Human immunodeficiency virus–related deaths and prevalent cases of acquired immunodeficiency syndrome in Chicago, Ill.
Age-adjusted human immunodeficiency virus–associated mortality rates, Chicago, Ill, by sex for 1980 through 1997.
Age-adjusted human immunodeficiency virus–associated mortality rates, Chicago, Ill, by race and ethnicity for 1980 through 1997.
Human immunodeficiency virus–acquired immunodeficiency syndrome case fatality rate, Chicago, Ill, by sex in 1995 through 1997. NH indicates non-Hispanic.
Whitman S, Murphy J, Cohen M, Sherer R. Marked Declines in Human Immunodeficiency Virus–Related Mortality in Chicago in Women, African Americans, Hispanics, Young Adults, and Injection Drug Users, From 1995 Through 1997. Arch Intern Med. 2000;160(3):365-369. doi:10.1001/archinte.160.3.365
Copyright 2000 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2000
Declines in human immunodeficiency virus (HIV)–related mortality between 1995 and 1996 were seen across the United States but were small to nonexistent among people in marginalized sectors who are most likely to contract HIV and die of its effects. No comprehensive analysis describing HIV-related mortality in 1997 was available.
To describe Chicago's HIV-related mortality trends up to and including 1997, with specific attention focused on marginalized populations.
An analysis of cross-sectional HIV-related mortality data with emphasis on the years 1995 through 1997 was conducted for Chicago, Ill. Numbers, proportions, and rates of declines in HIV-related deaths were examined for the city as a whole and also among those diagnosed at Cook County Hospital, as a proxy for people with very low socioeconomic status.
Between 1995 and 1996 there was an overall decline of 19% in HIV-related mortality in Chicago but small or no declines among women, African Americans, Hispanics, injection drug users, and people aged 20 to 29 years and more than 50 years. Between 1995 and 1997 there was an overall decline of 61%. At that time the declines were spread more evenly across diverse groups. There were almost no significant differences between the declines for these groups at Cook County Hospital and in the rest of Chicago.
The HIV-related mortality has fallen dramatically in Chicago since 1995, the year of its maximum. During 1997, declines were seen among all groups. Declines were also seen among the most disenfranchised of the city. Access to care and the new combination therapies are apparently sustaining life for many in Chicago.
THE NUMBER of deaths related to human immunodeficiency virus (HIV) increased every year in the United States between 1980 and 1995.1 The first decline occurred in 1996, followed by an even larger one in 1997.2 However, enthusiasm for these marked declines in national HIV-related mortality, most likely resulting from access to primary care and triple combination therapies, has been tempered by disparities in these declines among various subpopulations. In particular, there has been a discouraging lack of progress in declines among groups who are increasingly affected by HIV disease, ie, women, adolescents and young adults, African Americans, Hispanics, and injection drug users (IDUs).3,4 Contributing to the challenges in this area is the fact that neither the federal government nor any local agency has yet published comprehensive AIDS mortality data. Herein we report HIV-related mortality in Chicago, Ill, through 1997, with particular attention to subpopulations that have previously not shown notable decline in HIV mortality.
Deaths attributed to HIV from 1980 through 1997 for Chicago residents were abstracted from mortality files generated by the state of Illinois. Following the standards and protocols described by Greenberg et al,5 an acquired immunodeficiency syndrome (AIDS) death was defined as one for which the underlying cause was coded with an International Classification of Diseases, Ninth Revision,6 primary code of 136.3 (Pneumocystis carinii pneumonia) and 279.1 (deficiency of cell-mediated immunity) from 1980 through 1986 and 042 through 044 (AIDS, symptomatic HIV, asymptomatic HIV) from 1987 through 1997.
The HIV/AIDS Reporting System (HARS) was used to produce all other data about AIDS in Chicago, including the number of people living with AIDS at a given time and modes of transmission of HIV. The modes of transmission used were those defined and mandated by the Centers for Disease Control and Prevention.7 Illinois state law requires that any clinician treating a person with AIDS in Chicago must complete a case report form and deliver it to the Chicago Department of Public Health (CDPH) within 5 days. Although this deadline is rarely met, active surveillance suggests that most cases do eventually get reported. For example, CDPH regularly reviews discharge summaries of all area hospitals, death certificates, etc, and actively pursues all missing information on case reports. This includes such variables as mode of transmission, for which there exists an extensive follow-up protocol. Not every case makes it into HARS, but most do. For example, a study of HARS in 6 states found that it was more than 90% complete for all groups in all states.8
To classify deaths attributed to HIV by mode of transmission, we matched the mortality files with the HARS database. The match was implemented by the director of the Office of AIDS Surveillance of CDPH to ensure confidentiality. A matching utility of HARS was used to generate a "fuzzy match" based on name, date of birth, race, and sex. Among the 6100 decedents with HIV as the underlying cause of death listed in the vital records files since 1990, we were not able to match 613 (10.0%). Possible explanations for the nonmatches are failure of the matching algorithm and the fact that some Chicago residents who died of AIDS were not residents of Chicago when they were diagnosed with the disease. The latter would result in such people not being listed in the Chicago HARS.
In an effort to estimate the decline in HIV-related mortality among the marginalized sectors of Chicago, we examined mortality among patients who received their AIDS diagnosis at Cook County Hospital (CCH). This produced a coherent subset of AIDS cases and mortality that, in turn, could be compared with data from the rest of Chicago. Cook County Hospital is Chicago's only public hospital and serves the poorest sectors, including a large number of people with no health insurance. For example, more than 90% of the patients are African American or Hispanic, less than 3% have private insurance, and 90% of the women with HIV infection cared for at CCH live below poverty level. About 20% of HIV/AIDS patients are homeless.
Numbers of deaths, age-adjusted mortality rates, and case-fatality rates (CFRs) are included in the analysis, as each illuminates a different aspect of the decline in HIV-related mortality. Mortality rates are derived from vital records data and are age adjusted, using the 1940 US population as the standard. Since vital records coding of Hispanic ethnicity in Illinois only started in 1989, trend data are categorized as White and Black (regardless of Hispanic ethnicity) and Hispanic (regardless of race). In Chicago, about 93% of Hispanic people who self-identify with a race do so as White.9 Tests of significance for changes in the number of HIV-related deaths assumed that these deaths followed a Poisson distribution. Case-fatality rates are defined as the number of people who died in a given year of HIV-related causes divided by the number of people living with AIDS. Differences in CFRs were examined by means of the χ2 statistic with Yates correction. All tests were performed at the 95% confidence level.
Figure 1 presents the annual number of HIV-related deaths in Chicago since 1980, along with the number of people living with AIDS as of December 31 of each year. From 1993 through 1995 there was an average of 953 deaths a year. This number declined to 783 in 1996 and 377 in 1997. When compared with the 968 deaths that occurred in 1995, the largest total since the beginning of the epidemic, these correspond to decreases of 19% (P<.001) and 61% (P<.001), respectively.
The demographic distributions of these deaths from 1990 through 1997 are presented in Table 1, along with the percentage decline between 1995 and 1996 and between 1995 and 1997. Note that the number of deaths among females actually increased by 2% between 1995 and 1996, compared with a 22% decrease for males. Between 1995 and 1997 there was a 35% decline in female HIV-related mortality, smaller than the 65% decline for males but much larger than the 2% increase for 1995 through 1996. Table 1 also presents mortality changes for race/ethnicity groups, for age groups, and for transmission modes. Note that in each case, the most marginalized sectors had only small declines between 1995 and 1996 but much larger declines between 1995 and 1997. For example, declines among Black people increased from 12% to 52%; for IDUs, declines increased from 8% to 47%; and for people aged 20 to 29 years, declines increased from 7% to 63%.
Figure 2 contains the age-adjusted mortality rates by sex, setting these recent changes into a longer-term (18 years) perspective. For men, mortality rates rose steadily from 1980 until 1993, remained more or less constant until 1995, and then dropped sharply in 1996 and 1997. The trend for women was similar, although there was no decline until 1997. Figure 3 contains similar data for White, Black, and (since 1989) Hispanic people. Mortality rates for White people and for Hispanic people showed their first decline in 1994, while mortality rates for Black people actually increased through 1995 and then rapidly declined.
As is indicated in Figure 1, the prevalence of AIDS cases (obtained from HARS) continued to increase as the number of AIDS deaths were decreasing. To take this dynamic into account, Figure 4 presents annual CFRs by race/ethnicity for males and females. In 1995, when the number of HIV-related deaths reached its maximum, the CFR was 33% for non-Hispanic Black and non-Hispanic White males and 24% for Hispanic males; the corresponding figures for females were 27%, 25%, and 17%, respectively. The CFR declined in both 1996 and 1997 for all groups, though not at the same rate. In 1997 the respective CFRs were 11%, 6%, and 7% for men and 13%, 9%, and 5% for women. All 3 changes for men were statistically significant (P<.001), as were all 3 for women (P = .02). The overall CFR for Chicago declined from 31% in 1995 to 9% in 1997 (P<.001).
Table 2 presents percentage declines in mortality from 1995 through 1997 for people diagnosed as having AIDS at CCH and the rest of the city. Although the declines at CCH were somewhat smaller than they were for the rest of the city, only 2 differences were statistically significant: the overall decline of 49% at CCH (from 162 deaths in 1995 to 83 in 1997) vs 64% for the rest of the city (P = .03) and for those for whom the virus was transmitted by men who have sex with men, 48% at CCH vs 71% for the rest of the city (P = .02).
The data presented herein demonstrate large declines in HIV-related mortality in Chicago in 1997. The declines, in contrast to those seen in 1996, affect all demographic and transmission groups. Differences remain between marginalized groups and others, but are far smaller than those noted when 1996 HIV-related mortality was compared with that of 1995.
These data were further analyzed by examining a group with an increased likelihood of medical indigence, ie, patients diagnosed as having AIDS at CCH. As noted in Table 2, the findings were similar to the data for Chicago as a whole. In addition, these data are comparable with mortality rates for women in the Women and Children HIV Program at CCH, which declined by 47% (from 90 to 48 deaths) from 1995 through 1997 (M.C., B. Moore, unpublished data, July 26, 1998).
We have not been able to locate any comprehensive analysis comparing HIV and AIDS mortality for 1997 with previous years. The Centers for Disease Control and Prevention has announced a 61% decline in mortality between 1995 and 1997, which exactly matches the overall decline for Chicago for the same period.10 National declines between 1995 and 1996 (28%) were larger than Chicago's (19%), while national declines between 1996 and 1997 (46%) were smaller than Chicago's (52%). However, data comparing declines by sex, race/ethnicity, and transmission mode are not available. Additionally, we are not aware of any reports for cities (or states) that have compared mortality for 1997 with that of previous years.
It is interesting to compare our findings with those derived from comparatively affluent clinical settings. In a series of 1255 patients enrolled in 9 clinics in 8 cities in the United States, Palella et al11 recently reported declines in CFR from 29 deaths per 100 person-years in 1995 to 9 in 1997, a decline of 70%. Despite the clinical differences between the 2 patient groups (most of the patients included in the study by Palella et al were seen in private clinics by physicians who "routinely care for hundreds of HIV-infected patients and thus have extensive experience with HIV"11(p854)), the overall case fatality rates in Chicago (31% in 1995 and 9% in 1997) showed an almost identical decline.
Several factors may have influenced the inclusive declines in HIV-related mortality in Chicago and among those who were first diagnosed as having AIDS at CCH. Most important among them are services and programs that improve and enable access to primary care and support services and reliable access to medications. In an earlier era of HIV care in the United States, Chaisson et al12 demonstrated that declines in morbidity and mortality are no different by sex, race, or risk of HIV when access to care and treatment is ensured. Our findings of broad reductions in mortality that are inclusive of women, minorities, young adults, and IDUs support that finding, although some disparities remain between men and women and between White people and minorities.
Additional factors may influence the disparities within subpopulations. The HIV-related mortality in women is somewhat higher than in men after controlling for CD4 count and disease progression.13 One fifth of deaths in women with HIV are not HIV related, but rather are caused by such factors as chronic liver disease, chemical dependency, domestic violence, and suicide.14 The impact of a comprehensive family HIV program at CCH serving more than two thirds of the HIV-positive women in Chicago is seen by the favorable decline in mortality in the program and in the city.
Of particular interest in these data are the mortality declines among IDUs. Recent data from Baltimore, Md, and Vancouver, British Columbia, suggest that large percentages of IDUs, as high as 60% in one study, are not receiving triple combination therapy.15,16 Several programs for outreach to IDUs are active in Chicago, including an aggressive needle distribution program and a university-based program for street outreach and prevention counseling for IDUs. These programs have engaged thousands of IDUs and have been associated with improvements of social services and entry of IDUs into primary care.17,18 Nonetheless, we note that the lowest HIV mortality reduction in Chicago occurred in African American women (21%), for whom the leading mode of transmission is IDU. As in many cities in the United States, there are inadequate drug treatment slots for the estimated 50 000 IDUs in Chicago, and racial and sex inequities in the access to these services may contribute to these findings.
We found no age-related differences in these data, including young adults aged 20 to 29 years and adults older than 50 years. Asymptomatic adolescents and young adults with HIV infection are difficult to identify with traditional prevention strategies and to engage and maintain in care, and the care of symptomatic young adults with AIDS often includes poor adherence to medications and rigorous outpatient visit schedules.19 It will be important to further study the causes of death from 1995 to 1997 and to perform retrospective case-control analyses of living and deceased young adults in Chicago to identify contributing factors to the improvements in mortality, which include entry and retention in care and access to triple combination therapy. Adults older than 50 years have a higher mortality from AIDS than adults younger than 50 years20; we saw no evidence of a difference in the degree of mortality reduction in this group, which suggests that the benefits of primary care and triple combination therapy are also accessible to older people with AIDS.
Finally, adverse morbidity and mortality from HIV have been associated with limited experience and expertise in HIV care.21 Differences in HIV primary care provision are more likely to occur with subpopulations with complex medical and psychosocial needs, such as women with families and IDUs.22,23 For example, limited experience with IDUs may affect the complex interaction between clinician and patient regarding adherence to triple combination therapy.4 Culturally appropriate training and education for HIV clinicians with specific attention to the diverse needs of women, IDUs, and adolescents is needed for clinicians in HIV care. The presence of an active and aggressive AIDS Education Center and numerous local university-based programs for HIV education is a likely contributor to improving life expectancy for people with HIV in Chicago.
There are potential limitations of the methods in this report. First, risk behavior data derived from HARS are limited, fixed in time, and determined at the time of AIDS diagnosis. As a result, no conclusions can be reached regarding the prevalence of active vs past behaviors such as injection drug use.
Second, we were able to match only 90% of HIV-related deaths to HARS. This does not affect the number of AIDS deaths observed or the sex or race/ethnicity distributions of these deaths (since this information is contained in the vital records files), but it does indicate that for 10% of the deaths we were not able to determine mode of transmission.
Third, there is some arbitrariness about which people with HIV who die are given an underlying code indicating a death from HIV-related causes. We examined this matter by determining how many people listed in HARS who died did not receive an underlying cause of death coded as HIV related but rather had a secondary cause (there could be up to 4 of these) or no HIV-related cause at all. There were no notable trends in these numbers between 1993 and 1997, for men or for women.
Further investigation is needed to understand the nature of the mortality decline and to further increase it in the future. Key questions include the durability of the observed mortality changes, the impact that transmission of multidrug-resistant HIV will have on future responsiveness to highly active antiretroviral therapy, and the extent of further improvement in access to primary care and treatment. To what extent will the various sources of public support noted herein, including the AIDS Drug Assistance Program and the Ryan White CARE (Comprehensive AIDS Resources Emergency) Act, be sustainable in the future? We note that the participation and advocacy of people living with HIV has been critical in Chicago and Illinois in such public policy discussions as the level of state support for the AIDS Drug Assistance Program and the priorities for allocation of Ryan White Title I funds. This is certain to be an important factor in the answers to these questions in the future.
Accepted for publication April 22, 1999.
Mr Murphy is funded through an annual cooperative agreement (federal grant U62/CCU506232-03) by the Centers for Disease Control and Prevention, Atlanta, Ga.
Presented at the XII International Conference on AIDS, Geneva, Switzerland, July 1, 1998 (abstract 43433).
Reprints: James Murphy, MPH, Chicago Department of Public Health Office of HIV/AIDS Surveillance, 333 S State St, Room 2136, Chicago, IL 60604 (e-mail: email@example.com).