Portal de Periódicos da CAPES

Effectiveness of methadone treatment in reducing HIV risk behavior and HIV seroconversion among injecting drug users

1999; Lippincott Williams & Wilkins; Volume: 13; Issue: 14 Linguagem: Inglês

10.1097/00002030-199910010-00002

1473-5571

David R. Gibson, Neil M. Flynn, John J. McCarthy,

Cannabis and Cannabinoid Research

Introduction Drug addiction is a major risk factor for HIV infection in the United States and at least 80 other countries[1]. In the United States, injecting drug use now accounts for more than one third of new AIDS cases (nearly double the proportion in 1994), and for as many as three quarters of new HIV infections[2]. Behaviors associated with drug use that are specific for HIV transmission include shared use of injection equipment and other paraphernalia and unprotected vaginal or anal sex with injecting drug users (IDU). Interventions that can reduce the prevalence of these behaviors are, therefore, critical components of a comprehensive AIDS prevention policy. Strategies for HIV risk reduction include needle exchange, street-based outreach by peer educators, educational interventions, counseling and testing programs, community-level interventions to change norms regarding safer injection and safer sex, and treatment for drug abuse. Treatment programs, to the degree they are effective, are thought to reduce the risk of transmission by increasing abstinence from drug use and/or injection[3,4]. The purpose of this paper is to provide a comprehensive review of published evidence of the effectiveness of methadone treatment in preventing HIV risk behavior and HIV seroconversion among IDU. Methadone treatment is by far the most widely available treatment for addiction to heroin and other opiates. While three _previous papers reviewed evidence concerning the effectiveness of methadone in preventing HIV infection, one [5] drew only on studies conducted prior to 1991, and two others [6,7] did not attempt a comprehensive review. We have identified and will describe 38 studies published since 1987 that examined effectiveness. Identification of studies The scientific literature in the English language was systematically searched using Medline (National Library of Medicine, Bethesda, MD, USA) and PsychInfo (American Psychological Association, Washington, DC, USA) databases for the period 1985-1998. This was supplemented by reviews of bibliographies in articles that came to our attention as well as inquiries among colleagues. We included all studies of the effectiveness of methadone in reducing injection drug use, HIV risk behavior, and HIV seroconversion, regardless of design. While injection drug use is not, per se, a measure of risk behavior, several studies suggest that it is good proxy; for example, two studies [8,9] found relationships between methadone treatment and risk behaviors were attenuated when injection frequency was controlled. The HIV risk behaviors we examined included sexual risk behavior where such outcomes were reported. Regarding design, given that substantial evidence exists that methadone maintenance reduces drug use and criminality, promotes employment, and otherwise rehabilitates drug users, randomized trials of methadone are generally not possible for practical or ethical reasons. A few randomized studies are included in which investigators were able to deal with the practical and ethical issues. Studies with quasi-experimental designs that substituted statistical and other design controls in lieu of randomization are also included. These include prospective and longitudinal cohort studies, case-control studies, and observational studies with statistical controls for potentially confounding variables. Several studies without rigorous controls are also included because, although they do not provide definitive evidence of the effectiveness of methadone treatment, they do offer intriguing clues to an overall understanding of the pattern of results. Two of the authors (D.R.G and N.M.F.) independently reviewed each of the studies and abstracted essential information concerning design, procedures, results, statistical controls, and study limitations, including checks on the success of randomization, participation and attrition bias, and sample representativeness. They then reconciled their ratings of the studies, arriving at consensus description of each study. A formal meta-analysis was not attempted since outcome measures used in different studies were seldom identical. Of the 38 studies reviewed, five were randomized clinical trials, 11 were prospective or longitudinal cohort studies, two were case-control studies, and 20 were observational studies. Each of these sets of studies are discussed in turn. Randomized trials Although it has been frequently observed that randomized trials of methadone treatments are not ethically or practically possible, they have been conducted in circumstances in which the comparison condition is roughly equivalent to the standard of care, or where the relative effectiveness of the treatments being compared is not known (Table 1). In one instance[14], subjects were randomized to immediate methadone maintenance treatment or to a 4-month wait-list control group.Table 1: Randomized trials.Grella et al. randomized subjects to standard methadone maintenance or to an enhanced condition that included an array of added services such as case management, access to special groups, and on-site psychiatric services [10]; no differences in cocaine use were observed at follow-up interviews between the standard and enhanced conditions. Iguchi et al. [11] obtained similar results after randomizing subjects to 21 versus 90 days of treatment; individuals assigned to the two groups did not differ in outcomes, although at follow-up subjects in both groups reported substantial reductions in both injection-related and sexual risk behavior. By contrast, McLellan et al. [12] and Vanichseni et al. [13] found outcome differences between experimental and comparison treatments. In the McLellan study, while methadone maintenance without counseling or other psychosocial services was associated with reductions in opiate use, 69% of the subjects assigned to it had to be 'protectively transferred' from the trial because of 'unremitting use' of opiates or cocaine and/or medical/psychiatric emergencies, compared with 41% in standard methadone maintenance (with counseling) and 19% in enhanced methadone services (with counseling, psychiatric, and employment services). With regard to abstinence from heroin and cocaine, the group assigned to standard methadone services did better than that assigned to methadone alone, and the enhanced group did better than the standard group. In the Vanichseni study, subjects assigned to methadone maintenance were more likely than those assigned to methadone detoxification to abstain from using heroin and complete 45 days of treatment. Finally, Yancovitz et al. [14] randomized applicants to methadone maintenance to immediate entry to limited methadone services or to a wait-list control group; at a 1-month follow-up, the experimental group showed a 62 to 29% decrease from baseline to follow-up in urinalyses positive for heroin, while there was no change in the comparison group. No significant changes in cocaine use were reported for either group. At a 16-month follow-up, a greater proportion of the experimental group remained in treatment (72% versus 56% for the control group). Prospective and longitudinal cohort studies Of the prospective and longitudinal cohort studies that examined associations between methadone maintenance and HIV-related behaviors (Table 2), the study by Ball et al. [15,16] is probably the best known. Analyzing data gathered at six clinics in three East Coast cities in the United States, Ball found that 71% of patients who remained in treatment for 1 year or more ceased injection drug use; by contrast, the vast majority who dropped out of treatment relapsed to drug use. The treatment programs where the data were collected varied considerably in their effectiveness in reducing injection drug use: four of the programs were effective in stopping injection drug use in 75-90% of their patients by 1 year, while only about 45% of patients in the other two clinics stopped injecting. The investigators were fortunate in being able to investigate which aspects of the programs appeared to be responsible for these differences, holding the pretreatment characteristics of patients constant. In regression analyses using factor scores derived from principal component analyses that reduced 25 program variables to five program domains, the characteristics of the programs most responsible for decreasing injection drug use appeared to be those related to an effective program director and a strong rehabilitation and maintenance policy, and the adequacy of medical services. To the investigators' surprise, these program variables explained substantially more variance in injection drug use and other treatment outcomes that patient variables such as race, younger age of onset of drug use, history of cocaine use, and criminal behavior prior to admission to treatment. However, the use of only five program factors to differentiate six clinics may have exaggerated their importance. Interestingly, patients' dose levels appeared to have no relationship to treatment outcome.Table 2: Prospective cohort studies.In another study conducted at 17 clinics in eight cities in the United States, Condelli et al. [17] examined the effects of length of exposure to methadone maintenance treatment; those with continuous exposure (mean 725 days) were much less likely to use heroin at least weekly than those with short-term (mean 31 days) or long-term (mean 233 days) exposure: 17% of the continuous exposure group reported weekly use versus 39% and 40%, respectively, for those in the short- and long-term groups. Retention was highest among patients who attended programs that informed them of their methadone dose levels, were easy to access, and provided them with high-quality social services. This study did not attempt to control for baseline patient variables that may have accounted for differences in outcomes. In a study that was both prospective and retrospective, Greenfield et al. [18] tracked injection drug use and needle sharing in a mixed (in- and out-of-treatment) sample over a 12-month period; at each of six time points, the in-treatment group reported less injection and less needle sharing than out-of-treatment participants. The researchers did not attempt to control for potential confounders that might explain these relationships. In a study conducted in Verona, Italy, Martin et al. [19] followed members of a cohort from initial contact at entry to drug treatment for 6 months. During this time, syringe sharing in the cohort decreased from 35 to 12%; participants who were under continuous methadone treatment were less likely to share needles (18%) than those who interrupted treatment (39%). Continuous treatment was also associated with decreases in risky sex, number of sexual partners, and increases in condom use. Baseline differences between participants in continuous versus interrupted treatment were not controlled. Metzger et al. [20] followed a cohort of in- and out-of-treatment IDU prospectively for 18 months: in-treatment participants were a random sample of the clinic where the study was conducted; out-of-treatment participants were recruited through the social contacts of the in-treatment group and in many ways were similar to them. There were striking differences between the two groups by the time of the 18-month follow-up: 3.5% of the in-treatment group seroconverted for HIV compared with 22% among out-of-treatment drug users. The authors commented that while incidence in the untreated group seemed high for such a relatively short period, similar incidence rates had been reported from other East Coast cities in the United States. On the West Coast of the United States, Moss and colleagues [21] followed a 'hidden' cohort of drug users entering methadone treatment. While the members of the cohort were anonymous, they provided 'unique identifiers' that enabled the investigators to identify participants interviewed at multiple time points and to track their risk behavior and HIV seroconversion over time. They found that participants who had been in methadone treatment for less that 12 months were two to three times more likely to have seroconverted before being reached for follow-up interviews 2-5 years after baseline. In New York, Neaigus et al. [22] and Shore et al. [23] also found protective effects of methadone treatment. In the Shore study, increases in injecting frequency at follow-up was associated with interrupted treatment. In Philadelphia (United States), Watkins et al. [24] found out-of-treatment drug users to have more sexual partners and to exchange sex for drugs or money more often that those who were in treatment. Baseline differences between the two groups were not examined. In Seattle (United States) Wells et al. [25] found that men retained in treatment were much less likely to share injection equipment, and that women had fewer IDU partners and less sex for drugs or money, controlling for baseline risk scores, injection frequency, and age. In a small study in New Haven (United States), Williams et al. [26] found substantially less seroconversion (2%) in continuously treated patients that in patients who had interrupted treatment (19%), although this finding was only a trend (P_=_0.1) when differences in length of time to follow-up (29 and 53 months, respectively) were controlled. In a somewhat larger sample with a more appropriate (one-tailed) statistical test, these findings might well have reached significance. The Williams group also found that IDU in continuous treatment had fewer sexual partners. Zangerle et al. [25] found no seroconversion among in-treatment IDU a mean 18-months later, compared with 17% among untreated users. Case-control studies Two case control studies were identified, one from Italy the other from the Netherlands (Table 3). Serpelloni et al. [28] studied 40 seroconverters in a prospective cohort along with 40 seronegatives matched with cases by sex, age, duration of drug use, year of initiation into drug use, first entrance into drug treatment, and time to follow-up; the mean time to follow-up was 6 months, with a range of 3 to 18 months. For those not in treatment, the risk associated with every 3 months without treatment was 70%. The other study, conducted by van Ameijden et al. [29], examined associations between high-risk behavior and 'daily use' of 'low threshold' methadone posts, which meant that they were more permissive in terms of their requirements of clients and provided clients with lower doses of methadone. Use of the methadone posts appeared to be unrelated to the practice of borrowing or lending syringes.Table 3: Case-control studies.Observational studies The majority of the studies reviewed here were observational. While it is customary to give such studies less weight, a well-controlled observational study can be as rigorous as studies with quasi-experimental designs, since both rely on statistical adjustment for confounders. Even in a randomized trial, covariates are commonly used to correct for the failure of randomization to produce fully equivalent experimental groups. In an early study, Abdul-Quader et al. [30] examined differences between patients entering a New York City methadone maintenance program prior to 1982, between 1982 and 1984, and 1985 and thereafter (Table 4). Seroprevalence for HIV was higher for each succeeding cohort (35, 54, and 57%, respectively), and being in treatment was associated with less injection and less injection in shooting galleries, among other factors. There were no differences between the cohorts in high-risk sexual behavior. No attempt was made to control for potential confounders, such as age. A study from Baltimore (United States) conducted by Alcabes et al. [31] found no differences in HIV seropositivity between treated and untreated IDU. A well-controlled study conducted by Baker et al. [32] in Sydney, Australia also compared treated and untreated IDU, and found treated users injected considerably less frequently than previously or never treated IDU (partial regression coefficient of -0.25), although treatment did not significantly affect borrowing or bleaching of syringes. Potential confounders that were examined or controlled are shown in Table 4. The Baker group found no effect of current treatment on sexual risk. In a study conducted in Uppsala, Sweden, Blix and Grondbladh [33] found results similar to those of Zangerle et al. [27]: among patients entering methadone maintenance treatment before 1985, none had seroconverted to HIV by 1990; this contrasted with a seropositivity of up to 59% among patients admitted in each of the six succeeding years (1985-1990). No attempt was made to examine initial differences between drug users in the early versus later cohorts.Table 4: Observational studies.Contemporaneously, Brown and colleagues in one study [34] found an HIV seroprevalence rate of 61% for those in treatment a year or less versus 50% for 2 or more years; in another study [35] the breakdown was 66% and 56%, respectively, for patients in treatment less than versus more than a year. Neither study attempted to examined initial differences between patients in shorter versus longer term treatment. A well-controlled study from Sydney, Australia by Caplehorn and Ross [8] compared in-treatment and out-of-treatment IDU. Adjusting for the variables shown in Table 4, they found 45% decrease in borrowing of syringes associated with being in treatment. In contrast with the Brown findings, Chaisson et al. [36] in San Francisco (United States) found no differences in HIV seroprevalence between IDU in treatment less than versus more than a year; the sample, however, included many patients entering short-term 'detox.' and many with problematic cocaine use. In another study from New York, Kang and de Leon [37] compared patients just beginning treatment with patients in treatment for at least 3 years. The rates of injection drug use among those beginning treatment was six times that in those in long-term treatment, and in multiple regression analyses length of treatment was inversely related to injection frequency (partial regression coefficient = -0.53). The controls for depression, age at first arrest, and employment status may not have been adequate. In northwest England, Klee et al. [38] found that treatment had a protective effect mainly for IDU who were 25 years or older or who had been using for 5 years or more; among these users there was much less borrowing or lending of syringes if they were in drug treatment. Age was the only control variable examined. Two studies by Longshore and colleagues examined injection-related and sexual risk behavior among in- and out-of-treatment IDU. In the first study [9] 63% of in-treatment subjects reported borrowing syringes versus 79% for out-of-treatment users; this relationship persisted after injection frequency and drug user background characteristics were controlled. In the second study [39] treatment was associated with reduced numbers of sexual partners, although the magnitude of this relationship was small. Meandzia et al. [40] studied treated and untreated heroin, cocaine, and 'speedball' (heroin/cocaine mixture) users in New Haven (United States). Treated IDU reported fewer injections overall in the past 30 days and reduced cocaine and speedball injection as well. Sex and race were the only variables controlled. In New York City, Novick and colleagues conducted two investigations of HIV infection in treated and untreated IDU. In the first study[41], as in several other studies already described, HIV-positive patients had been in drug treatment for a much shorter period (average 3.5 years) compared with HIV-negative patients, who had been in treatment for an average of 6.7 years. In the second study[42], none of 58 long-term (12 years or more in treatment), socially rehabilitated patients tested positive for HIV, closely paralleling the findings of Blix and Grondbladh [33] and Zangerle et al.[27]. In neither study were controls employed. In another New York study, Schoenbaum et al. [43] found similar results: among patients admitted between 1978 and 1985, there was a 24% greater odds of seroconverting for each additional year of continued injection. Initial differences between members of the yearly cohorts were not examined. In yet another New York study, Selwyn and colleagues [44] compared the behavior of IDU in methadone maintenance treatment with that of the inmates of a correctional facility, reasoning that the second group would closely approximate the world of the street addict in terms of HIV risk; the study found the adjusted odds of needle sharing among detention inmates to be three and a half times that of methadone clients. Two studies conducted by Stark and colleagues in Berlin also found evidence of a protective effect of methadone. In the first study[45], out-of-treatment IDU reported significantly more borrowing of syringes that those in treatment; in the second study[46], investigators found an approximately twofold decrease in likelihood of borrowing syringes even when controlling for age, schooling, and duration of injection drug use. The first study did not apply controls. Finally, in a study by the van Ameijden group [47] in Amsterdam (not the study described above [29]), use of 'low threshold' methadone posts was found to have no_relationship with the borrowing, lending, or reusing of syringes. Discussion In summary, there is substantial evidence that methadone maintenance treatment, the most widely available treatment for opiate addiction, is effective in preventing HIV disease among IDU. Thirty-four of the 38 studies we reviewed contained evidence that methadone maintenance reduces HIV risk behavior and/or HIV seroconversion. Seventeen of these reported statistically significant differences between in-treatment and out-of-treatment IDU; nine showed that the benefits of treatment increase with treatment duration, and five showed that continuous as opposed to interrupted treatment was associated with decreased risk of infection with HIV. In addition, four studies provided evidence that the quality and intensity of the treatment provided can significantly to substantially affect the prevalence of injection drug use and high-risk injection-related behavior in treated populations. Of the four studies that did not find benefits of being in treatment, two studies [10,36] used the injection of cocaine as well as heroin as an outcome, and two studies[29,47], both from the same group, showed no benefits associated with attendance at a 'low threshold' methadone dispensary. With regard to the findings in which cocaine was implicated, it is well recognized that while methadone is effective in blocking withdrawal symptoms and craving for heroin (a central nervous system depressant) it confers no similar benefit for people who are addicted to stimulants such as cocaine (there is evidence, however, that IDU who inject a 'speedball' mixture of heroin and cocaine profit as much from methadone treatment as IDU who only inject heroin [48]). In the early to mid-1980s, cocaine was the drug most frequently abused by methadone maintenance patients in many cities in the United States [48]; it has since been eclipsed by crack cocaine and other drugs. Powdered cocaine, however, remains a drug of choice in other countries such as Italy and Brazil. Unfortunately, no effective pharmacotherapy for cocaine exists. Nevertheless, it is important that methadone treatment providers attempt to address cocaine abuse in developing psychosocial adjuncts to methadone; cocaine injection may be the single most important risk factor for infectious blood-borne diseases in patients receiving methadone maintenance. The study by Ball and colleagues [15,16] showed that the quality of methadone treatment can diminish risks associated with cocaine use; it found that patients in methadone programs with a strong rehabilitation orientation and adequate medical services were less likely to use cocaine. As we have seen, the quality and intensity of treatment and dose may be critical to the effectiveness of methadone maintenance in preventing HIV and other blood-borne diseases. The Ball study[15,16], for example, found large differences between the six clinics studied in the proportion of their patients injecting drugs, differences that were attributed to an effective program director and strong orientation to rehabilitation and medical services. Among the studies we have reviewed, four other studies provided evidence that quality or intensity of treatment can affect its effectiveness in preventing HIV risk behavior and HIV seroincidence. Four studies provided mixed evidence concerning the effects of dose levels. However, dose levels were not reported, and the lack of findings with regard to dose could have been a result of restricted variation in doses prescribed at some clinics. Numerous studies (for example [49]) have found dose level to be among the more important predictors of treatment outcome. Eight studies examined the impact of methadone treatment on drug users' sexual behaviors. Six of the eight found effects, most of which had to do with reduced numbers of sexual partners and the prevalence of exchanging sex for drugs and money. One study suggested that these effects were specific to females. Only one study found impact of methadone on use of _condoms or prevalence of high-risk sex. The studies reviewed here underscore the long-_recognized need for increasing access to drug treatment. In the United States, the National Academy of Sciences has urged that 'treatment on demand' be made available to IDU [50] and the National Institute on Drug Abuse has identified drug treatment as its principal strategy for preventing HIV and AIDS[51]. The focus of this review has been on methadone's effectiveness in preventing HIV, but it has also been shown to be effective in reducing, even eliminating heroin use [15,52] and reducing criminal behavior[15,52-54]. Methadone maintenance, which costs an average of US$4000-5000 per patient per year reduces criminal behavior associated with illegal drug use, _promotes health, and improves social productivity, all of which serve to reduce the societal costs of drug addiction[55,56]. Cost-benefit analyses indicates _savings of US$4-5 in health costs for every dollar spent on methadone maintenance treatment[55-58]. These estimates do not include the costs of treating HIV infections. A cost-effectiveness analysis of methadone maintenance treatment under epidemic conditions present in the northeastern cities of the United States in 1990 found five to seven HIV infections are averted for every 100 HIV-negative patients in methadone maintenance treatment for a year[59]. This is based on a two thirds reduction by patients in the frequency of needle sharing, estimated from literature review. The annual cost of treating methadone patients in 1990 was about US$3000 annually, or about US$50_000 over an average lifetime, substantially less than the US$190_000 it costs a to treat a person for HIV disease[59]. In closing, it is worth noting some of the limitations of the studies we have reviewed. Many of the studies, including both prospective cohort and observational studies, did not adequately control statistically for differences between groups being compared. Consequently, findings of differences attributed to treatment could be caused by unmeasured or unanalyzed third variables. There seems to be some consensus that the important control variables include age, sex, age at first injection, length of injection career, and history of incarceration[60]. However, in several studies at least (for example [32]) these variables did not appear to be related to outcomes. Moreover, there is evidence (for example [61,62]) that drug users who enter drug _treatment may be more poorly functioning at entry to treatment than drug users who are not in treatment, and comparisons of drug users who are in-treatment versus out-of-treatment may underestimate the differences between the two groups. Further research needs to be done to sort out these important issues. In the meantime, greater attention needs to be paid to adjustment for confounding in quasi-experimental comparisons of in-treatment and out-of-treatment drug users.