Redefining the Growth of the Heterosexual HIV/AIDS Epidemic in Chicago

A dramatic shift in the relative distribution of the five categories of heterosexual transmission for AIDS cases diagnosed in Chicago since 1991 prompted a mode-of-transmission validation study of what had become the most frequently reported heterosexual exposure: heterosexual relations with a person with AIDS (PWA) or documented HIV infection whose risk is not specified.

Cited Here...: For 395 cases with originally reported heterosexual exposure, one or more of three supplemental data sources were employed: medical records were reviewed, medical providers were interviewed, and patients or proxies (i.e., spouse, significant other, or family member) were interviewed when possible. When reported HIV exposure could not be validated or reclassified, the transmission category employed was"no identifiable risk" (NIR).

Cited Here...: Eighty-five percent (336 of 395 cases) were reclassified into different transmission categories. Most notably, 69% (272 of 395 cases) were reclassified into transmission categories that did not involve heterosexual contact, including NIR. The cumulative percentage of cases attributable to heterosexual contact declined from 8% to 5% as a result of reclassification. Additionally, reclassification resulted in a reduction of nearly 50% in the number of AIDS cases attributable to heterosexual contact diagnosed in 1993 and 1994.

Conclusions: In Chicago, an emerging problem in AIDS surveillance appears to be the use of an ambiguous heterosexual exposure category as a default when other information is not readily available. This study has found the growth in AIDS cases among persons exposed to HIV through heterosexual contact to be much slower than previously perceived. This finding may have important implications for the national debate over the extent to which heterosexual people are being infected and how funding and prevention strategies should be prioritized.

Throughout the course of the AIDS epidemic, the number of cases attributable to heterosexual contact seemed to increase steadily each year in Chicago, from 4% in 1989 to 13% in 1994 (Fig. 1). A similar growth has been seen in the United States overall ⁽¹⁾. Additionally, debates have arisen about the magnitude of the number of cases attributable to heterosexual contact and the implications of this magnitude for prevention and funding policies^(2,3). Further confusing the matter, recent Chicago data suggest that the rise in the number of reported heterosexually transmitted cases may be an artifact of inaccurate surveillance by AIDS case reporting sources.

Fig. 1

The "heterosexual contact" mode of HIV transmission is subcategorized by the Centers for Disease Control and Prevention (CDC) into five expanded modes of transmission⁽⁴⁾. Table 1 demonstrates that the majority of cases with the mode of transmission "heterosexual contact" diagnosed through 1991 were among persons with heterosexual partners who are injection drug users(hetsex w/ IDU). The other categories of heterosexual contact were all lesser contributors to the heterosexual subepidemic up until the early 1990s.

Table 1

Beginning in 1991, the proportion of persons with AIDS (PWAs) reported as having acquired HIV through heterosexual contact with an HIV-infected person whose risk was not known (hetsex w/ HIV risk unk) increased rapidly (Table 1). This category of heterosexual contact is reserved for cases in which a person's heterosexual partner is reported to be HIV infected but the partner's mode of exposure to HIV is not known⁽⁵⁾. In 1991, hetsex w/ HIV risk unk accounted for 39% of the cases associated with heterosexual contact. By 1994, this proportion had grown to 63% and the number of cases in this category had quadrupled. In light of this rapid growth, a study of 395 cases classified as hetsex w/ HIV risk unk was conducted to assess the validity of the report of this particular mode of transmission.

METHODS

Three hundred and ninety-five cases in the Chicago AIDS registry reported through July 1995 and classified hetsex w/ HIV risk unk were selected for the study. The goal of the study was to verify the reported mode of transmission for each case. The protocol for follow-up is provided in Figure 2. The medical record from which each case was reported was examined by a Communicable Disease Control Investigator. If a more specific risk was not identified, or if the originally reported risk could not be verified, care providers were then contacted and interviewed to elicit further information on the mode of transmission for the patient. Again, if a more specific risk could not be determined and the originally reported risk was not verified, an interview with the PWA, or a proxy (i.e., spouse, significant other, or family member) was attempted. Proxies for PWAs deceased before January 1995 were not contacted to request an interview. The interviews were conducted according to a standardized CDC protocol for conducting a"no identifiable risk" (NIR) interview⁽⁶⁾. An additional 30 cases were reported with this risk from sources outside the jurisdiction of the Chicago Department of Public Health. These were not included in the study.

Fig. 2

These investigations resulted in either the reclassification of mode of transmission, the coding of cases as "original report verified", or the closure of cases as NIR. The mode of transmission categories are presented in Table 1. Analysis of data was conducted using PRODAS version 3.2A ⁽⁷⁾.

RESULTS

The 395 cases were reclassified or verified to have been classified correctly in the original report at various stages of follow-up, as indicated by Figure 2. Consistent with the CDC protocol, 118 cases (30%) were reclassified as NIR after the exhaustive follow-up protocol was completed without verifying the originally reported mode of transmission or identifying any others⁽⁷⁾. These were classified in the following NIR closure categories: lost to follow-up (64 cases), PWA dead (49 cases), interview declined (3 cases), and interview did not identify risk (2 cases).

After the review, the cumulative number of cases classified as hetsex w/ HIV risk unk declined from 395 to 59 cases (Table 2). Thus, only 15% of the cases were correctly classified initially, and another 16% were reclassified into one of the other heterosexual categories. After reclassification, the cumulative percentage of persons in whom HIV infection is attributable to heterosexual contact dropped from 8% to 5%. Changes for recent years are even more dramatic: prior to reclassification, the heterosexual contact percentage grew from 4% in 1989 diagnoses to 13% in 1994. After reclassification, a more modest increase from 3% to 7% is noted during the same period (Fig. 1).

Table 2

Table 2 presents the distribution of the 395 cases according to classification or reclassification status, gender, and time. Two hundred and forty-one (61%) of the study cases were men. Excluding those cases found to be NIR, only 15% (24 of 160) of the cases among men were found to be correctly classified, leaving 85% reclassified into known transmission modes. Among women, 30%(35 of 117) of patients were correctly classified; 70% were reclassified into known transmission modes.

Subtracting the 118 cases determined to be NIR from the 395 examined in the study leaves 277 patients with known transmission modes. One hundred and twenty-three of these were valid heterosexual transmission cases, giving a 44% (123 of 277 patients) validity for report of heterosexual transmission. For men, the validity was 23% (36 of 160 patients); for women, the validity was 74% (87 of 117 patients). Table 2 also indicates that the classification or reclassification patterns did not vary much over time, although the largest number of study cases came from the most recent years of diagnosis.

DISCUSSION

The criteria for risk ascertainment for behaviors associated with the transmission of HIV are based on a hierarchy developed by the CDC⁽⁸⁾. This assignment is determined by what is reported by health care providers after interviewing patients about their history of HIV risk behaviors. For surveillance purposes, only one risk category characterizes each case, even if multiple risks are reported (with the exception of men having sex with men[MSM] and injection drug use [IDU]). Although risk ascertainment should be based only on patient self-report, medical history, or both, health care providers occasionally report risk behaviors based on their own assumptions. This practice, which should not be done, is continuously monitored by surveillance personnel.

The level of misclassification related to heterosexual transmission found in this study is consistent with the findings of two other reports. A study conducted in southern Florida found that 30% (50 of 168) of AIDS cases reported in a 13-month period with mode of transmission described as "heterosexual contact" had evidence of a known mode of transmission other than heterosexual contact⁽⁹⁾. This compares with the finding in our study that 39% (154 of 395) of the cases examined had available information on a known mode of transmission in addition to heterosexual contact; however, a notable difference in the methodologies exists. In the Florida study, cases in which no exposure information considered valid for CDC-defined modes of transmission could be identified, even after PWA interview, remained classified as heterosexual contact. The protocol in our study called for reclassification of similar cases as NIR.

The second study was conducted in Italy and examined the concordance of reported mode of transmission of HIV between two national AIDS surveillance systems⁽¹⁰⁾. Although this was not a validation study of the report of HIV exposure, the authors did find a low level of concordance between the two registries among cases reported with heterosexual contact as the mode of transmission. The concordance between the two registries for reported exposures was 94% for IDU and 88% for MSM, but only 69% for heterosexual contact. Thus, the findings suggest substantial misclassification in the reports of heterosexual contact in Italian AIDS registries as well.

Several factors may have an impact on the accuracy of the report of mode of transmission in AIDS cases. The authors of the studies in Florida and Italy hypothesize that patients may be more likely to self-report risk of exposure as heterosexual contact to avoid the social stigmas associated with MSM and IDU^(9,10). However, information on exposure associated with MSM and IDU was readily available from the medical record or care provider for 20% (77 of 395) of those cases examined in our study (Fig. 2). This suggests an additional dynamic, other than patient false-report, contributing to increased misclassification of heterosexual transmission.

The increased case reporting associated with the expansion of the AIDS case definition in 1993^(11,12) may also have contributed to the misclassification of the report of mode of transmission. In Chicago, for 5096 cases reported after the 1993 revision of the definition, 38%(1918 of 5096) would not have been reportable as AIDS cases at the time of report if the definition had not been expanded⁽¹³⁾. We can note that 83% (278 of 335) of the misclassified cases were reported as diagnosed after the expansion of the definition in 1993. Consistent with this,Table 2 demonstrates that for both genders, cases originally classified as hetsex w/ HIV risk unk were predominantly those diagnosed subsequent to the expansion of the case definition in 1993 and are represented by more recent years of diagnosis.

An additional contributor to misclassification may be a misunderstanding of the meaning of hetsex w/ HIV risk unk by AIDS case reporting sources. This is demonstrated by the fact that only 15% (59 of 395) of these cases examined had information available from one of the sources pursued to substantiate the originally reported exposure. However, in many of these cases, mention of multiple sexual partners was made without mention of the serostatus or history of HIV risk behaviors for any of the partners. Presently, multiple sex partners(without one documented as being HIV infected or as having a known history of HIV risk behaviors) is not considered a valid category of heterosexual transmission by the CDC⁽⁵⁾.

As the AIDS epidemic continues, it is natural that larger policy issues will be raised and debated as we learn more about the epidemic. Such issues will also be raised as changes occur in related social forces such as limited funding and the distribution of cases among gender, racial or ethnic populations, and geographic areas, among other considerations.

For example, this study in Chicago has resulted in a decrease of nearly 50% in cases attributed to heterosexual transmission (Fig. 1). Although we examined only data from Chicago, it is possible that a similar audit of reporting from the United States as a whole may produce analogous findings. In the United States, 7.9% (40,038 of 506,538) of AIDS cases reported cumulatively through 1995 are categorized as occurring through heterosexual contact, with 3.7% (18,710 of 506,538) further categorized as hetsex w/ IDU and 3.6% (18,055 of 506,538) categorized as hetsex w/ HIV risk unk⁽¹⁾. Thus, slightly less than one half of the heterosexual contact cases reported in the United States are of the same category as the ones examined in this study. Do these 18,055 cases reflect reality or, to some extent, an incorrect default assignment mode of HIV transmission when true exposure has not been ascertained?

As at least one book (2) and a prominent front-page article in the Wall Street Journal (3) suggest, the public s concerned with these questions. It is up to all of us-medical providers, surveillance personnel, and other interested investigators-to provide them with accurate answers.

Acknowledgments: Work for this paper was partially supported through cooperative agreement funding (Federal Grant#U62/CCU506232-03) provided to the Chicago Department of Public Health Office of HIV/AIDS Surveillance from the Centers for Disease Control and Prevention. We acknowledge the investigative staff with the Office of HIV/AIDS Surveillance-Juan Elias, Catherine House, Paulette Marlow, and Donald Pollard-for their assistance in reviewing medical records and interviewing care providers and persons with AIDS. Additionally, we thank Jade Dell, M.A., for her editorial assistance.