Key Issues In Methadone Maintenance Treatment. Chapter 3 "Treatment Effectiveness II: Observational Studies"

I. Observational Studies of Treatment Effectiveness

In the absence of a sufficient number of randomised controlled trials to draw strong inferences about treatment effectiveness, the assessment of the effectiveness of methadone maintenance depends upon the results of observational studies of patient outcome. In such studies the outcomes of treatment are observed among patients who have assigned themselves to different forms of treatment rather than being randomly assigned by a treatment researcher. Observational studies of treatment effectiveness comprise two major types. First, there are comparative studies in which the outcomes are compared in persons who selected themselves into different treatments (for example, methadone maintenance, therapeutic communities, and drug-free counseling). Secondly, there are pre-post evaluations of treatment in which a group of people entering a single type of treatment are assessed at intake and at some time after treatment, with the effect of treatment being assessed by changes in outcomes such as drug use between pre-treatment and post-treatment.

II. Controlled Comparative Studies

T he major problem with all observational studies is whether the people receiving different forms of treatment were comparable prior to treatment. As a consequence, it is difficult to rule out the possibility that apparent differences in treatment outcome arise because of differences in patient prognosis prior to treatment. The strategy of quasi-experimentation (Cook & Campbell, 1979) provides a way of making cautious inferences about treatment effectiveness from observational studies. This involves three processes. First, plausible rival hypotheses are generated which may explain any differences between treatments in outcome. Of these the most plausible is that the treatments differed in the number of patients who had a good or a poor prognosis regardless of treatment. Secondly, patients are measured on variables that may predict a better or worse outcome, such as prior history of drug use, degree of criminal involvement, and severity of drug dependence. Thirdly, statistical methods of control (e.g. stratification and covariate adjustment) are used to decide whether these rival hypotheses explain the differences in outcomes between treatments. That is, do the differences in treatment outcome persist when account is taken of pre-existing patient differences? If the differences in outcome persist after statistical adjustment, one can be more confident that there is a treatment effect.

The inclusion of evidence from the quasi-experimental studies in a review of the effectiveness of any form of treatment is unavoidable, even when there is abundant evidence of effectiveness from randomised controlled trials. In order to conclude from the evidence of the randomised controlled trials that the treatment under investigation is effective, a quasi-experimental comparison is required to justify the inference that the people who have been included in the randomised controlled trials are comparable in all relevant respects to the patients who receive such treatments in clinical practice. Moreover, a refusal to accept anything other than the evidence from randomised controlled trials (e.g. McMaster Department of Clinical Epidemiology, 1981) would have the unintended effect of endorsing minimally evaluated treatments for opioid dependence. Given the serious individual and societal consequences of opioid drug dependence, the community will insist upon offering some form of treatment, and in the absence of evaluation by randomised controlled trials, the proponents of all therapeutic approaches have an equal claim for public support.

III. Comparative Studies of Methadone Maintenance Treatment

III. a. Bale, Van Stone, Kuldau, Engelsing, Elashoff and
Zarcone (1980)

Bale and his colleagues (1980) planned to conduct a randomised controlled trial in which the outcomes of methadone maintenance and therapeutic communities would be compared with detoxification; however, ethical and practical problems prevented random assignment of subjects to treatment. The result was a study that compared the outcomes of patients who selected methadone maintenance treatment , therapeutic communities and detoxification at 12 months post-treatment.

There were several distinctive features of this study. First, subjects who entered methadone and therapeutic communities were very nearly comparable, as indicated by a comprehensive pretreatment assessment. The main reasons for this were that subjects were recruited from a common pool of potential patients (opioid-addicted veterans in the Veterans' Administration treatment system), and that staff from each of the treatment programs competed for the patients on conditions of near equality. All program staff had access to the potential subjects while they were in hospital, and subjects were encouraged to spend three weeks in the program to which they were assigned before they changed to the program of their choice.

Secondly, a number of different programs were represented within each treatment modality. There were three therapeutic communities with a variety of orientations, and two low-dose methadone maintenance programs that were compared with detoxification only, which was provided in the main treatment centre from which all subjects for the study were recruited. Thirdly, 93% of patients were followed up at six and 12 months. The results of treatment were therefore available for almost all who entered treatment, regardless of how long they stayed, and not just for the treatment successes. Fourthly, the 12-month outcome was assessed by an independent interviewer who was unaware of which treatment the subject had received, and efforts were made to validate self-reports of drug use and criminal activity.

The major results of relevance to this review are those that compared the outcomes of methadone maintenance with those of simple detoxification. The comparison between methadone maintenance treatment and therapeutic communities -- which failed to find any difference between the two in average effectiveness -- will not be discussed. The results indicated that the two methadone maintenance programs produced better outcomes than did detoxification when measured by reductions in opioid drug use during the past month, and the number of convictions recorded during the past year. Moreover, the differences in outcome between methadone maintenance and detoxification persisted after adjustment for 10 patient characteristics that had been shown to predict outcome.

A number of issues need to be considered in interpreting this study. First, the methadone maintenance programs provided in this study differed from that recommended by Dole and Nyswander. Both programs prescribed low doses of methadone and encouraged their patients to become abstinent from all opioids, including methadone. Secondly, the combination of a small sample size for methadone and the use of crude dichotomous measures of outcome reduced statistical power, and hence the sensitivity of the study to detect differences in outcome between treatment modalities. Thirdly, those who received detoxification only were entirely self-selected in that they consisted of people who had declined any other form of treatment.

Even allowing for these qualifications, the Bale et al. study provided evidence for the effectiveness of methadone maintenance which supported the results obtained in the three randomised controlled trials. The methadone maintenance programs produced better outcomes in terms of drug use and criminality than detoxification, and this difference in treatment outcome was not explained by the covariates that Bale et al. measured. In terms of the quasi-experimental strategy outlined above, this study provides qualified support for the conclusion that the differences in outcome between methadone maintenance and detoxification were caused by the difference in treatment.

III. b. Anglin and Associates

Anglin and his colleagues conducted a series of studies in California to evaluate the impact of treatment on the behaviour of patients in a number of methadone clinics (Anglin & McGlothlin, 1984). In each study, retrospective data were collected using a time line technique in which the interviewer went over a detailed chart marked with the subject's criminal and treatment history. After establishing the date of first opioid use and the date of first dependence, the interviewer and subject filled in details of opioid use, criminal activity, and other relevant outcome measures up until the time of interview. The authors claim that this technique yielded reasonably accurate, retrospective information.

III. c. Methadone Versus No Methadone Treatment

The authors originally set out to study a group of opioid-dependent people who were committed for seven years to compulsory inpatient treatment as an alternative to imprisonment during 1962- 64 as part of the California Civil Addict Program (CAP) (Anglin & McGlothin, 1984). Of the 439 subjects in this early study, 118 later entered methadone maintenance treatment when it was commenced in California in the early 1970s. By this time nearly all of the subjects in the CAP had finished their first commitment period (Anglin, 1988). Among the subjects who did not enter methadone maintenance treatment, two groups were defined on the basis of their opioid use post- CAP: a group of inactive recovered heroin users who gave up their addiction during CAP treatment; and a group of active users who relapsed to daily heroin use. For the purposes of evaluating the effectiveness of methadone maintenance treatment, the comparison group for the methadone patients was the active heroin-using group.

Figure 1 is a graph of retrospective time series data that shows the percentage of non-incarcerated time that subjects in each of the three groups (recovered, active heroin users and methadone patients) were involved in daily heroin use. The two dotted vertical lines at 4 years and 8 years mark out the time spent in the CAP. The year 0 on the abscissa represents the time of admission to methadone maintenance treatment for the methadone group. For the recovered and active groups, the median admission date for the methadone group was used to establish the comparative reference point (Anglin, 1988).

The effect of entry to methadone maintenance treatment for the methadone group is apparent when compared with the active group. An increase in heroin use is shown for both groups just prior to admission, which was probably due to a dramatic increase in the supply of heroin in the USA at that time. Entry into methadone maintenance brought about a marked reduction in heroin use which endured throughout the three-year follow-up period. Although not represented here, a similar pattern of results was found for criminal activity (Anglin & McGlothlin, 1984).

III. d. Involuntary Termination Of Methadone Treatment
Versus Methadone Maintenance

In 1976 the only methadone program operating in Bakersfield, California was closed. The nearest clinic was 70 miles away in Tulare. The closing of the Bakersfield program provided McGlothlin and Anglin (1981b) with the conditions for a natural experiment in which the Bakersfield patients could be compared to a group from nearby Tulare who were not involuntarily discharged from treatment. Follow-up interviews were conducted two years after the closure of the Bakersfield clinic. Overall, the Tulare group spent 73% of non-incarcerated time during the follow-up period in methadone maintenance compared with 8% for the Bakersfield group.

There were substantial improvements in heroin use and criminal activity reported by both groups when the pre-treatment and treatment periods were compared. In terms of the effect of the closure of the Bakersfield program, 60% of the men and 56% of the women became dependent again, excluding the eight patients who managed to transfer to another clinic. At the time of interview, an unexpected urine sample was taken, and the rate of morphine-positive urines was higher for the Bakersfield group. Of the 14 subjects who tested positive in the Tulare group only two were still on methadone. The Bakersfield group also had about twice the percentage of individuals arrested during the follow-up period when compared with the Tulare group. The overall outcome for the Bakersfield group was poor: 54% became dependent on opioids again, 73% were arrested, 61% were imprisoned for more than 30 days, and two died from drug overdoses.

Anglin, Speckart, Booth and Ryan (1989) conducted a similar study after the closure of the San Diego public methadone program, using a similar comparison group. However, the study found few differences between people in the involuntarily terminated and continuing methadone maintenance programs, largely because a substantial proportion of the San Diego group transferred to private methadone programs. Even so, it should be noted that the patients who did not transfer had poorer outcomes than those who did.

Overall, the studies reviewed here by Anglin and his colleagues include two reasonably powerful tests of methadone maintenance treatment: a comparison of the effects of the introduction of methadone maintenance treatment to a group of opioid-dependent patients with a no-treatment comparison group; and a unique study of the effects of involuntary cessation of methadone maintenance treatment with a comparison group of patients who remained in methadone maintenance treatment. Although the results of both studies showed the expected effects of the introduction and removal of treatment, our confidence that these results were due solely to the effect of treatment is somewhat reduced by the reliance on long-term retrospective self-reported drug use.

III. e. The Drug Abuse Reporting Program

The Drug Abuse Reporting Program (DARP) was a large-scale treatment outcome study that collected data on approximately 44 000 clients who applied for treatment at 52 drug treatment agencies in the USA and Puerto Rico during 1969 to 1973 (Simpson & Sells, 1982). The treatment modalities represented by the participating agencies were methadone maintenance, residential therapeutic communities, outpatient drug-free treatment, and short-term detoxification programs. Another category was created to include those people who applied for, but never began treatment. Bi-monthly status reports were received over a year for those clients who entered treatment.

Follow-up interviews took place five to seven years after initial assessment for treatment. A total of 4627 subjects were interviewed from each of three annual cohorts for the years 1969-71, 1971- 72, and 1972-73. Treatment outcome was assessed retrospectively by interviews in which subjects were asked about their behaviour during each month between the end of treatment and the time of the interview. The outcomes assessed in these interviews were: drug use, crime, employment status, alcohol consumption, living situation, and further treatment episodes.

The findings from the DARP have been reported in a series papers (e.g. Bracy & Simpson, 1982-83; Simpson, 1981; Simpson et al., 1982; Simpson & Sells, 1982). In terms of comparisons between treatments, patients in methadone maintenance, therapeutic communities and outpatient drug-free programs had better outcomes than those who went through detoxification programs or had no treatment at all (Simpson & Sells, 1982). This finding was apparent in the year immediately following treatment, and was still evident, although the differences had diminished, at the five-year follow-up (Bracy & Simpson, 1982-83). One DARP study, however, found that these differences in treatment outcome did not persist for such long periods (Simpson et al., 1982).

The finding of the DARP research that is most often quoted is the positive relationship between time spent in treatment and the post-treatment performance (see Chapter 9 for an extensive discussion of this issue). The length of time spent in treatment was predictive of improved treatment outcome for treatment periods of at least one year for methadone maintenance. In general, there was a linear relationship between improvement and treatment duration between three months and two years, the longest treatment period given the duration of the project (Simpson, 1981). The only other variable that predicted post-treatment performance was the pre-treatment criminal history of the person, in which case higher levels of criminal activity predicted poorer outcome in terms of opioid use, employment and crime (Simpson & Sells, 1982).

There are a number of problems with the DARP studies as evaluations of the effectiveness of methadone. The major problem was that the follow-up data on post-treatment outcome was collected retrospectively for the four years preceding follow-up. The credibility of month by month recollections of drug use over such a period is doubtful. Behaviour during periods closer to the follow-up period would be more reliable, but could be confounded by a number of unknown variables that may have nothing to do with treatment, as the authors acknowledge.

III. f. The Treatment Outcome Prospective Study

The Treatment Outcome Prospective Study (TOPS) (Hubbard et al., 1984; Hubbard et al., 1989) was a prospective study of over 11 000 illicit drug users who applied for treatment in 41 programs in the USA. The major drug treatment modalities represented by the participating programs were methadone maintenance treatment, residential therapeutic communities, and outpatient drug-free treatment. All the applicants for treatment in the participating treatment programs for the years 1979, 1980, and 1981 were interviewed about their drug use, criminality and other behaviour before treatment, and were then followed up during treatment.

The purposes of the study were to assess the effect of treatment on clients' behaviour, and to identify client and treatment factors that predicted different treatment outcomes. The key outcomes measured were illicit drug use, criminal activity, employment, depression and suicide. All outcomes were assessed by subjects' self-reports which were validated by a variety of procedures. An important feature of the TOPS study was the use of statistical techniques to control for the influence on outcomes of potential confounding variables such as sex, marital status, education level, age, race/ethnicity and number of prior admissions.

The study can be divided into two phases. In the first in-treatment phase all applicants for treatment were interviewed and followed up every three months while they remained in treatment. In the second phase, selected cohort subgroups for each admission year were followed up at three months, one year, two years and at three to five years after treatment (the length of the latter follow-up depending on their year of admission).

The results of the TOPS study confirmed those of previous studies in that all three treatment modalities were associated with a reduction in illicit drug use. TOPS also confirmed the observation that length of time in treatment was an important predictor of post-treatment behaviour for some of the outcomes measured (see Chapter 9).

Methadone treatment had the best retention rates of the three treatment modalities in TOPS. Patients in methadone maintenance treatment were less likely to drop out of treatment than those in drug-free outpatient and therapeutic communities: after three months, 65% of methadone patients remained in treatment, whereas less than 40% of the outpatient drug-free clients and 44% of the residents in therapeutic communities remained in treatment more than three months. At the end of six months 50% of patients were still in methadone maintenance treatment.

Patients in methadone maintenance substantially reduced their heroin use while in treatment, with less than 10% regularly using heroin (weekly or daily) after three months. Table 3 summarises the results of a logistic regression analysis which examined the likelihood of regular heroin and predatory criminal activity in the year after leaving treatment (or the past year in the case of patients who remained in long-term methadone maintenance) on the part of patients grouped by time in treatment. The comparison group consists of those patients who remained in methadone maintenance for less than one week. The numbers in the table are odds ratios which can be interpreted as follows. In contrast to the comparison group, odds ratios of less than one for each of the groups who spent more than one week in methadone maintenance represent a reduction in the likelihood of the outcome (for example, heroin use) and odds ratios greater than one indicate that the outcome is more likely. When the decreases or increases in the likelihood of the behaviour become statistically significant this has been indicated. The logistic regression analyses controlled for a variety of potential confounding variables and its results can be understood, in simple terms, as representing the patients' outcomes when all other important variables have been controlled for.

As can be seen from Table 3, a significant reduction in regular heroin use was observed among patients who spent more than a year in methadone maintenance but who had subsequently left at some time before follow-up interviews took place. Patients who left treatment within one week of entry were twice as likely to be regular heroin users as were those who stayed for a year or more. Similarly, compared with those patients who remained in methadone maintenance throughout the period to follow up, the group that stayed less than a week were four times more likely to be regularly using heroin.

Criminal activity was assessed by self-reported predatory crimes such as breaking and entering and robbery. Among patients in methadone maintenance, one-third reported committing a predatory crime in the year before treatment. This dropped to 10% during the first month of treatment. As Table 3 suggests, significant reductions in self-reported predatory crime were only observed while patients remained in methadone maintenance. Post-treatment criminal activity was predicted by level of pre-treatment involvement in crime but was unrelated to aspects of treatment. Methadone treatment, therefore, was associated with a reduction in criminal activity during treatment but did not permanently change the behaviour of the more criminally involved patients in the post-treatment period.

TOPS is the largest controlled prospective study of drug treatment to be conducted in recent times. It provides information on the behaviour of a large number of subjects before, during and after treatment in methadone maintenance, therapeutic communities, and outpatient drug-free programs. The use of statistical procedures to control for the influence of variables like client characteristics on treatment outcome lends more weight to the findings. The results of the study suggest that participating in methadone maintenance treatment is associated with marked and enduring reductions in heroin use and criminal activity.

IV. Pre-Post-Studies of Treatment Effectiveness

The interpretation of pre-post-observational treatment studies is even more problematic than the interpretation of comparative observational studies because of the absence of a comparison treatment condition. Inferences about treatment effectiveness from pre-post-studies are often made either by comparing the outcomes of people who dropped out of treatment with the outcomes of people who remained in treatment, or by examining the relationship between length of time in treatment and outcome. Such inferences are of uncertain value because of the existence of a plausible rival explanation of a positive relationship between length of time in treatment and patient outcome -- namely, that those with the best outcome (e.g. who were the least dependent on opioids, and the most motivated to discontinue drug use) were more likely to be retained in treatment. This is a form of 'selection' bias.

The quasi-experimental strategy can provide a limited evaluation of such alternative explanations. First, the hypothesis that patients with a good outcome were more likely to be retained in treatment can be tested by measuring the relevant characteristics (e.g. degree of dependence, previous treatment history, and motivation to change) of those who do and do not remain in treatment. Secondly, if selection bias is operating, statistical methods (e.g. covariate adjustment) can be used to discover whether the relationship between treatment duration and patient outcome persists when differences in patient characteristics are taken into account.

IV. a. Gearing and Schweitzer (1974)

Gearing and Schweitzer (1974) provided an independent evaluation of 17 500 patients admitted to Dole and Nyswander's long-term methadone maintenance program between January 1964 and December 1971. They identified four cohorts by date of admission, which defined changes in admission criteria over time, and a shift from inpatient to outpatient induction to the methadone program. Outcome was evaluated by changes in social productivity, arrests for predatory crime, and mortality rates.

The demographic characteristics of patients entering the program changed over the period of study. The average age declined from 33 to 29 years; the proportion of women increased from 15% to 23%; and the percentage of whites decreased from 40% to 32%, while the percentage of Hispanic persons increased from 19% to 26%. Despite these changes in patient characteristics, retention in treatment was high and relatively constant across the first three cohorts who had been enrolled for sufficiently long for it to be assessed, namely, 90% after one year, 80% after two years, and 75% after three years.

Retention in treatment was associated with improved social productivity, reduced crime and a reduced mortality rate. The percentage who were employed, attending school or homemakers increased with treatment for all three cohorts, although less so for later cohorts. The three cohorts showed similar decreases in rates of arrest with increasing time in treatment, namely, 6.5% in the first year, 4.6% in the second year, 3.1% in the third year, and 2.9% in the fourth year.

The only comparative component of the study was a comparison of mortality rates among 3 000 patients while in treatment, 850 patients who left methadone, 100 patients entering detoxification in 1965, and the general New York population in 1969 to 1970 in the age range 20 to 54 years. The rates among patients while in treatment (7.6 per 1000 population) were not substantially higher than those in the general population (5.6 per 1000 population) which is impressive given that the mortality rate among opioid users is generally higher than that in the general population. The mortality rate among those entering detoxification was almost 11 times higher than that of those in treatment (82.5 per 1000 population), while those who had left treatment had a rate that was almost four times higher than that of those who remained in treatment (28.2 per 1000 population). The percentage of deaths that were judged to be probably or possibly drug-related was 50% among those in treatment, 80% among those who died after leaving treatment, and 100% among those who entered detoxification.

Gearing and Schweitzer's study is uncontrolled and, with the exception of mortality, there is no comparison group with which to compare outcome in the absence of treatment. Nonetheless, their results are noteworthy in replicating the positive results for drug use and crime reported by Dole and Nyswander in their early reports, and showing that these positive outcomes were sustained over four cohorts of 17 500 patients who were admitted to their program over a period of eight years. The outcomes assessed were relatively objective, and the advantage in favour of methadone maintenance was substantial in the case of mortality where some comparative data were available.

IV. b. Ball and Colleagues

Ball and his colleagues (Ball et al., 1988; Ball & Ross, 1991) have recently reported the results of a large-scale outcome study of methadone maintenance treatment involving six methadone maintenance programs, two in each of Baltimore, Philadelphia and New York, over a three-year period between 1985 and 1987. During the winter of 1985-86, 633 male patients were interviewed, and 506 were re-interviewed a year later about their drug use history, their last period of injecting drug use, and their past and current criminal activity. The initial sample consisted of 113 new admissions and 520 patients who had been in treatment for at least six months. At follow-up 388 remained in treatment and 107 had left treatment at some time during the intervening year. The characteristics of the methadone maintenance programs were also extensively assessed to determine if there was any relationship between program characteristics and outcome.

The findings suggested that methadone maintenance had a dramatic impact on injecting drug use and crime among the 388 patients who remained in treatment during the follow-up year: 36% had not injected since the first month on methadone maintenance, 22% had not injected for a year or more, and 13% had not injected in the past one to 11 months. In all, 71% had not injected in the month prior to interview, and the rate of injection among the 29% who had injected in the past month was substantially less than before treatment.

The results also suggested that some programs were more effective at eliminating drug use than others: four of the programs reduced drug use by between 75% and 90%, whereas around 56% of patients in the other two programs were still injecting. Among the 107 patients who had left treatment by the time of follow up, 68% had relapsed to injecting drug use. The relapse rate increased linearly over time reaching a maximum of 82% among patients who had been out of treatment for more than 10 months. Those patients who had been in the less successful programs had higher relapse rates than those who had been in the more successful programs. Overall, these results suggested that methadone maintenance was effective at substantially reducing injecting drug use among the majority of patients, and that some methadone maintenance programs were more effective than others in achieving this goal.

The reduction of crime associated with retention in methadone maintenance also appeared impressive. The study sample had an extensive criminal history prior to entering methadone: a total of 4723 arrests, with a mean of nine arrests for the 86% of the sample who had been arrested. Sixty-six per cent of the group had spent some time in gaol, 36% having been incarcerated for two years or more. Although these figures indicate extensive criminal involvement, they seriously underestimate criminal activity which is better estimated by self-reported crime.

The sample admitted to 293 308 offences per year during their last period of addiction. Among those who admitted committing criminal acts, each person committed an average of 601 crimes per year (range 1 to 3588), and had committed criminal offences on an average of 304 days per year during their last addiction period. After entry to methadone, the number of self-reported offences declined to 50 103 crimes per year and the mean number of 'crime days' per year decreased from 238 in the year prior to entry to 69 crime days during the early months of methadone maintenance. The number of crime days continued to decline with the number of years spent in treatment. In terms of the number of crimes committed, the reduction during methadone maintenance was 192 000 offences per year. As Ball and Ross (1991) remark, such a substantial reduction in criminal activity among heroin users is usually only achieved by incarceration. As might be expected, given the relationship between drug use and crime, some programs were more successful than others in reducing crime.

According to Ball and Ross (1991) and Ball et al. (1988) the more effective programs in their study were characterised by the following features: they prescribed higher doses of methadone and had maintenance rather than abstinence as their treatment goal (see Chapter 6); they offered better quality and more intensive counselling services (see Chapter 8); they provided more medical services; they retained their patients in treatment and managed to achieve compliance in terms of regular clinic attendance; they also had close, long-term relationships with their patients; and they had low staff turnover rates.

Two important points emerge from this study. The first is that methadone maintenance treatment programs differ in effectiveness. The second is that, on average, methadone maintenance treatment is effective for the majority of patients while they are maintained on methadone; they relapse quickly once they leave treatment. The fact that the sample in this study was restricted to inner-city males with long histories of dependence (mean = 11.2 years), and long-standing criminal involvement, provided a stringent test of methadone maintenance treatment. It is reasonable to assume that if methadone maintenance treatment is effective in this difficult population, then it would also be effective with a less troubled group. As was the case with the DARP study, however, there is some concern about the reliance upon retrospective self-reports about drug use and crime. In the case of crime, the study may have overestimated the impact of methadone maintenance in that it compared crime during the last period of addiction with that during treatment. This may have exaggerated the difference between the amount of crime reported before and during methadone maintenance.

IV. c. General Accounting Office Study

In 1989 the Chairman of the Select Committee on Narcotics Abuse and Control asked the General Accounting Office of the United States Congress to evaluate the effectiveness of methadone maintenance treatment programs in reducing heroin and other drug use. The General Accounting Office staff selected 24 methadone maintenance programs in California, Florida, Massachussets, New Jersey, New York, Texas, and Washington State which had at least 200 patients enrolled, and had operated for at least five years. They obtained data on heroin and cocaine use by urine analysis from 5600 patients who had been enrolled for at least six months in methadone maintenance treatment. This included all patients from 21 programs and a random sample of patients from the other three programs. The effectiveness of the programs was evaluated in terms of whether less than 20% of patients who had been enrolled for at least six months were still injecting heroin. By this standard 10 of the 24 programs were judged to be ineffective.

The General Accounting Office study confirms Ball and Ross's (1991) findings that there was substantial variability between the programs' policies. The programs varied widely in the frequency with which urinalyses were conducted, in the consequences of continued drug use, and in the average dose of methadone. The majority of the programs (21 out of 24) provided sub- optimal doses of methadone, as defined by the minimum dose recommended by the National Institute for Drug Abuse, namely, 60 mg per day (Schuster, 1989). The mean dose in all 24 programs was 48 mg of methadone per day.

Given these differences in policies it is not surprising that there were also substantial differences in patient outcomes between programs. The proportion of patients in each program who had been retained in treatment at six months varied between 83% and 4%, with an average of 54%. The proportion that continued to inject heroin ranged between 13% and 67%, and the proportion who injected cocaine varied between 0% and 40%. Secondary analyses of the grouped data by Newman and Des Jarlais (1991) suggested that the mean dose of methadone in each program predicted both retention and continued illicit heroin use.

Overall, the results of the General Accounting Office study provide evidence that many methadone maintenance treatment programs in the United States are relatively ineffective in terms of reducing injecting drug use among those they retain in treatment. Nevertheless, the results also provided suggestive support for the Dole and Nyswander model of treatment in that the treatment programs that used adequate doses of methadone had the best outcomes in terms of patient retention and the frequency of injecting heroin use in treatment (Newman & Des Jarlais, 1991). (Top)

V. A Comparison of Results:
Randomised Controlled Trials and
Obsevational Studies

The observational studies of the effectiveness of methadone maintenance treatment generally support the results of the small number of randomised controlled trials in showing that methadone maintenance decreased heroin use and criminal activity. These studies also revealed two other features of contemporary methadone maintenance treatment. The first was that there was substantial variation between different programs in outcomes as measured by treatment retention and continued heroin and other illicit drug use, which most clearly emerged in the studies by Ball and Ross and the General Accounting Office. The second was that the average results of methadone maintenance treatment in recent observational studies are not as impressive as those reported from the randomised controlled trials. For example, the retention rates from the randomised controlled trials are usually of the order of 70% or more after one year whereas the retention rate in the DARP and General Accounting Office studies is approximately 50% after six months. Similarly, the early randomised controlled trials reported very little continuing heroin use among those who remained in treatment whereas the proportion who continued to use heroin in the General Accounting Office study, for example, was as high as 67% in some programs.

There are a number of candidate explanations for the differences in the apparent effectiveness of methadone maintenance between the randomised and observational studies. First, it is likely that randomised controlled trials have provided a somewhat optimistic estimate of treatment effectiveness. In order to produce clear results, such studies usually exclude some of the more difficult patients from entry, and they often have greater degree of control over the quality of the treatment that is provided than usually occurs under the ordinary exigencies of clinical practice. In addition, in many of the initial randomised controlled trials, patients who were denied access to methadone maintenance treatment were unlikely to receive it or any other form of treatment elsewhere. The comparison of methadone maintenance treatment with control treatment in such studies was not attenuated by the effects of treatment obtained elsewhere, as has more often been the case in recent observational studies, such as that of Bale and his colleagues.

Secondly, there is clear evidence that many current methadone maintenance treatment programs in the USA (and, as we shall see, in Australia) have departed from the original model of Dole and Nyswander in directions that are likely to reduce average effectiveness. As the data presented by D'Aunno and Vaughn (1992) and in the General Accounting Office report show, many programs have reduced average methadone dose and put pressure on patients to become abstinent from all opioids, including methadone. There has also been a wide variation in the practices that different programs follow, and an absence of any interest in evaluating their impact on treatment outcome.

Thirdly, there have been important historical changes in patterns of illicit drug use between the time when methadone was introduced and when the more recent observational studies were conducted. The most obvious of these has been the spread of poly-drug use among people presenting for treatment. In the USA, cocaine use in particular has become widespread among methadone patients. Methadone has no specific effects on cocaine use, neither blocking the effects of cocaine nor avoiding withdrawal symptoms, so it has had minimal impact on the use of non-opioid illicit drugs.

Fourthly, the context within which methadone maintenance treatment has been provided has changed dramatically in the past 20 years. The Federal financial support for methadone maintenance treatment in the mid-1970s has given way in the USA to fiscal restraint on program budgets with consequent reductions in the quality of treatment services, and to a steady decline in the number of treatment places on methadone maintenance, with no new clinics having opened during the past 10 years. This decline in the quantity and quality of treatment in the USA has been accompanied by an increase in Federal government regulations that have encouraged the reduction in average methadone dose and the introduction of time limits (usually two years) on treatment.

VI. Summary

The findings of the comparative observational studies of methadone maintenance are consistent with the results of the small number of randomised controlled trials in showing that methadone maintenance retained patients in treatment and substantially reduced illicit opioid drug use and involvement in criminal activity in comparison with those patients who did not enter treatment. The pre-post-studies generally agree with the results of the comparative studies in showing that the longer patients remain in treatment, the less likely they are to inject heroin or to engage in criminal activity. In those studies that have used a quasi-experimental strategy to evaluate rival explanations, these results have proved robust.

The observational studies also indicate that there is substantial variation between different programs in their effectiveness in retaining patients in treatment, and reducing their drug use and criminality while they are in treatment. Analyses of the characteristics that predict the variations between programs in retention, drug use and criminality have generally supported the original model of Dole and Nyswander in showing that programs with higher doses, a maintenance goal and ancillary services have better outcomes than programs that use lower doses and aim to achieve abstinence.

The average effects of methadone maintenance treatment in the observational studies have been lower than those observed in the randomised controlled trials. Among the more important reasons that can be identified for the decline in the average effectiveness of methadone maintenance have been: a systematic departure from the model of methadone maintenance proposed by Dole and Nyswander in the direction of lower dose and time-limited treatment; a decline in the quality of methadone maintenance programs in the face of fiscal restraint and federal regulations in the USA; and changes in the patient population of methadone maintenance treatment programs with the rise in poly-drug use.