Introduction
Since its introduction into widespread clinical use in 1996, highly active antiretroviral therapy (HAART) has become the standard of care for patients with advanced HIV infection living in industrialized countries [1]. Reductions in mortality since the introduction of HAART have been observed in all demographic groups with HIV infection in the USA, Canada, and Europe [2-7], and the number of AIDS cases has also decreased over time in both men and women, in all racial/ethnic groups, and in all HIV transmission risk groups [8]. Disparities in the utilization of appropriate care for HIV infection [9-11], and problems with adherence to complex HAART regimens could result in differences in the effectiveness of treatment among certain demographic groups. Preliminary assessments of HAART effectiveness at the population level have been conducted in Europe and north America on the basis of data from longitudinal studies of HIV [12-14]. Although the overall effectiveness of HAART has been established, studies comparing the effectiveness of HAART by transmission group and demographic factors such as sex, race, and age have produced mixed results [15-17].
In the Johns Hopkins Clinic Cohort, we previously found no differences by age, sex, race, and HIV transmission group in the progression of HIV infection to AIDS or death [18]. This analysis was conducted using data before 1996, and reflected antiretroviral therapy that was simpler to use and less effective than HAART. We wished to determine whether any demographic differences in the clinical effectiveness of therapy had now emerged since the introduction of HAART. In particular, we wondered whether women, African-Americans, and those with injecting drug use (IDU) as their mode of HIV transmission were having the same benefit as men, Caucasians, and those with a non-IDU mode of HIV transmission.
Methods
Study design
The Johns Hopkins AIDS Service provides care for a large proportion of HIV-infected patients in the Baltimore metropolitan area. Longitudinal primary and subspecialty care are integrated in one hospital-based HIV clinic. An observational clinical database has been part of the AIDS Service since 1990. Information from the clinical records is reviewed and abstracted by trained medical record technicians onto structured data collection forms, and then entered into an automated database. The clinic medical record, the main hospital medical record, and various institutional automated databases are abstracted. Comprehensive demographic, clinical, laboratory, pharmaceutical, and psychosocial data are collected at times corresponding to enrolment into the HIV Clinic and at 6 month intervals thereafter. Information on death is obtained from a death registry maintained by the clinic that receives reports from families, funeral homes, other medical institutions, and local coroners. In addition, death records of the Maryland Bureau of Vital Records and the national Social Security death index are regularly searched. Details of database design and the method of follow-up have been published previously [19].
Study population
The study population consisted of HIV-infected patients receiving care at the Johns Hopkins HIV Clinic who enrolled in the cohort after 1 July 1990 and before 30 June 1999 (n = 3547). Our HIV Clinic is the largest HIV clinical practice in Maryland, USA, and our clinic patient population is representative of the demography (age, race, sex, and HIV transmission group) of HIV-infected patients in Maryland [19,20].
Outcome variable
The outcome of interest in the analysis was the development of an AIDS-related opportunistic illness (OI) or death (by any cause). The definition of AIDS-related diseases was based on the 1987 Centers for Disease Control and Prevention revised surveillance case definition for AIDS among adults and adolescents [21]. This system does not include severe immunodeficiency (CD4 cell count < 200 cells/mm3) as part of the AIDS case definition.
Exposure variable
We compared the hazard of an event in those patients receiving care in different calendar periods corresponding to different HIV therapeutic regimens. This analysis provides information about the therapeutic effectiveness of antiretroviral drugs at a population level.
Therapeutic regimens were classified into three categories: monotherapy, combination therapy, and HAART. The patient had to receive at least 3 months of the therapy to be categorized as receiving that regimen. Monotherapy was defined as the use of one nucleoside reverse transcriptase inhibitor (NRTI), including zidovudine, stavudine, zalcitabine, lamivudine, and didanosine. Combination therapy was defined as the use of two NRTI. Consistent with 1999 United States Department of Health and Human Services guidelines, HAART was defined as the use of two or more NRTI with either a protease inhibitor (PI) or a non-nucleoside reverse transcriptase inhibitor (NNRTI), or the use of two or more PI [22]. PI included indinavir, ritonavir, nelfinavir, and saquinavir; NNRTI included efavirenz, nevirapine, and delavirdine.
Fig. 1 illustrates the prevalence of antiretroviral use by patients in this cohort. Follow-up was divided into two therapeutic eras corresponding to different HIV treatment regimens. Era 1 spanned 1 July 1990 to 31 December 1995; this era corresponded with the widespread use of monotherapy and combination antiretroviral therapy for the management of HIV infection. Era 2 spanned 1 January 1996 to 31 December 1999; this era corresponded with the availability and use of HAART.
Data analysis
Comparisons of characteristics of the patient sample between therapeutic eras was performed by continuity-adjusted chi-square (for categories) or Wilcoxon rank-sum test (for continuous). Survival curve estimates were generated using Cox proportional hazards methods to estimate the probability of developing a new AIDS-defining opportunistic infection or dying. Individual patients contributed as many observations for data analysis as calendar time periods (eras) in which they were observed at risk for events of interest. A patient who enrolled into our clinic during era 1 contributed to era 1 risk sets between enrolling and 31 December 1995, the end of era 1. A patient who enrolled into our clinic on or after 1 January 1996 contributed only to era 2. A patient who enrolled into the clinic during era 1, and did not have an event or censor during era 1 could contribute person-time to era 2. For these individuals, it is critical to compare the hazard of developing AIDS or death fairly in those with a similar duration of follow-up in order to minimize any length of survival bias. Our survival analysis incorporated staggered entry for individuals who enrolled in era 1 but contributed person-time in era 2 [23]. For example, a patient who had been followed in the clinic for 2 years before entering era 2 would not contribute any person-time to the risk sets until 2 years after the start of era 2. A patient who had been followed in the clinic for greater than 4 years would not contribute at all to the era 2 analysis, as he had already survived for a duration of time greater than the entire era 2 calendar period. Our survival analysis adjusted the variance estimates for the possible contribution of a single patient to more than one era [24].
Secondary analyses were also performed with an outcome of death only, and with an outcome of AIDS-related OI only. Variables included in the Cox regression models included sex, race, IDU as a risk factor for HIV transmission, baseline CD4 cell levels, and age. The therapeutic era was analysed as a time-dependent covariate. Terms describing the interaction between era and sex, era and IDU status, and era and race were incorporated into the model to assess the extent to which the effects of sex, race, and IDU on developing an AIDS-defining OI or death changed from era 1 to era 2. Survival estimates were stratified by baseline CD4 cell count (≤ 200 cells/mm3, > 200 cells/mm3). For era 1, the baseline CD4 cell count was the level obtained at entry into the clinic. For era 2, the baseline CD4 cell count was the level obtained at entry into the clinic if the patient was enrolled after 1 January 1996 or the CD4 cell count obtained closest to 1 January 1996 if the patient was enrolled before 1 January 1996.
Finally, an analysis was performed of achieving an undetectable HIV-1-RNA level in patients who received a HAART regimen. An undetectable HIV-1-RNA level was defined as less than 400 copies/ml. The HIV-1-RNA level was measured using reverse transcriptase polymerase chain reaction (Roche Molecular Systems, Branchberg, NJ, USA).
All statistical analyses were performed using the Stata 6.0 software package (Stata Corp., College Station, TX, USA).
Results
A total of 2016 patients contributed person-time to era 1 (total person-time 3637 years; median follow-up 2.4 years). A total of 2165 patients (823 enrolled before and 1342 enrolled after 1 July 1996) contributed person-time to era 2 (total person-time 5813 years; median follow-up 2.1 years). The characteristics of the Johns Hopkins HIV Cohort patients eligible for this analysis are presented in Table 1. Overall, the median age of the patients at entry was 37 years. Approximately 31% of patients were women and 77% were African-American. IDU was the most common mode of HIV transmission risk. Most of the patients in the cohort live below the federal poverty level in the United States, earning an annual income of less than US$10 000, and most live in an inner city environment. There were no significant differences in demographic patient characteristics between the two therapeutic eras, except for an increase in HIV transmission by heterosexual contact and a decline by homosexual contact. At enrolment, 42% of patients had a CD4 cell count of 200 cells/mm3 or less. There was a significant difference in the CD4 cell level (P = 0.03) between the therapeutic eras. Between 1 July 1990 and 31 December 1999, 1344 of the patients developed an AIDS-related OI and 1142 patients died (Table 2). Using the combined outcome measure describing the development of the first AIDS-defining OI or death (`event'), a total of 1703 events were observed in the cohort during follow-up. The specific AIDS-related OI also shown in Table 2. Almost all of the AIDS-related illnesses declined from era 1 to era 2 except for non-Hodgkins lymphoma.
During follow-up, 77% of the participants received any type of antiretroviral therapy. As illustrated in Fig. 1, the use of monotherapy declined from over 79% of the treatment in patients receiving therapy in 1990 to less than 5% of the treatment received in 1999-2000. Similarly, HAART regimens accounted for 0% of the treatment before 1995 and 70% of the treatment in 1999-2000.
Cox proportional hazards models were used to determine independent predictors of progression (Table 3). During era 1, only the baseline CD4 cell count was significantly associated with the progression of HIV infection. In era 2, the CD4 count was still associated with disease progression. Women were significantly more likely to experience disease progression than men after adjustment for other variables [relative hazard (RH) 1.25, 95% confidence interval (CI) 1.05-1.47;P = 0.01]. Injection drug users had a slight increase in the risk of progression that did not achieve significance. Increasing age was also associated with an increased risk of progression. In the model incorporating both treatment eras, increasing age was associated with an increased risk of an event, with a RH of 1.016 per one year increase in age. For IDU compared with non-IDU HIV transmission, the RH of an event in era 2 compared with era 1 was 1.39 (95% CI 1.12-1.71;P = 0.002). For women compared with men, the RH of an event in era 2 compared with era 1 was 1.34 (95% CI 1.06-1.69;P = 0.01). Neither race nor heterosexual HIV transmission were significantly associated with disease progression or death in the Cox proportional hazards model, and the interaction terms (race*era and heterosexual transmission*era) describing the differences in outcome in era 2 compared with era 1 were not significant, indicating that the era in which therapy was received did not modify the effect of race on disease progression. The interaction terms (sex*era and IDU*era) describing the effect of IDU and sex on outcome in era 2 compared with era 1 were highly significant, indicating that the effect of these factors on HIV disease progression was modified by the era in which patients received therapy. (Note: similar results were found for a Cox proportional hazards analysis that used only the development of an AIDS-related illness as the outcome event, used only death as the outcome event, or used Pneumocystis carinii pneumonia, Mycobacterium avium complex bacterium, toxoplasmosis, and cytomegalovirus disease as the outcome event. These results are not shown, but are available upon request.)
Fig. 2 and Fig. 3 show plots of the Cox proportional hazards estimates of disease progression by drug use and sex, stratified by the initial CD4 cell count. In both CD4 cell strata during era 1, IDU had a slightly lower risk of disease progression compared with non-IDU HIV transmission. In era 2, however, this situation was reversed. The median time to progression of HIV disease in injection drug users with CD4 cell counts of 200 cells/mm3 of less increased from 320 days in era 1 to 430 days in era 2, a 34% increase in disease-free survival (Table 4). For non-injection drug users with initial CD4 cell counts of 200 cells/mm3 or less, the corresponding median times to progression increased from 260 to 610 days, a 135% increase in disease-free survival. Similar results are seen in men and women and in both CD4 cell strata. Whereas women and drug users had a slightly longer disease-free survival in era 1, their relative improvement in era 2 was substantially less than for men and non-IDU HIV transmission, respectively.
Differences in treatment utilization and treatment success rates (defined as the suppression of HIV RNA to undetectable levels) by members of the cohort were assessed to explore the possible reasons for the variations in HIV disease progression by sex and IDU in era 2 compared with era 1. Stratifying by baseline CD4 cell count, the use of HAART during era 2 and the subsequent suppression of HIV-RNA levels to undetectable levels were examined by sex, race, and IDU. The results of this analysis are presented in Table 5. The percentage of patients on HAART differed significantly by sex and IDU, regardless of the baseline CD4 cell count. Lower rates of HAART utilization were observed among women and injection drug users. Of those receiving HAART, patients who achieved an undetectable HIV-1-RNA level differed only by IDU among those with a baseline CD4 cell count of 200 cells/mm3 or less.
The multivariate Cox proportional hazards analysis in Table 3 was repeated, adjusting for the receipt of HAART (as a time-dependent covariate) and for achieving an undetectable HIV-1-RNA level. The interaction terms of female sex*era (RH 1.17, 95% CI 0.96, 1.45;P = 0.09) and IDU*era (RH 1.18, 95% CI 0.98, 1.41;P = 0.08) were no longer as significant, suggesting that differences in the use and response to HAART were at least partly responsible for the differences by sex and IDU between eras 1 and 2 in disease progression.
Discussion
We have shown that among patients with HIV infection who are enrolled into our clinic cohort, disease progression and survival vary by sex and IDU, but not by race in the era of HAART. In the pre-HAART era, no difference in the time to the development of a first AIDS-defining illness or death was observed by demographic or risk group in this cohort [25]. The advent of HAART during era 2 brought a marked reduction in both the time to AIDS and death compared with era 1 across the entire population included in this analysis, confirming the findings of other studies examining the effectiveness of HAART [3-7,12-17], but the relative benefit to women and injection drug users is less than that seen in men and non-IDU HIV transmission in era 2 compared with era 1. Of note is the fact that there was no independent difference found by race or by non-IDU HIV transmission risk groups adjusting for the patient's sex and IDU HIV transmission.
The results of this analysis differ from reports recently published from several European studies of the effectiveness of HAART. The EuroSIDA Study Group [15] investigated the relationship between HIV risk group and HAART utilization, immunological and virological response to therapy, and survival. Injection drug users in this population were found to be less likely to start HAART regimens compared with non-injection drug users, but among those who did initiate HAART, no differences in immunological or virological response to therapy or survival were observed [15]. Findings from the Swiss HIV Cohort Study [16] were similar to those of the EuroSIDA study [15], indicating that despite lower rates of HAART utilization by HIV-infected injection drug users, progression to clinical AIDS and survival did not differ by risk group. In addition, no differences in disease progression or survival were detected by sex or educational level [16]. Another study published by the Lazio AIDS Surveillance Collaborative Group [17] examined temporal trends in survival by demographic and risk factors after the introduction of HAART. Similar to our findings, this group did detect a difference in the RH of death by sex, comparing outcomes in 1997-1998 with outcomes in 1993, but the difference was not statistically significant.
It is possible that the differences between those studies and the findings reported here from the Johns Hopkins HIV Clinical Cohort may be related to differences in the approach employed to examine temporal trends. The Swiss HIV Cohort Study group [16] examined trends in disease progression and survival by sex and HIV risk group subsequent to the availability of HAART in 1996. Although the Swiss HIV Cohort began enrolling participants in 1988, outcomes before and after HAART became available were not compared. Likewise, the EuroSIDA Study Group [15] did not compare trends in disease progression and survival before and after HAART became available as enrolment in the study began in mid-1994. The Lazio AIDS Surveillance Collaborative Group [17], by contrast, did examine trends in disease progression and survival pre- and post-HAART. Follow-up of the study population of the Lazio AIDS Surveillance Collaborative Group ended in mid-1998, however, less than 2 years after HAART became available. This may have limited their ability to detect emerging differences in disease progression and survival by sex and HIV risk group.
As our data are from a single HIV care site, it is possible that our results may not generalize more broadly, although they may be most likely to generalize to urban US HIV care. To support this, we would point out that the HIV Cost and Services Utilization Study [10] reported that access to antiretroviral therapy and appropriate care for the management of HIV infection varied significantly by sex, HIV risk group, race, and insurance status in the United States from 1996 to 1998. Although PI and NNRTI use increased across all groups during the follow-up period, women, African-Americans, and injection drug users were less likely to receive PI or NNRTI therapy and were less likely to use needed HIV services in both 1996 and 1998 [10]. The results from our cohort may be demonstrating the clinical impact of these disparities on women and injection drug users.
The differences in disease-free survival by IDU and sex that are now apparent in our cohort may be caused by a combination of factors, including differences in the utilization of therapy and possibly problems with adherence to complex drug regimens. HIV-infected women and injection drug users in our cohort are less likely to be on any form of antiretroviral therapy compared with men and non-injection drug users. In general, the non-use of antiretroviral therapy has been shown to be more frequent in active injection drug users without symptomatic disease who have less contact with the healthcare system [25]. Even in settings that provide free HIV therapy, low rates of utilization among eligible injection drug users have been observed [26]. Provider concerns of non-compliance because of the unstable living conditions of injection drug users and the potential for the subsequent development of drug resistance may also result in the under-prescribing of HAART regimens in this subgroup of HIV patients [25]. Optimal results of HAART require adherence to complex dosing regimens that may overwhelm many patients, especially current injection drug users and women who lack social support [27]. In addition to problems with adherence, we have shown that injection drug users in our cohort are more likely to miss clinic appointments, a risk factor for virological failure on HAART [28].
We do not know how hepatitis C co-infection might have influenced our results. Since the measurement of hepatitis C antibody became available in the mid-1990s, we have learned that approximately 90% of IDU patients are co-infected with hepatitis C. We suspect that co-infection with hepatitis C was likely to be almost as prevalent in our IDU patients in era 1, before we could measure it reliably. Hepatitis C may accelerate HIV progression [29], although such an effect is not as clear in our cohort [30].
Conclusion
HIV can be successfully managed with treatment regimens that are both expensive and demanding of patients. Since HAART became available, the rates of opportunistic infections and death have plummeted across all patient groups; however, these gains may not be uniformly distributed among demographic and risk groups. Disparities in therapy utilization can have a negative impact on HIV disease progression in traditionally disadvantaged groups. As differences in disease-free survival have already become manifest, an improvement in antiretroviral and other health service utilization among women and those with a history of IDU is needed.
References
1. Carpenter CC, Cooper DA, Fischl MA. et al. Antiretroviral therapy in adults: updated recommendations of the International AIDS Society - USA Panel. JAMA 2000, 283: 381 -390.
2. Moore RD, Chaisson RE. Natural history of HIV infection in the era of combination antiretroviral therapy. AIDS 1999, 13: 1933 -1942.
3. Mocroft A, Vella S, Benfield TL. Changing patterns of mortality across Europe in patients infected with HIV-1. Lancet 1998, 352: 1725 -1730.
4. Palella FJ Jr, Delaney KM, Moorman AC. et al. Declining morbidity and mortality among patients with advanced human immunodeficiency virus infection. N Engl J Med 1998, 338: 853 -860.
5. Forrest DM, Seminari E, Hogg RS. et al. The incidence and spectrum of AIDS-defining illnesses in persons treated with antiretroviral drugs. Clin Infect Dis 1998, 27: 1379 -1385.
6. 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: 365 -369.
7. Detels R, Munoz A, McFarlane G. et al. Effectiveness of potent antiretroviral therapy on time to AIDS and death in men with known HIV infection duration. JAMA 1998, 280: 1497 -1503.
8. Centers for Disease Control and Prevention. HIV/AIDS surveillance report, 1999, vol. 11. 1999; pp. 1 -43.
9. Moore RD, Stanton D, Gopalan R, Chaisson RE. Racial differences in the use of drug therapy for HIV disease in an urban community. N Engl J Med 1994, 330: 763 -768.
10. Shapiro MF, Morton SC, McCaffrey DF. et al. Variations in the care of HIV-infected adults in the United States. Results from the HIV Cost and Services Utilization Study. JAMA 1999, 281: 2305 -2315.
11. Moore RD, Hidalgo J, Sugland B, Chaisson RE. Impact of zidovudine on the natural history of acquired immune deficiency syndrome. N Engl J Med 1991, 324: 1412 -1416.
12. Gebhardt M, Rickenbach M, Egger M. Impact of antiretroviral combination therapies on AIDS surveillance reports in Switzerland. AIDS 1998, 12: 1195 -1201.
13. Hogg RS, Yip B, Kully C, Craib KJ, O'Shaughnessy MV, Schechter MT, Montaner JS. Improved survival among HIV-infected patients after initiation of triple-drug antiretroviral regimens. Can Med Assoc J 1999, 160: 659 -665.
14. Hogg RS, Heath KV, Yip B, Craib KJ, O'Shaughnessy MV, Schechter MT, Montaner JS. Improved survival among HIV-infected individuals following initiation of antiretroviral therapy. JAMA 1998, 279: 450 -454.
15. Mocroft A, Madge S, Johnson AM. et al. A comparison of exposure groups in the EuroSIDA Study: starting highly active antiretroviral therapy (HAART), response to HAART, and survival. J Acquir Immune Defic Syndr 1999, 22: 369 -378.
16. Junghans C, Low N, Chan P, Witschi A, Vernazza P, Egger M. Uniform risk of clinical progression despite differences in utilization of highly active antiretroviral therapy. Swiss HIV Cohort Study. AIDS 1999, 13: 2547 -2554.
17. Porta D, Rapiti E, Forastiere F, Pezzotti P, Perucci CA. Changes in survival among people with AIDS in Lazio, Italy from 1993 to 1998. Lazio AIDS Surveillance Collaborative Group. AIDS 1999, 13: 2125 -2131.
18. Chaisson RE, Keruly JC, Moore RD. Race, sex, drug use, and progression of human immunodeficiency virus disease. N Engl J Med 1995, 333: 751 -756.
19. Moore RD. Understanding the clinical and economic outcomes of HIV therapy: the Johns Hopkins HIV Clinical Practice Cohort. J Acquir Immune Defic Syndr 1998, 17 (Suppl.) : S38 -S41.
20. Centers for Disease Control and Prevention. HIV/AIDS surveillance report, vol. 12. 2000; No.1.
21. Centers for Disease Control and Prevention. Revision of the CDC surveillance case definition for acquired immunodeficiency syndrome. MMWR 1987, 36 (Suppl.) : 1S -15S.
22. Guidelines for the use of antiretroviral agents in HIV-infected adults and adolescents. MMWR 1999, 48(RR-13):1-27.
23. Munoz A, Hoover DR. Use of cohort studies for evaluating AIDS therapies. In:AIDS clinical trials. Finklestein DM, Schoenfeld DA (editors). New York NY: John Wiley and Sons Inc.; 1995. pp. 423 -446.
24. Lin DY, Wei LJ. The robust inference for the Cox proportional hazards model. J Am Statist Assoc 1989, 84: 1074 -1078.
25. Celentano DD, Vlahov D, Cohn S, Shadle VM, Obasanjo O, Moore RD. Self-reported antiretroviral therapy in injection drug users. JAMA 1998, 280: 544 -546.
26. Strathdee SA, Palepu A, Cornelisse PG. et al. Barriers to use of free antiretroviral therapy in injection drug users. JAMA 1998, 280: 547 -549.
27. Cheever LW, Keruly JC, Moore RD. Antiretroviral use and adherence predicted by active illicit drug use. 39th Interscience Conference on Antimicrobial Agents and Chemotherapy. San Francisco, CA, 26-29 September, 1999 [Abstract 591].
28. Lucas GM, Chaisson RE, Moore RD. Highly active antiretroviral therapy in a large urban clinic: risk factors for virologic failure and adverse drug reactions. Ann Intern Med 1999, 131: 81 -87.
29. Greub G, Ledergerber B, Battegay M. et al. Clinical progression, survival, and immune recovery during antiretroviral therapy in patients with HIV-1 and hepatitis C virus coinfection. The Swiss HIV Cohort Study. Lancet 2000, 356: 1800 -1805.
30. Sulkowski M, Thomas DL, Mehta S, Moore RD. Effect of HCV infection on HIV disease progression and survival among HIV-infected adults. 8th Conference on Retroviruses and Opportunistic Infections. Chicago, Illinois, 4-8 February, 2001 [Abstract 34].
© 2001 Lippincott Williams & Wilkins, Inc.