Introduction
In 1996, protease inhibitors (PIs) became available to treat individuals with human immunodeficiency virus (HIV) infection, and were demonstrated to reduce plasma HIV-RNA to below detectable levels [1–3]. Use of highly active antiretroviral therapy (HAART) was quickly adopted in several populations. In the Multicenter AIDS Cohort Study of homosexual and bisexual men, HAART use at the population level rose from 3% in 1996 to 65% in mid-1998 [4]. In a random sample of HIV-infected individuals in the HIV Cost and Services Utilization Study (HCSUS), 85% of participants eligible for therapy (CD4+ cell count less than 500 × 106 cells/l) reported receiving a PI or non-nucleoside reverse transcriptase inhibitor (NNRTI) treatment by 31 January 1998 [5]. In this national study, inadequate HIV care was more common among Blacks and Latinos, the uninsured and Medicaid-insured, women and risk groups other than men who had sex with men, even after adjusting for CD4+ cell counts. These data indicate that use of HAART may be less common in those with poor access to health care.
Two recent cross-sectional studies of HAART use among IDUs have been reported. In Baltimore, Maryland, we showed that 14% of treatment-eligible IDUs reported HAART use between July 1996 and June 1997; 49% reported no treatment [6]. Factors associated with reporting no antiretroviral therapy use included active drug use, sub-optimal HIV health care, not receiving drug treatment and recent incarceration. During the same period, in Vancouver, British Columbia, where antiretroviral therapy is offered free to all HIV-infected persons who meet International AIDS Society–USA panel (IAS–USA) guidelines [7], only 40% of eligible IDUs received any antiretroviral therapy, and 27% received HAART [8]. Younger individuals, females, those not currently enrolled in drug treatment, and those with inexperienced physicians were less likely to be receiving HAART. Using our cohort of well-characterized IDUs and factors that were found to be associated with HAART use in our previous cross-sectional study [6], we identified IDUs who were treatment-eligible between 1 January 1996 and 30 June 1999 and examined factors associated with the time to initiating HAART.
Methods
Study design and follow-up
A total of 3360 persons who were at least 18 years old were recruited in 1988–1989 and 1994 into a longitudinal study of the natural history of HIV infection in IDUs in Baltimore, Maryland, known as the AIDS Links to the Intravenous Experience (ALIVE) study. All participants were AIDS-free at the time of recruitment and gave a history of non-medical injection of illicit substances within the previous 10 years. Subjects provided informed consent (approved by the Committee on Human Research, Johns Hopkins School of Hygiene and Public Health) and were followed at intervals of 6 months with repeat interviews, physical examinations and blood specimen collection. The interviews included current information regarding demographic and health care utilization factors, current drug and sexual behavior, AIDS symptoms, and use of specific antiretroviral agents. HIV seroconverters were prospectively identified during the study by routine serologic HIV antibody tests confirmed by Western blot. AIDS diagnosis was made by confirmed physician diagnosis or hospital discharge summary. Deaths were monitored by follow-up and through annual review of death records. Details of the study methods have been published [9].
Study population
The study population was made up of HIV-infected IDUs who were alive on 1 January 1996 (n = 565). Of these, 528 (93.5%) had a CD4+ cell count less than 500 × 106 cells/l between 1 January 1996 and 30 June 1998 and were thus eligible to receive HAART according to 1996 IAS–USA guidelines [10]. These 528 participants contributed 931 person–years of observation prior to censoring at first HAART use or at the end of study follow-up.
Outcome variables
We determined the time from having a CD4+ cell count less than 500 × 106 cells/l to initiation of HAART. Those who did not report HAART during follow-up were censored at 30 June 1999 and deaths were censored at their last study visit during follow-up. Antiretroviral therapy use was classified as HAART, non-HAART, and not on therapy. Using IAS–USA guidelines, HAART was defined a priori as one of the following regimens [7,10,11] : two nucleoside reverse transcriptase inhibitors (NRTI) + PI; two PIs; two NRTIs + NNRTI; or PI + NRTI + NNRTI. Non-HAART regimens were defined as: dual NRTI, any other two-drug combination (PI + NRTI or NNRTI + NRTI), or monotherapy.
Antiretroviral therapy use in the 6 months prior to each semi-annual visit was based on participants’ self-report. Using an interviewer-administered questionnaire, participants were asked, ‘Did you ever take [specific medication] during the past 6 months?’ The question was repeated for use of combinations of antiretroviral drugs.
Predictor variables
Substance use in the 6 months prior to each semi-annual interview was the main exposure variable of interest. The principal measure was self-reported injection at each visit. Consistent substance use was defined as reporting injection at two consecutive visits (as compared to abstinent and intermittent users) during follow-up. This variable as well as other exposures were evaluated and allowed to change over time. Other variables of interest were health care access and utilization. Factors influencing access to care included having health insurance, identifying a regular source of health care over a 2 year period (other than ALIVE study visits), and having continuity of care (using the same health care provider at least 90% of the time). Three measures of inadequate utilization were constructed on the basis of results from the HCSUS study: (1) fewer than two ambulatory visits in each 6 months, (2) at least one emergency department visit not leading to hospitalization, and (3) at least one hospitalization.
On the basis of previous findings in this population, other factors potentially influencing HAART initiation include gender, ethnicity, age, employment, homelessness, incarceration, and income [6]. Methadone maintenance treatment was hypothesized to be positively associated with initiating HAART, whereas recent detoxification was hypothesized to be negatively associated with starting HAART, as it reflects active drug use [6,8] Finally, poorer health status [lower CD4+ cell count, AIDS diagnosis, or self report of two or more symptoms commonly associated with AIDS (symptoms that last longer than 2 weeks: diarrhea; unexpected weight loss of 10 pounds or more; shortness of breath; fatigue; and oral thrush)] was hypothesized to be associated with commencing HAART. Rate of CD4+ cell count change was calculated from longitudinal counts measured between semi-annual visits and expressed as an annual rate (number of cells lost or gained per year) to account for the variable time lag between visits.
Statistical methods
The study population was classified into four groups according to self-reported antiretroviral therapy use during follow-up: (1) no antiretroviral therapy use; (2) only non-HAART use; (3) both HAART and non-HAART use; and (4) only HAART use. A simple descriptive comparison was then performed between the groups, using χ2 tests for categorical variables and analysis of variance for continuous variables.
We evaluated the influence of exposure variables as both fixed and as time-dependent covariates. Time to initiating HAART was described using Kaplan–Meier survival methods that estimate the proportion that did not receive HAART at each point in time over the follow-up period. To assess the association of factors with the probability of initiating HAART, we utilized a Cox proportional hazards model with fixed and time-dependent covariates. Covariate values were determined by assigning the measurement at each visit to the preceding time interval. Clinical variables (such as CD4+ cell count) were lagged one visit. Univariate associations were evaluated first, and the relative hazards (RHs) with 95% confidence intervals (CIs) were calculated. The RH approximates the relative risk for each factor. To avoid the term hazard, in the context of a positive outcome, we prefer to interpret a RH > 1 as having a higher probability of initiating HAART whereas RH < 1 indicates reduced probability of starting HAART.
Results
A total of 528 individuals were eligible for HAART during follow-up and attended at least one follow-up visit over the 3.5 year period (mean = 4.3 visits). Overall, 178 IDUs initiated HAART before 30 June 1999. Of the remaining 350 who did not start HAART during follow-up, 175 were seen at least once in 1999, 73 participants were last seen in 1998, and 46 participants were last seen in 1997. A total of 65 IDUs died during follow-up and were censored at the time of last visit.
Figure 1 shows the estimated survival curve of time to first HAART use over the study interval beginning at the time when a participant became treatment-eligible with a CD4+ cell count less than 500 × 106 cells/l. After 1 year, 6% (95% CI, 4–8%) began HAART; by 2 years, 23% (95% CI, 20–28%); by 3 years, 38% (95% CI, 34–43%); and 59% (95% CI, 45–73%) by 30 June 1999. These estimates take into account the different lengths of follow-up of participants and censoring.
;)
Fig. 1.:
Kaplan–Meier estimates (and 95% confidence intervals) for initiation of highly active antiretroviral therapy (HAART) among Baltimore injecting drug users, 1 January 1996 to 30 June 1999.
Table 1 shows the Cox univariate analysis of socio-demographic, injection drug use, health care utilization, and clinical factors on initiating HAART. Among the socio-demographic factors, older age (P = 0.002) and higher education (P = 0.017) were associated with increased probability of beginning HAART, whereas incarceration (P < 0.001) and consistent injection drug use (P < 0.001) were not associated with HAART initiation. Higher income (P = 0.10) and homelessness (P = 0.020) were associated with commencing HAART, although they were marginally significant. Overall, the probability of initiating HAART did not differ by gender, ethnicity, or employment.
Table 1: Cox univariate hazard ratios for initiating highly active antiretroviral therapy, 1 January 1996 to 30 June 1999.
Any injection of drugs was strongly associated with not initiating HAART (P < 0.001). Relative to abstinence, consistent injection drug use was associated with a reduction of over 50% in the probability of beginning HAART (P < 0.001), whereas intermittent use was associated with a 20% reduction in probability, although this was not significant (Fig. 2). Overall, enrolment in methadone maintenance was associated with a higher probability of initiating HAART (P < 0.013). However, this effect was limited to male participants only. In our cohort, although women were more likely than men to be enrolled in a methadone maintenance program [odds ratio (OR), 2.76; 95% CI, 2.11–3.61], their participation was not associated with initiating HAART (RH, 0.51; 95% CI, 0.24–1.10) (Fig. 3).
Fig. 2.:
Kaplan–Meier estimates for initiation of highly active antiretroviral therapy (HAART) among Baltimore injecting drug users by different patterns of illicit drug use, 1 January 1996 to 30 June 1999.
Fig. 3.:
Kaplan–Meier estimates for initiation of highly active antiretroviral therapy (HAART) among Baltimore injecting drug users by gender and methadone enrollment, 1 January 1996 to 30 June 1999.
On average, IDUs with health insurance were over three times more likely to begin HAART (P < 0.001). Not surprisingly, participants reporting having continuity of care were almost four times more likely to initiate HAART (P < 0.001). Similarly, having a regular source of health care (other than ALIVE study visits) was strongly associated with starting HAART (P < 0.001). Among health care utilization measures, participants with less than two ambulatory care visits were, on average, 80% less likely to initiate HAART (P < 0.001). Finally, all clinical variables [AIDS symptoms (P = 0.00), AIDS diagnosis (P < 0.001), and low CD4+ cell count (P < 0.001)] were significantly associated with initiation of HAART in Cox univariate analysis.
An interaction term for gender and methadone treatment was entered into the final multivariate model, shown in Table 2. The main effect of gender was not entered into the model since the effect of methadone increased the probability of initiating HAART for men but had no effect on initiating HAART for women in our study.
Table 2: Cox multivariate hazard ratios for initiating highly active antiretroviral therapy, 1 January 1996 to 30 June 1999.
In the Cox multivariate model, consistent drug use was independently associated with a 58% decrease in probability of initiating HAART after adjusting for other variables (P = 0.001). Men enrolled in methadone maintenance were 80% more likely to commence HAART, whereas no such effect was observed for women (P = 0.052). Health care access, defined as having health insurance and having a regular source of care (at least 90% health care visits are to the same provider), were associated with initiating HAART (RH, 2.05 and 1.74, respectively). In addition, after adjusting for factors in the model, IDUs who reported antiretroviral therapy use prior to 1996 (monotherapy and dual combination therapy) were 2.3 times more likely to initiate HAART during follow-up than those who were treatment-naive (P < 0.001). With respect to clinical health status, each increase in CD4+ cell count of 100 × 106 cells/l was associated with a 15% decrease in probability of starting HAART after adjusting for other factors (P = 0.011).
In addition to exploring the factors associated with initiation of HAART, we further characterized IDUs who reported using non-HAART only and those who reported using no therapy at all. In Table 3, we characterize the participants in our cohort by antiretroviral therapy experience throughout the study period. Overall, 32% of participants remained treatment-naive over the 3.5 year follow-up. Over one-third reported non-HAART use only and only 6% reported exclusive use of HAART. Twenty-eight percent reported using both non-HAART and HAART during follow-up, with the majority (138 of 148 or 93%) reporting a switch from a non-HAART to a HAART regimen. Among participants who initiated HAART during follow-up, 159 (89.3%) initiated HAART that contained a PI while 19 (10.7%) started HAART regimens that were PI-sparing.
Table 3: Descriptive comparison (percentages and means) of injecting drug users by antiretroviral therapy (ART) during follow-up: 1 January 1996 to 30 June 1999.
A close examination of Table 3 reveals that the observed trends in the presence of various determinants across the four treatment groups agree with the results of the univariate Cox models shown in Table 1. The only exception is past hospitalizations where the percentage hospitalized is similar between non-HAART and HAART groups.
Comparison of mean visit-to-visit changes in CD4+ cell counts over the entire study period indicated that those not on therapy had a steeper decline (−7.3 cells/year) compared with those on therapy. Participants receiving non-HAART only had mean changes in CD4+ cell counts (−7.1 cells per year) that were similar to those receiving no ART (−7.3 cells per year). As expected, participants receiving HAART during follow-up had increasing CD4+ cell counts (+24.9 cells per year).
Antiretroviral therapy experience prior to 1996 was strongly associated with subsequent initiation of HAART. Among participants who received antiretroviral therapy (monotherapy and/or dual combination therapy) before 1996, 50% began HAART by June 1998 (mean 2.56 years follow-up), and 85% began HAART by 30 June 1999. Only 39% of treatment-naive IDUs initiated HAART during follow-up.
Discussion
Since PIs became widely available in 1996, many IDUs have initiated HAART. As of 30 June 1999, 59% of our cohort had a history of receiving HAART. It is encouraging that this percentage increased since our previous cross-sectional analysis [6]; however, the majority of patients initiating HAART were previously on zidovudine monotherapy or dual combination therapy. Although the proportion of individuals receiving these non-HAART regimens decreased significantly between 1996 and mid-1999, it remains unacceptably high, at 34%. These regimens are no longer recommended because they are unable to maximally suppress viral replication and therefore may lead to the development of drug resistance that would limit future treatment options. It has been shown that individuals receiving HAART who have had prior treatment experience are less likely to achieve an undetectable viral load relative to those who were previously treatment-naive [12].
As longitudinal viral load data were not available for all participants in our study, treatment eligibility was based on CD4+ cell count alone. Although plasma HIV-RNA level is recognized to be an important prognostic indicator and a determinant of the need for therapy, this marker was not widely available in early 1996 [10], the start date of this analysis. Although the 1996 IAS–USA guidelines acknowledged the value of viral load measurements in guiding the decision of whether to initiate therapy, symptomatic disease and a CD4+ cell count of less than 500 × 106 cells/l were felt to be the primary indicators of the need for treatment. Therefore, the data utilized in this study closely reflects that which was used for clinical decision-making when HAART became available. It was not until 1997 that the IAS–USA recommended using viral load to guide the decision on starting therapy [7]. Given that the population of IDUs in our study had a mean CD4+ cell count of 270 × 106 cells/l (interquartile range of 150 to 370 × 106 cells/l) it is unlikely that we excluded individuals who were treatment-eligible from the analysis.
Antiretroviral therapy use in the ALIVE study is based on self-report. Although this is a limitation in our analysis, there is an indication that use of specific antiretrovirals was not misreported. Among the 528 participants (2247 person–visits) in our analysis, only 23 individuals (29 person–visits) had reported regimens containing dual-NRTI combinations that are less commonly prescribed and not recommended due overlapping toxicity (zalcitabine + didanosine, zalcitabine + stavudine), antagonism (stavudine + zidovudine) or lack clinical data (zalcitabine + lamivudine). If data collected on specific use of medications were inaccurate, then presumably a greater number of participants would report the use of rarely prescribed antiretroviral combinations. However, it is possible that antiretroviral therapy use may be under-reported.
Approximately one-third of the cohort remained treatment-naive over the study period, although all were treatment-eligible on the basis of CD4+ cell count. Among these participants, many denied injecting drugs during follow-up, suggesting that they may have poorer access to care in comparison with individuals receiving some form of therapy. Ironically, these treatment-naive IDUs would have potentially received greater treatment benefit than IDUs who have previously received non-HAART [12]. Nevertheless, poorer access to care may not be the only reason many IDUs are treatment-naive, as these patients tend to be healthier than those on therapy: they are significantly less likely to have an AIDS-defining illness, experience HIV-related hospitalizations less frequently, and have higher CD4+ cell counts. The lack of treatment may reflect patient or provider decisions to delay treatment until it is deemed necessary or when the patient is ready to commence HAART [13].
Data from the HCSUS represents individuals receiving ongoing care; the sampling strategy used thus under-represents those with access barriers to care [14]. Although 20% of all HCSUS participants were uninsured and 48% were covered by Medicaid, among IDUs these percentages were 15% and 71%, respectively [14]. In our cohort, a larger percentage (33%) of IDUs are uninsured, and fewer (54%) are covered by Medicaid, underscoring the strong association between health care access and the initiation of effective treatment. Unlike the national data, however, our participants who reported Medicaid insurance were as likely to receive HAART as those who were privately insured. The level of HAART use was substantially lower among our cohort of drug users than the HCSUS drug users; this could be due to differences in sampling strategies in the two studies.
Poor utilization of health care among IDUs can be partially explained by the fact that HIV-infected IDUs seek medical attention later in the course of disease [15–19]. In the present study having a low CD4+ count during the previous 6 months was associated with receiving HAART and recently seeing a physician.
Once access to care has been addressed, non-HAART use may reflect provider caution in prescribing HAART. Physician experience with HIV care has been shown to be associated with IDUs receiving optimal therapy [8]. Although we lack data on the characteristics of the HIV care providers utilized by our cohort, less experienced providers may continue to treat patients with non-HAART regimens until the patient demonstrates adherence with their medication and clinic visits and substance abuse is under control. Some providers may believe that if IDUs are less likely to attain undetectable viral loads than non-IDUs because of poorer adherence [20] there could be potential transmission of multi-drug resistant HIV [21,22]. A comparison of exposure groups in the EuroSIDA study revealed that IDUs were indeed less likely to commence HAART, but among those who did, their response to therapy was similar to other risk groups [23]. Due to the rapid development of new drugs and continually evolving treatment strategies, staying abreast of the most advanced information can be challenging for providers. As a result, patients seen by primary care providers who see few HIV-infected patients may receive sub-optimal care [24,25]. On the other hand, it is possible that less experienced clinicians may follow treatment guidelines too rigidly, prescribing HAART to all patients without first assessing their ability to adhere to therapy [13]. However, to ensure long-term success of antiretroviral treatment in patients who can wait to initiate HAART, experienced providers recommend delaying therapy, until adherence-related issues are addressed [13,26]. Such issues, which include drug and alcohol use, active mental illness, including depression, and homelessness, require referrals to substance abuse treatment, psychiatrists, and social workers. Other factors that may impact adherence to antiretroviral therapy include poor knowledge about HIV infection and treatment and the lack of belief in the efficacy of antiretroviral therapy, which require continual counseling by providers.
Studies have shown improved outcomes among HIV-infected patients treated by experienced physicians [24,27,28]. In our study, participants who reported a usual source of care and consistently reported illicit drug use on two semi-annual visits were less likely to report HAART than those without a concurrent history of illicit drug use. In this case, clinician awareness of patient drug use may lead to more conservative management.
The study period covered in this analysis is from 1 January 1996 to 30 June 1999. Since then, expert opinion regarding the optimum time to start therapy has changed somewhat. Today, many providers feel that CD4+ cell count may be a better indicator of the need for therapy than viral load, and patients could wait longer before starting. Whereas many suggest waiting until CD4+ cell count has dropped to 350 × 106 cells/l, some will even suggest waiting for a nadir of 200 × 106 cells/l. The rationale for delayed therapy is based on the following considerations: (1) early therapy may lead to premature drug resistance, due in part to non-adherence; (2) the long-term toxicity of antiretroviral therapy may prevent patients from taking these regimens continuously for decades; (3) most patients who begin therapy with CD4+ cells counts above 200 × 106 cells/l have excellent virologic and immunologic responses to therapy. Thus, many IDUs in whom therapy was deferred (regardless of the reason) will find themselves in the enviable position of being treatment-naive, whereas all too many of those who were treated according to the guidelines at that time have already run out of treatment options or are experiencing difficult toxicity.
Integrating antiretroviral therapy regimens and drug abuse treatment may offer one avenue for the effective management of these two related medical problems. Current treatment guidelines call for addressing issues of substance abuse as an integral component of HIV management. Participants who reported being on methadone maintenance may be viewed by their providers as being stable and more likely to be adherent to HAART regimens, although in our analysis, this may not be the case for women. Although the confidence interval around our point estimate includes 1.0 for women who reported methadone use, the relative hazard was substantially lower than 1.0, suggesting that this subgroup of women are actually less likely to initiate HAART. It is believed that methadone treatment may encourage health-seeking behaviors, including antiretroviral therapy utilization, among drug users who had previously not sought care. Given our observation, this may be truer for men than women, perhaps reflecting a woman's reluctance to seek medical care to avoid disclosing their drug use. Alternatively, this may reflect a tendency of providers to be less willing to prescribe HAART to female IDUs. As providers become more experienced in the HIV treatment of drug users, there appears to be increased willingness to prescribe more complex and aggressive therapies for this population.
References
1. Hammer SM, Squires KE, Hughes MD.
et al. A controlled trial of two nucleoside analogues plus indinavir in persons with human immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. N Engl J Med 1997, 337: 725–733.
2. Gulick RM, Mellors JW, Havlir D.
et al. Treatment with indinavir, zidovudine, and lamivudine in adults with human immunodeficiency virus infection and prior antiretroviral therapy. N Engl J Med 1997, 337: 734–739.
3. Gulick RM, Mellors JW, Havlir D.
et al. Simultaneous vs. sequential initiation of therapy with indinavir, zidovudine, and lamivudine for HIV-1 infection. JAMA 1998, 280: 35–41.
4. Detels R, Muñoz 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.
5. 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.
6. Celentano DD, Vlahov D, Cohn S, Shadle VM, Obasanjo O, Moore RD. Self-reported antiretroviral therapy in injection drug users. JAMA 1999, 280: 544–546.
7. Carpenter CCJ, Fischl MA, Hammer SM.
et al. Antiretroviral therapy for HIV infection in 1997: updated recommendations of the International AIDS Society- USA panel. JAMA 1997, 277: 1962–1969.
8. Strathdee SA, Palepu A, Corneliesse PG.
et al. Barriers to use of free antiretroviral therapy in injection drug users. JAMA 1998, 280: 547–549.
9. Vlahov D, Anthony JC, Muñoz A.
et al. The ALIVE study, a longitudinal study of HIV-1 infection in intravenous drug users: description of methods and characteristics of participants. NIDA Res Monogr 1991, 109: 75–100.
10. Carpenter CCJ, Fischl MA, Hammer SM.
et al.
Antiretroviral therapy for HIV infection in 1996.Recommendations of an international panel. International AIDS Society-USA. JAMA 1996, 276: 146–154.
11. Carpenter CCJ, Fischl MA, Hammer SM.
et al. Antiretroviral therapy for HIV infection in 1998: updated recommendations of the International AIDS Society-USA Panel. JAMA 1998, 280: 78–86.
12. Lederberger B, Egger M, Opravil M.
et al. Clinical progression and virological failure on highly active antiretroviral therapy in HIV-1 patients: a prospective cohort study. Lancet 1999, 353: 863–868.
13. Gallant JE. Strategies for long-term success in the treatment of HIV infection. JAMA 2000, 283: 1329–1334.
14. Bozette SA, Berry SH, Duan N.
et al.
The care of HIV-infected adults in the United States. N Engl J Med 1998, 339: 1897–1904.
15. Dorucci M, Pezzotti P, Phillips AN, Alliegro MB, Rezza G. Antiretroviral treatment and progression to AIDS in HIV seroconverters from different risk groups. AIDS 1997, 11: 461–467.
16. Stein MD, Piette J, Mor V.
et al. Differences in access to zidovudine (AZT) among symptomatic HIV-infected persons. J Gen Intern Med 1991, 6: 35–40.
17. Roseberg PS, Gail MH, Schrager LK.
et al. National AIDS incidence trends and the extent of zidovudine therapy in selected demographic and transmission groups. J Acquir Immune Defic Syndr Hum Retrovirol 1991, 4: 392–401.
18. Thompson C, Salvato PD.
Intravenous drug use and the immune system.XII International Conference on AIDS. Geneva, June–July 1998 [abstract 32437].
19. Solomon L, Frank R, Vlahov D, Astemborski J. Utilization of health services in a cohort of intravenous drug users with known HIV-1 serostatus. Am J Public Health 1991, 81: 1285–1289.
20. Bangsberg D, Tulsky JP, Hecht FM, Moss AR. Protease inhibitors in the homeless. JAMA 1997, 278: 63–65.
21. Imrie A, Beveridge A, Genn W, Vizzard J, Cooper DA, and The Sydney Primary HIV Infection Study Group. Transmission of human immunodeficiency virus type 1 resistant to nevirapine and zidovudine. J Infect Dis 1997, 175: 1502–1506.
22. Hecht FM, Grant RM, Petropoulos CJ.
et al. Sexual transmission of an HIV-1 variant resistant to multiple reverse-transcriptase and protease inhibitors. N Engl J Med 1998, 339: 307–311.
23. 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 Syndr22:369–378
24. Dykeman MC, Sternberg C, Jasek J.
et al. A model for the delivery of care for HIV-positive clients. AIDS Patient Care STD 1996, 10: 240–245.
25. Kitahata MM, Koepsell TD, Deyo RA, Maxwell CL, Dodge WT, Wagner EH. Physicians’ experience with the acquired immunodeficiency syndrome as a factor in patients’ survival. N Engl J Med 1996, 334: 701–706.
26. Bangsberg D, Moss AR.
Protease inhibitors (PI) in the HIV+ homeless and marginally housed (H/M): good adherence but rarely prescribed.XII International Conference on AIDS. Geneva, June–July 1998[abstract].
27. Brosgart CL, Mitchell TF, Coleman RL, Dyner T, Stephenson KE, Abrams DI. Clinical experience and choice of drug therapy for human immunodeficiency virus disease. Clin Infect Dis 1996, 28: 14–25.
28. Paauw DS, Wenrich MD, Curtis JR, Carline JD, Ramsey PG. Ability of primary care physicians to recognize physical findings associated with HIV infection. JAMA 1995, 274: 1380–1382.
