exposure modes except injection drug use decreased.

From 1988 through 1993, the number of AIDS cases attributed to primary heterosexual contact increased rapidly and until the first half of 1993 accounted for most heterosexually acquired AIDS cases (Fig. 1). While the relative proportion of heterosexually acquired AIDS cases attributed to primary heterosexual transmission has decreased, the number and relative proportion attributed to secondary heterosexual transmission have increased rapidly, from 24% of heterosexually acquired AIDS cases in 1988 to 57% in 1995.

Fig. 1

From 1988 through 1995, heterosexually acquired AIDS accounted for 4% of all AIDS cases among men and 44% among women. Although only a small proportion of total AIDS cases among men were attributed to heterosexual contact, men accounted for 35% of all heterosexually acquired AIDS cases. The incidence of heterosexually acquired AIDS increased steadily among men (from 480 cases diagnosed in 1988 to 3200 in 1995) and women (from 1100 cases diagnosed in 1988 to 5700 in 1995) (Fig. 2A). The average annual increase was 32% among men and 27% among women. Because the number of cases among men increased at a slightly faster rate than cases among women, the overall female-to-male ratio decreased slightly, from 2.3:1 in 1988 to 1.8:1 in 1995 (Table 2). The incidence rate of heterosexually acquired AIDS (cases per 100,000 population) increased from 0.5 to 3.1 among men and from 1.1 to 5.2 among women.

Table 2

Fig. 2

The incidence of heterosexually acquired AIDS cases increased among persons in all age groups (Fig. 2B). Although the median age of persons with heterosexually acquired AIDS was the same as that of all persons with AIDS (35 years), a higher proportion of persons with heterosexually acquired AIDS were younger than 30 years (25%) and older than 49 (14%) than were those with AIDS attributed to other exposure modes (17% younger than 30; 10% older than 49). Among persons between the ages of 13 and 19 years with heterosexually acquired AIDS, 91% were women (compared with 51% to 76% in other age groups), 73% were black (compared with 55% to 59% in other age groups), and 61% lived in the South at the time of AIDS diagnosis (compared with 43% to 50% in other age groups). Heterosexual contact was the predominant mode of transmission among women between the ages of 13 and 29 years and women older than 49 years. Overall, 54% of all AIDS cases among women in these age groups were attributed to heterosexual contact: 68% of cases among women 13 to 19 years old, 54% among women 20 to 29 years old, 55% among women 50 to 59 years old, and 52% among women older than 59.

The number of heterosexually acquired AIDS cases increased among blacks, Hispanics, and whites(Fig. 2C). From 1988 through 1995, the incidence rate of heterosexually acquired AIDS increased among all three racial or ethnic groups but most rapidly among blacks (Fig. 3). In 1995, the rate among blacks (22.1) was 18 times that among whites (1.2), and the rate among Hispanics (8.1) was almost seven times that among whites. Although blacks and Hispanics composed 21% of the U.S. population and 51% of all AIDS cases, they accounted for 75% of heterosexually acquired AIDS cases. AIDS cases among American Indian, Alaska native, Asians, Pacific Islanders, and persons with missing race or ethnicity information totaled 400 cases, and no significant trends were observed.

Fig. 3

Heterosexually acquired AIDS cases increased in all geographic regions of the United States (Fig. 2D), with the South reporting the most rapid increases(average annual increase of 33%) and the largest absolute increase in incidence (3500 cases).

Among persons with heterosexually acquired AIDS, 47% of women and 41% of men reported heterosexual contact with an IDU. While the average annual increase in the number of heterosexually acquired AIDS cases among persons who reported heterosexual contact with an IDU was 18%, cases increased even more rapidly among persons who reported heterosexual contact with an HIV-infected partner whose risk was not specified (average annual increase of 46%) (Fig. 2E). The female-to-male ratio over the 8-year period was lower among persons who reported heterosexual contact with an HIV-infected partner whose risk was not specified (1.4:1) than among persons who reported sex with an IDU (2.2:1)(Table 2).

DISCUSSION

The popular and scientific press have published conflicting reports about the distribution and magnitude of the heterosexually acquired AIDS epidemic^(16-19). Our data clearly demonstrate that from 1988 through 1995 heterosexually acquired AIDS increased rapidly in the United States. These trends in AIDS incidence suggest that in preceding years the incidence of heterosexual HIV transmission also increased rapidly. From these data, HIV seroprevalence data, and estimates of HIV prevalence, we forecast that AIDS cases attributed to heterosexual contact will continue to be reported⁽²⁰⁾. Barring a significant impact of community-based efforts to reduce high-risk behavior in the foreseeable future, HIV transmission from a large prevalence pool of 650,000 to 900,000 HIV-infected adults and adolescents⁽²¹⁾ to their sexual or needle-sharing partners will continue.

Heterosexually acquired AIDS is increasing most rapidly among persons who report sexual contact with an HIV-infected partner whose risk is unreported, unrecognized, or unknown. The leveling of AIDS trends among persons whose heterosexual partners were bisexual men or recipients of HIV-contaminated blood or blood products is consistent with the leveling of trends in those primary risk groups. Persons with heterosexually acquired AIDS remain disproportionately women who report sexual contact with a male partner who injects drugs. With the highest increases observed in the South, Midwest, and West from 1990 through 1994 and the largest proportion and number of cases in the Northeast, AIDS cases among IDUs continue to increase⁽²²⁾ and are likely to increase concomitantly among their heterosexual partners.

The disagreement concerning the potential for more widespread HIV transmission among the heterosexual population reflects the lack of data that clearly differentiate between two different populations of heterosexually infected persons-those infected through primary and those infected through secondary heterosexual transmission. Most cases of heterosexually acquired AIDS are attributed to primary heterosexual transmission, such as sex with an IDU. Persons who are at risk of becoming HIV infected through secondary heterosexual transmission may not be easily identified. Because they may not belong to any of the recognized risk groups targeted by established education and prevention programs, many persons will not perceive themselves or their partners to be at risk and may represent a potential for more widespread heterosexual transmission in the United States.

From these surveillance data, we cannot definitively establish whether the increasing number of AIDS cases observed among persons who report heterosexual contact with an HIV-infected partner whose risk is not specified represent unknown primary heterosexual transmission. Validation of increases in secondary heterosexual transmission among persons in previously unrecognized risk groups (i.e., among heterosexually active persons with no other risks for HIV exposure) will require more careful follow-up to ascertain the risk of these persons and particularly the risk of their partners. Although the HIV-infection status of the patient's partner may be documented with increasing frequency as an increasing proportion of the HIV-infected population is tested for HIV infection, information on the partner's risk behavior is often not readily available in the patient's medical records, which are the primary source of risk information for more than 90% of AIDS cases. The risk of a patient's partner may not be documented in the medical record unless the patient volunteers information on their partner's behaviors or the provider solicits and records such information as deemed relevant to the patient's care.

Because most persons with AIDS attributed to primary risk factors have been men, most persons infected through primary heterosexual contact have been women. Although HIV is transmitted less efficiently from women to men than from men to women^(23,24) secondary heterosexual transmission of HIV from the increasing reservoir of HIV-infected women to their male partners may partly explain the lower female-to-male ratio among persons with AIDS who report sex with an HIV-infected partner whose risk is not specified. The efficiency of HIV transmission may be increased by cofactors such as sexually transmitted diseases (STDs) and genital lesions in either partner^(25-28) and a low CD4⁺ T-cell count or symptomatic HIV disease in the infected partner^(29,30). Although 65% of persons with heterosexually acquired AIDS are women, the steady decline in the female-to-male ratio is consistent with an increasing number of heterosexually active HIV-infected women and the increased efficiency of HIV transmission to men from the growing number of women who have progressed to severe HIV disease^(31,32). Another factor that may account for increasing AIDS incidence among the heterosexual population is the increasing number of persons who exchange sex for drugs, especially crack cocaine, for which an association with STDs and heterosexual HIV transmission has been shown^(33-36).

The overrepresentation of racial or ethnic minorities among persons with heterosexually acquired AIDS in these data is similar to that in surveillance data for other sexually transmitted diseases, including gonorrhea, syphilis, and chancroid, which in the United States have the highest rates among blacks and Hispanics^(37,38).

Although the median age at AIDS diagnosis of persons reported with heterosexually acquired AIDS is the same as that of all persons with AIDS (35 years), a higher proportion of heterosexually acquired AIDS cases are among persons younger than 30 years. These data are consistent with model-based estimates that suggest that between 26% and 50% of all adults and adolescents with heterosexually acquired AIDS were infected with HIV in their teens and early twenties⁽³⁹⁾. These estimates suggest that a substantial proportion of education and prevention programs should target adolescents. Fifteen percent of persons with heterosexually acquired AIDS, however, are older than 49 years, indicating that heterosexually active persons of all ages are at risk.

The incidence of heterosexually acquired AIDS increased most rapidly in the South. These findings are similar to data from seroprevalence studies that report increases in HIV seroprevalence among adult and adolescent childbearing women living in the South^(40,41). Although seroprevalence studies are limited by their lack of complete demographic or risk information, increases in some geographic areas (e.g., the South) in the number of childbearing women with more recently acquired HIV suggest that trends in heterosexual transmission of HIV are continuing to increase in those areas⁽⁴²⁾.

The trends reported in this article may overestimate the role of heterosexual transmission in the AIDS epidemic if exposure modes have been substantially misclassified. Concerns about stigmatization and discrimination may cause patients to deny other risk factors such as injection drug use or male-to-male sex. Increasing surveillance caseloads may prohibit medical and public health personnel from completely ascertaining or documenting exposure modes. Furthermore, patients may deny or be unaware of their partner's risk factors.

Because the completeness and accuracy of risk ascertainment varies by case and geographic location, the Centers for Disease Control and Prevention, in collaboration with six state and local health departments, has implemented a risk-validation project. Information on partner's risk, the extent of exposure mode misclassification, and underascertainment of multiple modes of exposure are being studied by reabstracting medical records, reviewing ancillary records(e.g., STD clinic records), and interviewing health care providers and patients. Preliminary analysis validated heterosexual exposure in 1402 of 1446(97%) AIDS cases initially reported as heterosexually acquired⁽⁴³⁾. Of these, 239 cases (17%) found to be associated with additional risks were reclassified based on the hierarchy of exposure categories.

We may have underestimated the number of heterosexually acquired AIDS cases by analyzing only cases among persons whose most likely mode of exposure was heterosexual contact. Although multiple possible modes of HIV exposure are reported for some persons with AIDS⁽²⁾, the exact mode by which such persons become infected is unknown. For example, a person with AIDS who reports heterosexual contact and injection drug use is not classified as having heterosexually acquired AIDS because injection drug use is considered a more likely mode of HIV transmission. An unknown proportion of cases hierarchically attributed to injection drug use may have been acquired by heterosexual contact with an infected partner^(44,45). Of an estimated 105,500 AIDS cases attributed to injection drug use (among women and among men who did not report sex with other men) from 1988 through 1995, 13,400(13%) were also reported with heterosexual exposure to HIV.

The reported incidence rates most likely underestimate the true impact of HIV infection in the heterosexual population. Because the denominators for each HIV exposure category are unknown, we calculated rates for heterosexually acquired AIDS cases based on the entire U.S. adult and adolescent population. However, not all persons are sexually active, and not all sexually active persons are exclusively heterosexual.

A proportion of cases reported with NIR may result from unrecognized heterosexual transmission among persons who are not aware of their partner's HIV infection or increased risk for HIV infection. Although NIR cases were redistributed in these analyses, the redistribution weights were based on historical data from follow-up of AIDS cases diagnosed from 1989 through 1993⁽¹⁴⁾. Imputation was based on the current distribution of exposure categories among cases initially reported as NIR from 1988 through 1993 which had been reclassified into an exposure category or had undergone complete follow-up without having an exposure mode identified. Using data from these earlier years to impute risk categories is likely to have generated somewhat conservative estimates of heterosexually acquired AIDS.

Currently, the highest incidence of HIV infection appears to be among specific populations at high risk: young men who have sex with men, IDUs and their sex partners, and heterosexual persons with multiple sex partners and high incidence of STDs^(44-47). In agreement with seroprevalence data^(20,40), the AIDS trends presented here indicate that heterosexual transmission is playing an increasingly prominent role in the epidemiology of HIV infection in the United States.

Because a disproportionate and increasing number of heterosexually acquired AIDS cases are among blacks and Hispanics (from 68% in 1988 to 78% in 1995), black and Hispanic communities at risk for HIV infection should be considered a high priority for HIV prevention and education programs specifically targeting heterosexually active adolescents and adults. This target population should also include heterosexually active IDUs who may present multiple opportunities for HIV transmission, whether or not they actively share needles or syringes. Secondary transmission of HIV will continue to be difficult to establish and track with routine HIV and AIDS surveillance methods and will require enhanced behavioral surveillance. To better control and prevent the further spread of HIV among heterosexually active adolescents and adults, epidemiologic and behavioral research of HIV risk factors and evaluation of prevention programs are urgent public health priorities.

Acknowledgments: The authors gratefully acknowledge the contributions of HIV/AIDS surveillance staff from state and local health departments. The authors also thank Phyllis Moir and Dr. Harold Jaffe for their review of the manuscript.