Louisiana Resource locator

More facts on alcohol and other drug addiction

Chapter 1 
Chapter 2
Chapter 3
Chapter 4
Chapter 5
Chapter 6
Chapter 7
References



CHAPTER 6.  EVALUATIONS OF DRUG ABUSE TREATMENT

6.1    Introduction

      Our goals for this chapter were similar to those in the previous chapter on alcoholism treatments.  Based on a review of the extant literature, we hoped to learn the answers to the following questions:

(1)    What treatments are being used to treat drug abusers?

(2)    Are these treatments effective?  Do they work?

(3)    Which treatments are the most cost effective?

(4)    What are the benefit/cost ratios for these treatments?

(5)    Which treatments are the most effective for the different types of patients?

      At best, we were only able to find partial answers to these questions.  The reason for this is that evaluation of drug abuse treatment effectiveness is very complex, involving an assessment of the nature of treatment provided, the problems and characteristics of clients, and the multiple outcomes attributable to interaction between treatments and clients (Hubbard et al. 1989).  Because treatment programs have different goals, they use different outcomes to measure success.  All of the programs seek to reduce or eliminate drug use among patients and many seek to lessen criminal activity as well.  Some programs attempt to increase their patients’ employment and earning opportunities, while others try to change their patients’ lifestyles, social behavior, and mental health.  This review will focus on the program’s ability to reduce drug use and criminality and to improve patients’ employment and earnings.  Patients, their families, and health care administrators are particularly interested in the decline in drug use because of the health benefits, the reduction of stress, and the associated decline in diseases such as AIDS, TB, and hepatitis spread by the use of needles (McLellan, Woody, and Metzger 1996).  Taxpayers and government officials are, perhaps, more interested in the effects of treatments on criminality, employment and earnings for obvious reasons.

      Part of the complexity of the study of drug abuse treatment effectiveness arises from the diversity of approaches and settings in which treatment occurs.  Drug abuse treatment is provided in a number of modalities by hospitals, community public health agencies, and a variety of independent organizations (Hubbard et al. 1989).  Prior to 1990, there were four major modalities (detoxification programs, methadone maintenance programs, residential drug free, and outpatient drug free) and effectiveness results were frequently reported at this level.  In the 1990s, a number of new modalities were introduced and some of them have recently been evaluated (NIDA 1999).  Comparisons of the relative effectiveness of drug abuse modalities is difficult because they have different goals and outcome measures, they provide different treatment services, and they serve different types of patients (DeLeon 1984, Hubbard et al. 1989, and NIDA 1999).

      The outline of the chapter is as follows.  Section 6.2 discusses the major types of drug abuse treatments and the various types of modalities.  In the next section, we outline the major methodological problems involved in estimating the effectiveness of drug abuse treatments.  Section 6.4 reviews the literature on the effectiveness of drug abuse treatments.  We begin with an overview of the data sources and the overall effectiveness of such treatments.  Then we consider modality effectiveness.  Studies have shown that the modalities treat different types of clients and that outcomes are sensitive to the problems and characteristics of clients.  We shall investigate this literature next.  After this, we shall examine the literature on the effects and determinants of length of treatment on program effectiveness.  This is followed by a discussion of the effectiveness of individual treatment and program components.  We shall also discuss the “MATCHING” hypothesis.  The section ends with a discussion of the problems of relapse and readmission.

      In Section 6.5 we shall review the sparse literature on the cost-effectiveness of drug abuse treatment programs.  The few existing benefit/cost studies of drug abuse treatment programs will be considered in the next section.  Section 6.6 critically reviews some recent benefit/cost studies of state drug abuse programs from the taxpayer’s point of view.  The final section summarizes our findings and conclusions on drug abuse treatment evaluations.

6.2    Drug Abuse Treatments and Modalities

      According to Anglin and Hser (1990), there is no simple cure for drug dependence.  Drug dependence is a chronic condition, and relapse is often the rule.  Many biological, socio-cultural, economic, and psychological factors are believed to contribute to drug abuse.  For this reason, the treatment of drug abuse is not simply a medical issue; it involves a wide spectrum of social considerations.  Because of the complexity of addiction, drug addiction treatment involves many components.  Some of those components focus directly on the individual’s drug use, but others focus on restoring the addicted individual to productive membership in the family and society (NIDA 1999).  Treatment for drug abuse and addiction is delivered in many different settings, using a variety of behavioral and pharmacological approaches.

6.2.1  Drug Abuse Treatment Modalities

      Research studies on drug addiction have typically classified treatment programs into several general types or modalities (NIDA 1999).  These include the following: 

(1)     Methadone Maintenance Treatment for opiate addicts usually is conducted in outpatient settings.  These programs use a long-acting synthetic opiate medication, usually methadone or LAAM, administered orally for a sustained period at a dosage sufficient to prevent opiate withdrawal, block the effects of illicit opiate use, and decrease craving.

(2)     Narcotic Antagonist Treatment Using Naltrexone for opiate addicts usually is conducted in outpatient settings.  Naltrexone is a long-acting synthetic opiate antagonist that blocks the effects of self-administered opiates, including euphoria.

(3)     Outpatient Drug-Free Treatment programs differ in the types and intensity of services offered, such as drug education and counseling.  Such treatment costs less than residential or inpatient treatment and often is more suitable for individuals who are employed or who have extensive social supports.

(4)     Long-Term Residential Treatment provides care 24 hours per day, generally in non-hospital settings.  The best-known residential treatment model is the therapeutic community (TC), which focuses on the “re-socialization” of the individual and uses the program’s entire “community” as active components of treatment.  Addiction is viewed in the context of an individual’s social and psychological deficits, and treatment focuses on developing personal accountability and responsibility and socially productive lives.  Some residential treatment programs employ other models, such as cognitive-behavioral therapy.

(5)     Short-Term Residential Programs provide intensive but relatively brief residential treatment based on a modified 12-step approach.  These programs were originally designed to treat alcohol patients, but during the cocaine epidemic of the mid-1980’s, many began to treat illicit drug abuse and addiction.

(6)     Medical Detoxification is a process whereby individuals are systematically withdrawn from addicting drugs in an inpatient or outpatient setting, typically under the care of a physician.  Detoxification is sometimes called a distinct treatment modality but is more appropriately considered a precursor of treatment, because it is designed to treat the acute physiological effects of stopping drug use.

(7)     Prison-Based Treatment Programs

Offenders with drug disorders may encounter a number of treatment options while incarcerated, including didactic drug education classes, self-help programs, and treatment based on therapeutic community or residential milieu therapy models.

(8)     Community-Based Treatment for Criminal Justice Populations

A number of criminal justice alternatives to incarceration have been tried with offenders who have drug disorders, including limited diversion programs, pretrial release conditional on entry into treatment, and conditional probation with sanctions.  For example, drug courts mandate and arrange for drug addiction treatment, actively monitor progress in treatment, and arrange for other services to drug-involved offenders.  The Treatment Accountability and Safer Communities (TASC) program provides an alternative to incarceration by addressing the multiple needs of drug-addicted offenders in a community-based setting.  These programs include counseling, medical care, parenting instruction, family counseling, school and job training, and legal and employment services.  The key features of TASC include (1) coordination of criminal justice and drug treatment; (2) early identification, assessment, and referral of drug-involved offenders; (3) monitoring offenders through drug testing; and (4) use of legal sanctions as inducements to remain in treatment.

6.2.2  Drug Abuse Treatments

      Most treatment programs usually provide a combination of service components besides their modality-specific treatment.  The following treatment components have been identified by Anglin and Hser (1990) and the National Institute of Drug Abuse (1999).

(1)    Drug Counseling is the primary support service provided to clients in most drug-treatment programs and is common across all modalities.  Drug counseling focuses on assisting the client in practical problem solving for day-to-day living.

(2)    Drug Education contributes to the client’s understanding of the biological, familial, psychological, and social factors that contribute to drug dependence.

(3)    Pharmacotherapy interventions with serious drug dependence typically involve a period of detoxification for the purpose of medically managing drug withdrawal systems.  Agents such as desiprimine and bromocriptine have been used to treat depression with dysfunctional levels of cocaine use.  Long-term use of narcotic antagonists such as naltrexone or clonidine have been used to treat heroin addiction.  Methadone substitution for illicit opiate dependence is perhaps the most lengthy pharmacotherapy applied in drug treatment.

(4)    Psychotherapy components vary widely across the programs.  The unit of the intervention may be the client, the client and spouse, or the family.  Group counseling is commonly used.

(5)    Education and Vocational services are often provided because of clients educational, employment, and legal problems.  Job skill training can enhance employment and earnings opportunities.

(6)    Urine testing provides one objective measure of compliance with the treatment goals of reducing primary drug use.  Such testing is common in all modalities.

(7)    Relapse-prevention training can be used to insulate the carefully cultivated attitudes, skills, and intentions derived from the treatment process from individual and community influences.  The relapse prevention approach to the treatment of cocaine addiction consists of a collection of strategies intended to enhance self-control (NIDA 1999).

(8)    Social and Community Support assists recovery from drug dependence during treatment after discharge.  This includes self-help or mutual support organizations like Narcotics Anonymous and Cocaine Anonymous.  Group homes can provide a useful intermediate step from residential care to community care.

(9)    The Matrix Model draws heavily from other treatment approaches.  It includes relapse prevention training, family and group therapies, drug education and self-help participation.  The therapist functions simultaneously as teacher and coach, fostering a positive encouraging relationship with the patient and using that relationship to reinforce positive behavior change.

(10)                       Supportive-Expressive Psychotherapy has two main components:  supportive techniques to help patients feel comfortable in discussing their personal experiences and expressive techniques to help patients identify and work through interpersonal relationship issues.  Special attention is paid to the role of drugs in relation to problem feelings and behaviors, and how problems may be solved without recourse to drugs.

(11)                       Motivational Enhancement Therapy is a client-centered counseling approach for initiating behavior change by helping clients to resolve ambivalence about engaging in treatment and stopping drug use.  This approach employs strategies to evoke rapid and internally motivated change in the client, rather than guiding the client stepwise through the recovery process.

6.3    Methodological Problems in Evaluating Drug Abuse Treatments

      The number of evaluation studies of drug abuse treatment programs has increased rapidly since the early 1970s.  These studies vary in scope and methodology, which makes it difficult to interpret and compare their results (French 1995 and Anglin and Hser 1990).  The methodological problems involved in evaluating drug abuse problems are quite similar to those in evaluating alcoholism treatments, which were discussed in Section 5.3.  Therefore, only a brief summary of these problems will be presented in this section.

Problem 1.  Standardizing Treatment Protocols

      There is widespread disagreement among clinicians and researchers with respect to the causes of drug addiction and how it should be treated.  Therefore, treatments, processes, and procedures vary from program to program.  Evaluation studies of drug abuse programs rarely provide detailed information on these components so we are only beginning to learn the relative effectiveness of these components (see Section 6.4.5 below).  Recently, the National Institute of Drug Abuse (1999) put out a manual explaining how to measure and improve costs, cost-effectiveness, and cost benefit for substance abuse treatment programs based on a cost-procedure-process-outcome analysis (CPPOA) model.  The manual is intended to help program managers understand why their programs are effective and to find the cost-effectiveness of individual processes and procedures so that programs can be improved.

Problem 2.  Standardized Outcome Measures

      The primary goal of most drug abuse treatment programs is abstinence or at the very least reduction in drug use.  But studies define and measure success with respect to abstinence and reduced drug use in different ways, making it difficult to compare the relative effectiveness of the programs.  Drug abuse programs have a number of other goals and associated output measures including:  “decreased levels of illegal activities such as drug trafficking, property crime, or prostitution; increased employment and decreased reliance on social service agencies; improved social and family functioning; improved psychological functioning, and decreased mortality and improved physical health” (Anglin and Hser, 1990, pp. 415-416).  Because program objectives and output measures are not identical across programs, it is difficult to measure their relative effectiveness.

Problem 3.  Patient Variation

      Over time we have learned more about the heterogeneous nature of the drug-dependent population.  We have learned that certain types of patients self-select into different types of drug abuse modalities and types of treatments (Anglin and Hser 1990).  Some clients respond to treatments better than others and some patient characteristics are significantly correlated with successful outcomes (see Section 6.4.3 below).  It is difficult to separate treatment effects from patient variation when comparing the effectiveness of different programs.

Problem 4.  Costs of Treatments

      Average costs per patient are typically reported (if at all) at the program level rather than actual individual patient costs.  Because most programs tailor treatments to individuals, and individuals remain in treatment for different lengths of time, the average cost per patient is a poor proxy for actual patient cost (NIDA 1999).  In addition, most evaluation studies ignore indirect costs such as individual’s transportation expenses, lost work time, donated time and equipment, and care taking (French 1995).  The omission of such costs makes it difficult to estimate the cost-effectiveness or the cost-benefits of drug abuse programs.

Problem 5.  Research Design

      Like alcoholism treatment evaluations, drug abuse treatment evaluations do not use the experimental design, which randomly assigns patients to experimental patient groups and to a control group which does not receive treatment services, because it is considered unethical to deny services to drug abusers (Anglin and Hser 1990).  In the absence of randomly selected “no treatment” control groups, it is impossible to determine the “true” effectiveness of drug abuse treatments.

      Random assignment reduces or eliminates the problem of patient variation discussed above.  There have been a few attempts to randomly assign patients in drug abuse studies, but many patients refuse random assignment and attrition rates have been so high that the results cannot be trusted (Anglin and Hser 1990).

      Many of the early 1970s evaluation studies of drug abuse programs did not include control groups.  For reasons discussed earlier, we have little confidence in these studies’ findings.  Later studies have used the pre/post research design with patients serving as their “own” control group.  These studies produce results that are favorable biased because of the “ramp up” and “regression-to-the-mean” and “spontaneous recovery” problems discussed earlier.

      A few studies have used a matched control group drawn from participant dropouts or waiting lists.  Unfortunately, we cannot be sure that individuals chosen from these sources have the same motivation to change even if they are matched on other patient characteristics.  Finally, there are a few natural experiment studies where the control group seems quite comparable to the treatment group (Anglin and Hser 1990).  These studies come closer to measuring the “true” effectiveness of drug abuse treatment programs than the other evaluations.

Problem 6.  The “Ramp Up Effect” and “Regression-to-the-Mean Problem

      Anglin and Hser (1990), French (1995), and Estee and Nordlund (2001) have noted that the abnormally high levels of drug use, crime, or both among drug abusers in the period just prior to treatment might cause the observed “improvement” in the post-period to represent no more than a “regression-to-the-mean effect, particularly when the pre-treatment period was short.  Also, the crisis resulting from the high levels of drug use and crime might cause the individuals to seek treatment or to change their behavior without treatment (i.e., “spontaneous recovery”).

Problem 7.  Follow-Up Analysis Problems

      Drug abuse evaluation follow-up periods tend to be short (6 months or one year).  Studies using no control group or “own” control group research designs tend to produce highly favorable results in short follow-up periods because of the “regression-to-the-mean” and “spontaneous recovery effects”.  Studies with longer follow-ups (2 years or more) provide a better indication of a program’s effectiveness.  Follow-ups also face the problem of sample selection due to high attrition rates in drug abuse programs (which often exceed 50%).

Problem 8.  The Relapse Issue

      Drug dependence is a chronic illness and many patients that successfully complete drug abuse treatments will relapse and later re-enter new drug-abuse treatment programs.  Most drug abuse evaluations ignore the issue of readmission.  They focus on individual treatment episodes.  Without tracking patients who return for more treatment, treatment cost and effectiveness measures could be seriously biased (French 1995).  Observed differences in abuse treatment effectiveness could represent different mixes of new and readmitted patients rather than “true” treatment effects.

Problem 9.  Spontaneous Recovery

      We know that some drug abusers recover (i.e., lessen their drug use and reduce their health care expenditures) without treatment (Estee and Nordlund 2001) and follow-up studies often show some degree of improvement for drug abusers that receive only minimal treatment (Anglin and Hser 1990).  For these reasons, it is difficult to accept at face value the highly favorable results reported by evaluation studies that have no control group or that use an “own-patient” control group.  If a spontaneous recovery effect is present, their estimates of success are biased upward by an unknown amount.

Problem 10. Statistical Analysis

      Most research data obtained in drug-treatment evaluation have been analyzed at the descriptive and comparative level.  Simple descriptive analysis of means can detect average changes in outcome variables between periods, but it cannot explain or predict the causes of the changes.  We need more advanced statistical techniques such as multivariate regression analysis to help explain why a certain result occurred (French 1995).  Multivariate analysis can be used to estimate the effects of patient characteristics and program components on post-treatment outcome measures.  With advanced statistical techniques, we can begin to learn why programs and individual drug abuse treatments are effective.  Over time, drug-treatment evaluations have become more sophisticated in terms of research design and statistical analysis, but there is still a long way to go before we are able to measure the “true” effectiveness of drug abuse treatment programs.

6.4    The Effectiveness of Drug Abuse Treatments

      A considerable body of evaluative research on the treatment of drug abuse has been generated over the past thirty years, by clinical experience and project research and from three large evaluative studies.  The Drug Abuse Reporting Program (DARP) was the first comprehensive, nationally based evaluation of drug abuse treatment effectiveness.  DARP examined the admission records of over 44,000 clients in 52 NIDA-supported agencies during the period from 1969 to 1974 (Hubbard et al. 1989).  At that time, the drug of choice was opiates.  Effectiveness measures were reported in DARP studies based on one, six, and twelve-year follow-ups.

      The second national evaluation study, Treatment Outcome Prospective Study (TOPS), was also funded by the NIDA.  The TOPS study included 11,750 clients in three annual admission cohorts – 1979, 1980, and 1981.  This study collected more information about the nature of drug abuse treatment and the characteristics and behavior of abusers prior to treatment.  Evaluation reports were conducted during treatment and based on, one, two, three, and five-year follow-ups.  Most TOPS studies use the pre/post research design with no separate control group.  The studies do report outcomes for patients who received minimal (less than 3 months) treatments.  The TOPS research also includes studies of the outcomes for clients involved with the criminal justice system, particularly those referred to treatment by the Treatment Alternatives to Street Crime (TASC) programs (Hubbard, Rochal, Craddock, and Cavanaugh, 1984).  The TOPS studies included far more non-opioid drug abusers than did the DARP studies.

      Finally, the NIDA supported the Drug Abuse Treatment Outcome Study (DATOS) which tracked 10,010 admissions to 96 programs in 11 cities for 1991-1993.  This includes 2,774 admissions to long-term residential programs; 2,574 to outpatient drug free programs, 3,122 to short-term inpatient programs, and 1,540 to outpatient methadone treatment (DATOS 2003b).  DATOS analysis was limited to clients who had stayed in treatment for at least a month.  Evaluations were based on one and five year follow-ups.  The DATOS studies did not include a control group.  DATOS collected a large amount of data on clients, treatments, and the program environment, which was intended to help us understand how drug abuse treatment works (Franey and Ashton 2002).

      Literally hundreds of drug abuse treatment effectiveness studies have been published.  There was no way to incorporate all of these studies in this review.  Our analysis of drug abuse treatment effectiveness draws heavily from seven survey papers (Tims and Ludford 1984, Hubbard et al. 1989, Anglin and Hser (1990), McLelland, Woody, and Metzger 1996, NIDA 1999, Franey and Ashton 2002, and numerous DATOS sources 2003).  A number of individual studies were also reviewed and they will be incorporated into the discussion.

6.4.1 The Overall Effectiveness of Drug Abuse Treatments

      Early in the literature, most researchers concluded that, in general, drug abuse treatments are  effective in reducing patient’s drug use and criminal behavior, but somewhat less successful in improving patient’s social behavior, employment opportunities and mental health.  For example, see the statements by Joffe, Hubbard et al., Sevay, and Tims and Holland in the NIDA volume edited by Tims and Luford (1984).

      According to Hubbard et al. (1989), the body of research emerging from the DARP study provides convincing evidence of the effectiveness of drug abuse treatment in community-based settings.  Significant reductions in drug use were reported.  For example, 82% of DARP clients reported they frequently used heroin in the year before treatment.  One year after treatment, only 63% reported any opioid use and 47% reported daily use.  Six years after treatment, only 42% reported any use and 25% reported daily use.  Between 6 to 12 years after treatment, 23% increased their frequency of opioid use and 23% decreased their frequency of opioid use or stopped entirely.

      Although fewer in number, cocaine users in the DARP sample also reduced their consumption following treatment.  During the pre-treatment period, 38% of their sample of opioid users has used cocaine.  One to six years after treatment only 18 to 32% reported using cocaine.  However, in the 12 year follow-up, 39% said they used cocaine.  Finally, the DARP studies indicated that patient’s use of marijuana and alcohol increased after treatment raising the possibility that these drugs were being substituted for opioids, a troubling prospect.  The DARP studies reported a significant decline in criminal activity following treatment.  We shall review these findings in our discussion of modality effectiveness presented later.

      TOPS studies also reported significant declines in drug use and criminal activity following drug abuse treatment (Hubbard et al. 1989).  TOPS compared the use of four types of drugs (heroin, cocaine, psychotherapeutic drugs, and marijuana) for the year prior to treatment, during the treatment period, and one and five years after treatment.  They reported results for patients who stayed in treatment less than 3 months and for patients who received treatment for more than 3 months.  Our current discussion will focus on the latter group.  Relative to usage in the pre-period, the TOPS studies reported a significant decline in cocaine, heroine, psychotherapeutic drugs, and marijuana during the treatment and post-treatment periods.  The results varied across the different modalities and we shall examine the specific outcomes below.  The TOPS studies reported that heavy alcohol use declined significantly during treatments, but it almost returned to pre-period rates shortly after treatment.

      Based on self-reported data, TOPS found that predatory criminal activity (robbery, burglary, larceny, etc.) declined significantly from pre-treatment levels during and after treatments.  The declines varied across the modalities and will be discussed later.

      One of the goals of drug abuse treatments is to get drug abusers back to work and perhaps increase their earnings.  Several TOPS studies examined this issue.  Hubbard (1989) examined the effects of treatments on full-time employment defined as working 35 or more hours per week for at least ¾ of the weeks in the period.  He reported a small but significant increase in employment between the pre-treatment period and the 5 year follow-up for residential and outpatient drug-free patients, but not for outpatient methadone patients.  Several later studies based on TOPS data and using multivariate statistical analysis reported that drug abuse treatments have a small but statistically significant negative effect on illegal earnings and a small positive effect on legal earnings and employment (French, Rachal, Harwood, and Hubbard 1990; French, Zarden, Hubbard, and Rachal 1991; ad. French and Zarkin 1992).

      After reviewing the existing literature and noting the absence of the “ideal” control group, Anglin and Hser (1990), concluded “research on drug abuse treatment demonstrates significant declines in drug use and criminal behavior by drug-dependent clients as a result of treatment” (p. 443).  In their review of the literature, McLellan, Woody, and Metzger (1991) concentrated “on the studies that have used the most rigorous evaluation methods, including random patient assignment, an intent-to-treat design and data collection by independent evaluation, the same standards and methods that are typically applied by the FDA in evaluating new drugs and medical devices” (p. 84).  They concluded that substance abuse treatments dramatically reduce alcohol and drug use, improve patient’s medical and psychological function, sometimes improve earnings from employment and reduce utilization of medical and social services, and reduce AIDS risk behaviors and drug-related crime.  The authors stress that these outcomes are found both in controlled clinical trials of experimental interventions and in large-scale (i.e., DARP and TOPS) evaluations of standard treatments in “real world” settings (p. 84).  The NIDA (1999) echoed these sentiments a few years later.  The institute contends that drug addiction treatment is as successful as treatment of other chronic diseases such as diabetes, hypertension, and asthma.  According to them, research shows that drug treatment reduces drug use by 40 to 60% and significantly decreases criminal activity during and after treatment by 40% or more.  Treatments also reduce the risk of HIV infection and improve the prospects for employment.

      Finally DATOS (2003) one-year follow-ups indicted that drug abuse treatments reduced drug use, illegal activities, and psychological distress on average by 50%.  The success rate varied across the different modalities, which will be discussed below.  Some patients also experienced a 10% increase in full-time employment.  The results from the DATOS (2003d) 5-Year Outcomes were not quite as promising.  Weekly cocaine use was reported by 25% of the sample in the fifth year of follow-up, slightly higher than the 21% for the first year after treatment.  Illegal activity was 25% in the fifth year compared to 16% in the first year after treatment, but this is still well below the 40% level at intake.  Based on these studies, it must be concluded that some drug abuse treatments are effective and that effectiveness varies across treatment modalities.

6.4.2  The Effectiveness of Treatment Modalities

      We shall begin with a discussion of the four historically important treatment modalities and then consider the effectiveness of newer modalities.

(1)   Methadone Maintenance Programs

      Early on, it was learned that methadone maintenance treatments for opioid addicts led to a significant reduction in the use of those drugs (Cooper et al. 1983).  DARP studies confirmed this result.  Because DARP considered multiple outcomes, they defined two types of outcomes for purposes of analysis:  (1) highly favorable outcomes are defined as no use of illicit drugs (except for less-than-daily marijuana use) and no arrests–or- incarcerations in any one or more months during the year and (2) moderately favorable outcomes are defined as no daily use of illicit drugs and no major criminality (i.e., no more than 30 days collectively in jail or in prison, and no arrests for crimes against persons or crimes of profit) (Simpson 1984).  These two outcomes are well below the pre-treatment levels of drug use and criminal activity levels.  At the end of the first year after treatment in methadone maintenance programs, it was reported that 27% of white males had a highly favorable outcome and 41% of black males had a moderately favorable outcome.

      The results for methadone maintenance programs in the TOPS studies were also  favorable, as indicated in Table 6.1.  The percentage of drug use for each of the four drugs declined significantly for methadone patients from the one-year pre-period to the 4

Table 6.1

Changes in Drug Abuse in the TOPS Study

By Modality and Treatment Duration

Source:  Hubbard et al. (1989, p. 181).

post-periods.  For example, for MM patients having 3 months or more of treatments, the percentage of regular heroin users declined from 63.5% in the pre-treatment period to 16.7% at the one-year follow-up.  This represents a 73.7% decrease in drug use.  In the 3-to-5 year follow-up, the figure had risen slightly to 17.5%.  Even MM patients with less than 3 months of treatment experienced a significant decline in drug use.  For this group, the percentage of regular heroin users declined from 65.0% in the pre-period to 31.2% in the one-year follow-up and 24.9 in the 3-to-5 year follow-up.

      TOPS MM patients with 3 or more months of treatments also experienced a decline in criminal activity in the post-treatment period.  In this group, the percentage of patients engaging in predatory crimes was 31% in the pre-period and only about 19% in the one-year follow-up (a 39% decline) and in the 3-to-5 year follow-up it had declined further to about 17% (Hubbard 1989).  Employment gains were more modest for the TOPS MM patients with 3 or more months of treatment.  This group’s percentage of full-time employment was 24.2 in the pre-period.  It declined to 20% in the first year follow-up and then rose to nearly 30% in the second year follow-up.  In the 3-5 year follow-up it was 17.7%, which was below the pre-treatment level (Hubbard et al. 1989).

      Anglin and Hser (1990) were also sold on the effectiveness of methadone maintenance in reducing drug use and criminality.  They were particularly impressed by two natural experiment studies that investigated the effects of the termination of two different methadone maintenance programs, which they had conducted.  NIDA (1999) reviewed the literature on methadone maintenance programs and narcotic antagonist treatment programs using naltrexone and concluded that naltrexone can help patients hold jobs, avoid crime and violence, and reduce their exposure to HIV.  Finally, DATOS (2003b) reported that a one-year follow-up of outpatient methadone treatment showed a 47.6% reduction in cocaine use, a 68.5% reduction in heroin use, a 6.6% increase in heavy alcohol use, a 51.7% reduction in illegal activity, and a 3.5% gain in full-time work compared with pre-treatment levels.

(2)   Long-Term Residential Drug Free Programs

      DeLeon (1984) was the first to survey the literature on the effectiveness of therapeutic communities (TCs) or long-term residential drug free programs.  He only considered published studies of traditional TC programs that had at least 12 months of planned treatments.  He reported that immediate and long-term status of clients in terms of drug use and criminality declined significantly, while measures of employment and/or school involvement increased.  DeLeon did not provide any quantitative estimates of the average outcome effects for the programs surveyed.  He cautioned, however, that these conclusions should be regarded as tentative because of serious methodological problems.  The studies did not include a control group and the follow-up samples may be self-selected to seek, remain in and benefit from TC; or, perhaps, to improve without any treatment.

      DARP studies also reported favorable outcomes for TCs.  At the first year follow-up, 28 percent of adult white male TC patients were reported to have highly favorable outcomes and 40% of adult black males to have moderately favorable outcomes, as defined above (Simpson 1984).

      The TOPS data also showed that residential programs are highly affected.  As shown in Table 6.1, there was a significant decline in residential patients use of all four drugs following treatment.  For example, for those patients having 3 or more months of treatment, the percentage of regular heroin users declined from 30.9% in the year prior to treatment to 11.5% at the 1-year follow-up and to 11.8% at the 3-to-5 year follow-up.  Even those patients having less than 3 months of treatment reported significant declines in drug use at the 1-year and 3-to-5 year follow-ups.

      Criminal activity also declined for TOPS residential patients following treatment.  A little over 60% of residential patients engaged in predatory crime before treatment.  At the first year follow-up, less than 30% reported doing so and in the 3-to-5 year follow-up only 20% reported such activity (Hubbard et al. 1989).

      TOPS residential patients having more than 3 months in treatments showed significant gains in long-term employment.  The percentage of patients working full-time prior to treatment was 15.3%.  At the first-year follow-up, it had risen to 25% and in the 3-to-5 year follow-up it was 38%.

      Subsequent review studies confirm the effectiveness of TCs.  Citing DARP, TOPS, and other studies, Anglin and Hser (1990) concluded that TCs reduce patient’s drug use and criminal activity and increase their employment and social behavior.  DATOS (2003e) reported favorable outcomes for patients in long-term residential treatment.  From the pre-period to the first-year follow-up cocaine use declined 66.7%, heroin use declined 64.7%, heavy alcohol use declined 52.5%, illegal activity fell 60.9% and full-time employment rose 12.5%.

(3)   Outpatient Drug Free Treatment

      DARP data indicated favorable outcomes for outpatient drug free treatment.  At the first year follow-up, 24% of adult white males had highly favorable outcomes and 33% of adult black males had moderately favorable outcomes (Simpson 1984).

      TOPS data for patients with 3 months or more treatment in outpatient drug-free treatment also showed favorable outcomes.  As reported in Table 6.1, drug use declined for all 4 drugs following treatment.  The percentage of regular heroin use declined from 8.6% in the year prior to treatment to 4.9% in the first year follow-up and to 4.6% in the 3-to-5 year follow-up.  Table 6.1 also shows that outpatient drug-free patients having less than 3 months of treatment also reported significant declines in drug use in the post-period follow-ups.

      TOPS data also show a reduction in predatory crime activity for outpatient drug-free patients having more than 3 months of treatment.  For this group, the predatory crime rate was 33% in the year before treatment.  It declined to 19% in the first year follow-up and to 8% in the 3-to-5 years follow-up (Hubbard et al. 1989).

      TOPS data showed highly significant gains in employment for this group as well.  The percentage of patients in full-time employment rose from a pre-treatment level of 27% to 38% in the first year follow-up and to 49% in the 3-to-5 years follow-up (Hubbard et al. 1989).

      DATOS data also confirm the effectiveness of outpatient drug-free treatments.  From the pre-period to the first year follow-up, DATOS outpatient drug-free treatment patients experienced a 57.1 decrease in cocaine use, a 64.0% decrease in marijuana use, a 51.6% decrease in heavy alcohol use, a 36.4% decrease in illegal activity, and a 7.3% increase in full-time work (DATOS 2003e).

(4)   Detoxification Programs

      It was concluded early on that outpatient or inpatient detoxification treatments produce no lasting effects for those addicted to opioids (Cooper et al. 1983).  However, this treatment is effective in reducing drug use temporarily and there is a demand for it (Anglin and Hser 1990).  Furthermore, as Senay (1984) points out from a clinical point of view, detoxification programs are needed to treat emerging episodes, reduce the length and severity of “runs” and attract addicts into the treatments system generally.   A recent NIDA (1999) study echoed these sentiments.

(5)   Criminal Justice Treatment Programs and Clients

      Anglin and McGlothlin (1984) summarized the results their research group obtained from evaluations of the California Civil Addict Program (CAP).  They used a pre/post time series design with a matched comparison group.  They found evidence of positive “ramp up” effects in the pre-period for drug use and criminal activity and negative “ramp-down” effects for employment.  They find strong evidence that CAP effectively reduces drug use and crime, and to a lesser extent, increases employment and family responsibility.  The strength of their findings was supported by two natural experiment studies that analyzed the effects of closing two methadone maintenance treatment programs.

      The TOPS study analyzed the effects of drug abuse treatments on clients who were referred to outpatient drug-free or residential treatment programs under the Treatment Alternatives to Street Crimes (TASC) Act.  The TOPS data indicated that in terms of reductions in drug use and criminal activity, the criminal justice clients do as well or better than other clients in drug abuse treatment (Hubbard et al. 1989).

      After reviewing the relevant literature, Anglin and Hser (1990) concluded clients entering treatment under legal coercion do as well by most outcome criteria as volunteer clients and may stay in treatment longer.  They specifically referred to the evidence from the evaluations of criminal justice civil commitment programs conducted by Anglin and McGlothlin (1984).  Anglin and Hser also reported that research results indicate that correctional drug-treatment programs can have a substantial effect on the behavior of chronic drug-abusing offenders.

      Along these same lines, NIDA (1999) reported that research has shown that combining criminal justice sanctions with drug treatments can be effective in decreasing drug use and related crime.  They also reported that prison-based treatment programs can be effective if patients are separated from the general prison population and if they continue treatment after returning to the community.  Finally, NIDA (1999) concluded that individuals who enter treatment under legal pressure have outcomes as favorable as those who enter treatment voluntarily.

(6)   Short-Term Inpatient (STI) Treatment

      DATOS data showed that patients receiving short-term inpatient treatments have favorable outcomes.  First year follow-ups showed a 69% drop in the number of weekly cocaine users, a 63% reduction in the number of weekly marijuana users, a 58% decline in heavy drinkers, and an insignificant 4.5% increase in full-time work (DATOS 2003e).

6.4.3        The Relative Effectiveness of Treatment Modalities

      Simpson (1984) noted that the DARP outcome differences reported for methadone maintenance, therapeutic communities, and outpatient drug-free treatment programs were not statistically significant.  He noted there were significant differences in the types of clients served and in the dropout rates in the programs.  For instance, methadone maintenance frequently deals with older addicts with longer histories of opioid use and criminal involvement and the treatment strategy is designed to deal with these historical entrenched behavioral patterns.  Tims and Holland (1984) argued that few meaningful differences are likely to be found in outcomes among treatment modalities that are not either a function of client differences or of time in treatment.  They also noted that attempts to randomly assign well-defined homogenous client pools to modalities have been frustrated by clients crossing over to their treatment of choice or withdrawing from treatment (see also Hall 1984).  Finally, Anglin and Hser (1990) also concluded that comparisons between modalities are necessarily restricted because no two modalities necessarily have similar client populations.  We know that outcomes are correlated with patient’s problems and characteristics.

6.4.4        Patient Characteristics and Outcomes

      The drug abuse treatment population is heterogeneous; clients are not characterized by a common set of demographics or problems.  They vary in age, gender, race, social and economic background, drug-dependence, health status, and psychological well being (Hubbard et al. 1989).  They have different criminal and treatment histories.  The literature indicates that some of these factors are highly correlated with post-treatment successful outcomes.  Jaffe (1984), DeLeon (1984) and McLellan, Woody, and Metzger (1996) concluded in their reviews that a stable family background has a positive affect on drug abuse patients’ favorable outcomes.  Anglin and Hser (1990) reported that having an intact marriage also has a positive impact on drug abuse treatment outcomes.  They further concluded that having a job was positively correlated with favorable outcomes.

      Jaffe (1984), Simpson (1984), and DeLeon (1984) reported that drug abuse patients having a more extensive criminal history are likely to report less favorable outcomes.  DeLeon (1984) reported that drug abusers with more extensive history of treatment fared as well as other patients, but Franey and Ashton (2002) reported they fared less well.

      The severity of drug dependence and the extent of drug use are reported to be negatively correlated with drug abuse treatment favorable outcomes (Simpson 1984 and McLellan, Woody, and Metzger (1996).  Drug addicts that use alcohol or poly drug use are likely to have less favorable outcomes as well (Anglin and Hser 1990). 

      Finally, a number of studies have reported the greater the severity of the psychiatric disorder at intake, the less favorable the treatment outcomes for drug abusers (Jaffe 1984, DeLeon 1984, Anglin and Hser 1990,McLellan, Woody, and Metzger 1996, and DATOS 2003d).  McLellan, Woody, and Metzger (1996) added that treating drug abusers having an antisocial personality diagnosis is particularly problematic.

      Given these findings, it is important to gather information on these factors at the intake interview or as soon as possible.  If data is collected on these factors, multivariate statistical analysis can be used to isolate the effects of treatment from these confounding factors.

6.4.5        Length of Treatment and Outcomes

      The most consistent predictor of successful outcome has been length of stay in treatment.  This is true across all drug abuse treatment modalities (DeLeon 1984, Simpson 1984, Anglin and McGlothlin 1984, Senay 1984, Tims and Holland 1984, Hubbard et al. 1989, Anglin and Hser 1990, McLellan, Woody, Metzger 1996, NIDA 1999, Franey and Ashton (2002).  The positive relationship between length of stay in drug abuse treatment and favorable outcomes is interpreted as further evidence that drug abuse treatment is effective.  This is particularly important since treatment evaluations do not use randomly selected “no treatment” control groups.  However, the observed positive relationship between length of treatment and more favorable outcomes is not conclusive proof of the effectiveness of these programs.  It is possible that these results merely reflect the outcomes of the most motivated (thus longer staying and most compliant) patients (McLellen, Woody, and Metzger 1996).  Hubbard et al. (1989) says this may be true in part but several studies with random assignment and evaluations of abrupt closures of methadone programs provide evidence that programs do produce effects independent of client motivation to remain in treatment.  Also, their multivariate analysis supports this interpretation.  Because of the importance of length of stay to favorable outcomes, a number of studies have attempted to model the determinants of retention.  The efforts to date have not been highly successful (DeLeon 1984, Anglin and Hser 1990, NIDA 1990, and Franney and Ashton 2002). 

6.4.6        The Effectiveness of Individual Treatments and Program

            Components

      Having concluded that drug abuse treatments work (i.e., produce favorable outcomes for some patients), researchers have begun to look inside the black box of drug abuse treatment programs to determine why they work or to determine which treatments and program components lead to favorable outcomes. 

(1)    Quality of Staff and Counseling

      Hubbard et al. (1989) report that many clinicians believe that the quality of the staff is an important determinant of the success of a program.  They say there is no clear evidence on the effects of professionally trained counselors versus counselors who are ex-addicts.  Nevertheless, they note there is a clear upward trend in the formal education and credentials of counselors.  In this same regard, Simpson (1984) maintains that successful treatment requires that counselors establish rapport with and influence over their clients.  Anglin and Hser (1990) report that there is weak empirical evidence supporting the hypothesis that the quality of staff has a positive effect on treatment outcomes.  It is surprising that the quality of staff does not have a stronger influence on outcomes because other studies reported that drug counseling services have a significant positive effect on drug abuse treatment outcomes (NIDA 1999 and Franney and Ashton (2002).  Obviously there is more than credentials required for effective counseling.

(2)    Psychotherapy and Related Treatments

      A number of studies have reported that psychotherapy has a favorable effect on drug abuse treatment outcomes (Hall 1984; O’Brieny, Woody and McLellan 1984, Senay 1984, Anglin and Hser 1990, and NIDA 1999).  For example, O’Brieny et al. (1984) reviewed eight controlled studies of psychotherapy with opiate addicts.  In six of the eight studies, the clients improved under the psychotherapy condition as compared to controls.  The NIDA (1999) also referred to later studies that showed that patients receiving supportive-expressive psychotherapy had better outcomes than a control group that only received drug counseling.  Hall (1984) reported that contingency contracting also had a favorable effect on outcomes, but that relaxation therapy and biofeedback therapy did not.

(3)    Vocation, Education and Support Services

      Hall (1984) reported that vocational rehabilitation interventions have a favorable impact on drug abuse treatment outcomes.  He specifically noted that work projects, regardless of their content, decrease dependence on public assistance and increase employment.  He concluded that skill training is a promising modality for enhancing vocational effectiveness.  DeLeon (1984) reported that patients enrolling in college courses showed greater improvement on psychiatric outcome measures.  Finally, Franey and Ashton (2002) noted that providing transportation increased retention rates at non-residential treatment programs, which in turn hopefully improve outcomes.

(4)    Privileges and Sanctions

      Hubbard et al. (1989) cite a number of studies indicating the use of privileges and sanctions is a fundamental method of motivating behavior through the stages of the therapeutic community experience.  They contend that giving small rewards to patients in methadone maintenance programs for engaging activities has improved retention rates and outcomes.  The NIDAC (1999) reported that studies show that patients receiving vouchers for drug-free urine samples achieved significantly more weeks of abstinence and significantly more weeks of sustained abstinence than patients who were given vouchers independent of urinalysis results.  Finally, Anglin and Hser (1990) report the use of take-home privileges as a reinforcement for attendance at counseling sessions has been shown to be successful in some studies.

(5)    Program Policy Decisions

      Programs with flexible policies produce better results than inflexible programs (Senay 1984, Anglin and McGlothlin 1984, and Anglin and Hser 1990).  Specific program policies are also important.  A number of studies have reported that higher doses are required for methadone maintenance treatments to be effective (Anglin and McGlothlin 1984, Hubbard et al. 1989, Anglin and Hser 1990).  There is some disagreement concerning the policy of urine testing on patient outcomes.  Anglin and McGlothlin (1984) and Senay (1984) reported that urine testing had a favorable effect on treatment outcomes.  Hubbard et al. (1989) felt that urinalysis had a small positive effect on outcomes, but it was not clear it was worth the cost.  Anglin and Hser (1990) said the evaluations of urine testing had mixed results but that it was more effective in a criminal setting because of legal coercion.

(6)    Relapse Prevention Therapy and After Care

      As noted earlier, drug-dependency is a chronic illness and the possibility of relapse is always there for successfully treated drug-addicts.  The NIDA (1999) reports that research indicates that the skills individuals learn through relapse prevention therapy remain after the completion of treatment.  Other studies indicate that after care is important for the continuing success of treated drug-abusers (Franey and Ashton 2002).

(7)    New Approaches To Drug Treatment

      The NIDA (1999) reviewed the relevant literature with respect to two relatively new approaches to drug treatment.  It reported that patients treated with the Matrix model demonstrate statistically significant reductions in drug and alcohol use, improvements in psychological indicators, and reduced risky sexual behaviors associated with HIV transmission.  The NIDA also reported that the community reinforcement approach (CRA) plus vouchers approach has been successfully used in outpatient detoxification of opiate-addicted adults and with inner-city methadone maintenance patients who have high rates of intravenous cocaine abuse.

6.4.7        The Matching Hypothesis

      Recognizing the heterogeneous nature of the drug addict population and the fact that modality patient groups are different, a number of researchers have called for a matching of the different types of patients and modalities to achieve the most beneficial results.  Similarly, other researchers have recognized that some groups of patients respond better to some types of treatments and program characteristics than others.  They believe outcomes can be improved by matching patients with the right treatments.  Unfortunately, not much progress has been made in achieving such matches (Anglin and McGlothlin 1984, Simpson, 1984, and Anglin and Hser 1990).  Efforts are currently being made by the DATOS evaluation team to determine the optimal matches between patients, modalities, and treatments.

6.4.8        Relapse and Readmission

      Drug addiction is a chronic disease and many successfully treated addicts relapse and return to treatment.  The life history of drug abusers is often marked by numerous episodes of abuse and treatment (Hubbard et al. 1989).  Simpson (1984) reported that approximately 60% of the DARP follow-up sample of opioid addicts re-entered some type of drug abuse treatment within the first 4 years after leaving DARP.  Readmission rates at the first-year follow-up were about 38%.  Hubbard et al. (1984) reported that a substantial proportion of TOPS clients were in treatment at the first-year follow-up.  Specifically, 60% of methadone maintenance clients, 38% of residential, and 11.5% of outpatient drug-free clients received some treatment in the first year after leaving TOPS treatment programs.  According to DATOS (2003), in the year after discharge from DATOS, 52% of cocaine abusers had relapsed to drug use.

      Hubbard et al. (1989) reviewed the literature on the determinants of relapse and return to treatment.  They reported that sociodemographic factors are not significant.  The number of prior admissions is positively related to the return to drug treatment.  This highlights the chronic nature of the disease.  They noted clients in residential programs and outpatient drug free programs are less likely to return to treatment than outpatient methadone clients.  This probably has more to do with the types of patients at these modalities than the quality of treatments.  Hubbard et al. (1989) reported that time spent in treatment was among the strongest predictors of post-treatment readmission.  They interpret this to indicate that drug abuse treatment is a recurrent, chronic phenomenon.

      Hubbard et al. (1989) suggests that because of the problem of relapse, the issue for treatment programs becomes how to increase effectiveness by maximizing long-term rehabilitation and minimizing the likelihood of relapse and return to treatment.  While this may be true, to date drug abuse treatment evaluation studies have focused on the short-term (6 months to 1 year) effects of single episodes of treatment.  They have generally ignored the issue of readmission in their analysis.  This stands in sharp contrast to other health care evaluations that focus on the lifetime effects of interventions.

6.5      CEA and CBA Evaluations of Drug Abuse Treatments

      As reported in the previous section, research has shown that drug abuse treatment is effective, at least for some groups of drug addicts.  Nevertheless funding is often lacking for treatment because these programs compete for scarce resources with other important and effective social programs (French, 1995, p. 111).  If advocates are going to secure more public funds for drug abuse treatment programs, they need to show these programs are cost-effective.  Policy makers want to know whether resources devoted to treatment yield benefits in excess of treatment costs in the short and long terms (Hser and Anglin 1991, p. 67).  That is, they want to know if a drug abuse treatment program’s benefit/cost ratios exceed the value of one. In the absence of cost/benefit evaluations, policy makers would at least like to know which programs and treatments are the most effective.  For this, they need cost-effectiveness analysis (CEA) of the different programs and treatments.  In addition, CEA evaluations of individual treatments and program components can help practitioners to redesign their programs to improve their overall efficiency. 

      Despite their obvious value, only a limited number of CEA or CBA evaluations of drug-abuse have been conducted.  According to French (1995), research is lacking in this area primarily because the costs and dollar benefits of treatment are difficult to conceptualize and even harder to estimate (p. 112).  Apsler (1991) and Apsler and Harding (1991) present seven reasons for the surprising lack of cost-effectiveness studies of drug abuse treatment.  First, numerous conceptual problems exist due to the absence of satisfactory definitions for key terms such as “drug abuse,” “dependence,” and “addiction.”  Second, strongly held beliefs about treatment goals divide the treatment community.  They range from abstinence to psychiatric improvement.  Third, there are differences in the outcomes selected and how they are defined.  Some studies focus on drug use and criminal behavior; others consider interpersonal relations as well as physical and mental health.  Fourth, programs disagree on the appropriate length of treatment.  This is important because length of treatment has an important effect on observed outcomes.  Fifth, programs vary in terms of the treatments offered (by type and amount) and the types of clients served.  This variability severely limits the generalizations that can be drawn from research conducted on only a few programs.  Sixth, the high dropout rates typical of many drug abuse treatment programs make it difficult for researchers to interpret outcome results.  Finally, the reliance on self-reported outcomes in follow-ups creates a host of problems.  Although difficult and costly to conduct, there is a pressing need for more CEA and CBA evaluations of drug abuse treatments if we are to make progress in this area.  Before we review the literature on CEA and CBA evaluations, it is useful to consider the literature on the costs and benefits of drug abuse treatments.

6.5.1 The Costs of Drug Abuse Treatments

      As noted in Section 3.5, there are three main categories of drug-abuse treatment costs:  health care sector costs (c₁), patient and family costs (c₂), and other sector costs (c₃).  C₁ includes the cost of organizing and operating the program as well as fixed and overhead costs.  C₂ includes transportation expenses, lost work time, donated time and equipment, and care taking.  C₃ refers to costs borne externally to the treatment sector, patients, and their families (French 1995).

       In theory, the “true” costs of treatment are the opportunity cost value of resources used or lost.  Most evaluations use monetary costs recorded in accounting data.  They tend to ignore the indirect costs (i.e., those not recorded) such as those included in C₂ and C₃.  Such studies understate the “true” costs of treatment.

      Most evaluations simply report cost figures for the whole program.  Accounting data on income and expenditures are usually obtained for a period (i.e., a year), and the total number of treated patients is divided into total program to estimate an average cost per patient or “slot” costs (Cartwright 1998).  Unfortunately, the average cost or “slot” cost is a poor approximation for actual episodes of treatment and for a measure of services used by patients (Richman 1983 and Cartwright 1998).  Most programs tailor treatment to the needs of each patient, and most patients use treatment resources to different degrees.  Some patients leave treatment after a few weeks or less, whereas others stay through the period (e.g. 12 months)  during which costs are assessed.  For these reasons, costs vary for patients.  Our review of the effectiveness literature revealed that favorable outcomes are positively related to the time in treatment, which is to be expected.

      Individual behavioral changes and outcomes vary with the treatments received.  Some patients change with a small amount of treatment; others change only a little with a lot of treatment.  To analyze cost-effectiveness and cost-benefit accurately, costs as well as effectiveness, and benefit must be measured separately for each patient (NIDA 1999).  This is almost never done in the published literature.

      To increase our knowledge of cost-effectiveness, we must begin to link the costs and outcomes for individual patients.  We need to learn what treatments are cost-effective for each type of patient so that we can redesign our programs and direct patients to the most beneficial treatments.  Unfortunately, we are only beginning to explore empirically the implications of the “MATCHING” hypothesis (Hubbard and French 1991).  Recently, the National Institute of Drug Abuse issued a manual to help researchers and practitioners systematically explore the link between individual treatments, program components and outcomes at the individual unit (NIDA 1999).

      The relevant costs to be considered in any evaluation depend on the viewpoint of the analyst (see Table 3.1).  For example, the cost of travel to and from a clinic is a legitimate cost from a client’s perspective, but it is not a cost for the treatment provider.  Medicaid payments for treatment are a cost for the paying government (i.e., taxpayer), a gain to the patient, and neither a cost nor a gain but a transfer to society (French 1995).  In some treatment programs, patients cannot hold employment, whereas in others they can.  An evaluation conducted from the client’s viewpoint would include foregone income as a treatment cost so they might conclude the latter programs are more cost-effective than the former.  From the treatment provider’s view, the patient’s lost income is not a cost so they might conclude the former programs are more cost-effective than the latter.

      There are few published studies on the cost of drug abuse treatments.  Several studies, based on crude estimates of direct costs from the provider’s view have concluded that the “slot” cost of residential treatment is about three times the cost of outpatient methadone or outpatient drug-free treatment (Harwood et al. 1988 and Wallack 1990).  In the few studies that have estimated the costs of drug abuse treatment, the methods and results are not consistent (Bradley 1996 and NIDA 1999).  Recently, Anderson et al. (1998) provided detailed cost statistics for a variety of treatment procedures at several levels of specificity.  They show that more than 97 different services in a treatment bundle may be identified and that the estimated costs of treatment vary with the specified bundle.  The huge potential variations in the estimated costs of treatment make it almost impossible to measure the relative cost-effectiveness of treatments and programs.  NIDA (1999) has called for a standardized method of measuring costs so that CEA and CBA evaluations of treatments and programs can be compared in a meaningful way.  There are similar problems when it comes to identifying the outcomes and benefits of drug abuse treatments.

6.5.2 Drug Abuse Treatment Outcomes and Benefits

      We identified the nature and the problems involved in measuring drug abuse treatment outcomes in Section 6.4.  Here we should note that some of these outcomes are measured objectively and some are not.  Biological measures such as urine or blood tests are sometimes used.  Biological measures of infections are used in the case of health care outcomes such as AIDS or TB.  Arrests or convictions can be used to measure criminal activity.  Most programs, however, rely on the subjective self-reported measures obtained in follow-up interviews.  We generally have more confidence in the validity of the objective measures than the subjective measures.  As noted earlier, the variations in the methods of estimating costs and outcomes makes it impossible to determine the relative cost-effectiveness of different programs.  It is likely that programs which offer specialized treatments such as vocational and jobs skills or psychotherapy will have better cost-effectiveness scores for employment outcomes and psychological functioning than programs which do not offer these treatments.

      Outcome measures have to be translated into dollar values if the researcher wants to conduct cost/benefit analysis.  Section 3.6 identified the major benefits associated with health care interventions such as drug abuse treatments.  If treatments cause abstinence or reduced drug use, the individual may enjoy immediate health benefits and reduced risk of future ill health and premature death.  Treatments may improve social or emotional functioning of individuals as well as their health.  This, in turn, could improve marital and family relations.  All in all, drug-abuse treatments can change the quality of life of patients.  The individual may also benefit from improved future employment and higher earnings, which would raise government’s tax revenue.

      Other parties may benefit from successful drug-abuse treatments through avoided costs from continued drug abuse (French 1995).  These avoided costs (i.e., cost-savings) are:  (1) reduced future health care costs, (2) reduced criminal activity and gains for crime victims, (3) reduced criminal justice costs, and (4) reduced social care costs.  Employers and governments may also benefit from increased productivity and enhanced tax revenue.

      Some of these benefits are more easily measured than others.  Direct benefits such as treatment and health care expenditures are relatively easy to measure.  Indirect benefits such as quality-of-life improvements, increase in productive activities, and reduced criminal activities are more difficult to measure and value (French 1995).  The data and the methods necessary to estimate the full range of avoided costs from drug-abuse treatments are described in French et al. (1996).  They organized these methods into three groups.

(1)   Cost-of-illness method calculates the dollar cost of drug abuse as the sum of medical resources used to diagnose and treat the disorder and the value of lost productivity due to morbidity and mortality.

(2)   Averting behavior models quantify the cost of actions individuals take (e.g., participating in Narcotics Anonymous, purchasing security devices, altering social activities, changing residence) to mitigate the adverse consequences of drug abuse.

(3)   Utility valuation models are designed to estimate the intangible costs (e.g., pain and suffering, family disruption) of drug abuse.  To date, no study has undertaken the task of estimating all of the benefits suggested by French et al. (1996).

      Many of the benefits reported are based on the monetized values of subjective outcomes self-reported in follow-up studies.  For example, self-reports on criminal activity are used to estimate the decrease in the number of crimes during a post-treatment period.  This figure is then multiplied by national figures on crime costs and incarceration costs, and criminal justice system costs to obtain a dollar gain in crime-cost savings.  If the self-reported decrease in criminal activity is not accurate, then the estimated value of crime cost-savings also is not accurate.

      On the other hand, objective monetary benefits of substance abuse treatment can be obtained from the following sources (NIDA 1999).

·        Financial records from accountants, funders, and tax agencies of legal employment after, versus before, substance abuse treatment.

·        Records of welfare benefits paid during and after, versus before, substance abuse.

·        Records of public health services used during and after, versus before, substance treatment.

·        Records of funds spent on arrests, convictions, and other interactions of the patient criminal justice system during and after, versus before, substance abuse treatment.

      As just seen, treatment outcomes and benefits can be measures in a number of different ways and they vary from study to study.  If comparable estimates of the cost-effectiveness and cost/benefits of treatments are to be made, a standard methodology for calculating costs and measuring treatment benefits in economic terms must be developed (French 1995).

6.5.3 Cost-Effectiveness Analysis (CEA) of Drug Abuse Treatments

      Cost-effectiveness analysis considers both cost and outcome by forming and comparing ratios of cost and effectiveness, both of which are difficult to identify and measure.  Costs are measured as dollars spent, whereas effectiveness or outcome is measured as changes in patients’ behaviors (e.g., drug use or criminal activity), thoughts, feelings, or biology (NIDA 1991).  Cost-effectiveness analysis is appealing because it considers the possibility of improved outcomes in exchange for using more resources.

      A single program cost-effectiveness ratio means little by itself.  The cost-effectiveness of a particular strategy needs to be compared with the cost-effectiveness of alternative strategies that would use same resources (see chapter 2 or French 1995).  If a number of alternative treatment programs calculated cost-effectiveness ratios in the same way, we could determine which program was cost-effective assuming the clients are randomly drawn from the same population.  If treatments, costs of treatment and outcomes were followed on an individual basis, cost-effectiveness ratios could be calculated for each patient in a program and averaged to describe the typical cost-effectiveness of treatment (NIDA 1999).

      Some evaluations of treatment outcomes exclude patients who have dropped out of treatment early on because they feel that these patients have not received the minimum amount of treatment to change their behavior.  This is not proper.  The patients who dropped out consumed resources during intake and during treatments.  It would be inaccurate to distribute those costs across patients who stay in treatment.  When costs are examined, that approach would penalize programs that have higher dropout rates (NIDA 1999).  All patient’s costs and outcomes should be included in the evaluation.

      For meaningful comparisons of cost-effectiveness to be made, they must be calculated the same way.  But there is no single standard for “cost-effective.”  Both costs and outcomes are measured in many different ways.  The National Institute of Drug Abuse recently recommended the use of cost per drug-free day as a preferred cost-effectiveness ratio.  The NIDA (1999) claims cost per drug-free day suggests  a standard metric that will be better (lower-cost per drug-free day) if either (a) less money is spent per patient, (b) more patients are free of drugs for a day, or (c) an individual patient is free of drugs for more days.

      The relevant costs and outcomes to include in a CEA depend on the analyst’s perspective as always.  As discussed in Chapter 2, a CEA can be conducted from the client, program manager, taxpayer, or society point of view.

      New efforts are being made to link treatment costs and outcomes at the patient level so that cost-effectiveness ratios for each treatment for each individual can be calculated (NIDA 1999).  If this is eventually accomplished on a widespread basis, we can begin to learn how treatments work and what treatments should be provided to each type of patient.  This will take at least a decade.

6.5.4 Cost/Benefit Analysis of Drug Abuse Treatments

      Although cost-effectiveness analysis (CEA) is useful to decision makers, it does not explicitly assess whether the outcomes are worth the cost; it merely compares the costs of achieving the same outcome.  Cost/benefit analysis (CBA) forces an explicit comparison between costs and benefits by measuring both in dollars (French 1995).  A program is efficient only if the dollar benefits exceed the dollar costs.  CBA can be applied to a single program or modality, but it can also be used to compare programs.

      There are several ways to report the relationship between costs and benefits (NIDA 1999):

·        The net benefit of a program can be shown by subtracting the costs of a program from its benefits.  For example, if a substance abuse treatment program cost $100,000 per year but generated in the same year $500,000 in increased patient income, increased tax payments by patients, and reduced expenditures for social and criminal justice services, the net benefit of the program would be $500,000 minus $100,000, or $400,000, for that year.

·        The ratio of benefits to costs is found by dividing total program benefits by total program costs.  For example, dividing the $500,000 benefit of the program by its $100,000 costs yields a cost-benefit ratio of 5:1.

·        Because neither net benefits nor cost-benefit ratios indicate the size of the cost (initial investment) required for treatment to yield the observed benefits, it is important to report this as well.  We cannot assume that the same exact relationships between costs and benefits will exist at different levels of investment.  Sometimes an increase in cost allows new, more productive procedures to be used for treatment, increasing benefits dramatically.  For example, increasing a program budget to allow hiring of a community liaison, vocational counselor, or physician might dramatically increase patient outcome.  Therefore, it often is best to report the initial investment, the net benefit, and the cost-benefit ratio.

·        The time to return on investment (the time it takes for program benefits to equal program costs) is yet another indicator used in cost-benefit analysis.  For programs, benefits and costs occur at the same time, or at least in the same year.  For individual patients, however, the investment in treatment may pay off substantially only after several months or years.  Costs usually occur up front, but program benefits may take time to reach the point where they exceed costs.

      The benefits of drug abuse treatment, as noted above, accrue primarily through avoided costs from continued drug abuse.  In effect, economists calculate the avoided costs resulting from positive treatment and count these as the dollar benefits of drug abuse treatments (French 1995).  The major benefits of drug abuse treatment programs are indirect such as reduction in crime-related costs, including property losses, medical services required by victims, time taken off from work by victims, and criminal justice costs of apprehending, trying, and incarcerating offenders (NIDA 1999).

      The most common approach for calculating benefit/costs is to establish “critical periods” of interest, determine behavior of clients during these periods, and assign cost values.  Most studies consider three periods:  pre- post- and after-treatment periods.  This approach corresponds to a level of analysis at the individual addiction career.  Hser and Anglin (1991) use Figure 6.1 to illustrate how cost-benefit analysis can be accomplished within this perspective.  Time-related cost imposed on society by an untreated drug user are shown by the height of the top line.  This line slopes down to the right based on the assumption that some users “naturally” recover over time even if they receive no treatment.  Many of the empirical CBA studies ignore this “natural” or “spontaneous” recovery effect.  Lifetime costs are indicated by the area under  the curve.  The treatment profile indicates that during treatment, social costs are positive but lower than without treatment.  After treatment, costs rise as relapse and other negative behaviors occur, but costs typically do not attain the level they would have without intervention.  The gross gains from treatment are measured by the difference between the total lifetime social costs of an untreated drug user less those costs for one who is treated.  The costs of treatment  are  indicated  by  the  area  of  the  shaded  rectangle,  and the net benefit from

Figure 6.1  Cost-Benefit Model

treatment is equal to the gross gains less the treatment program expense.  Although treatment may not be 100 percent successful at curbing drug use and the associated social costs, it can achieve a substantial savings.  Notice this model may be applied to all social costs or to any component of social cost such as criminal cost-savings or health care cost-savings, etc.

6.5.5 Individual Evaluations of Drug Abuse Treatments

      Because of the difficulties involved, only a relatively small number of studies have attempted to evaluate the cost-effectiveness of drug abuse treatments.  We shall review ten major published studies in chronological order.

(1)   Hser and Anglin (1991) reported on two earlier studies conducted by McGlothlin and Anglin in 1981.  These studies examined the effects of involuntary termination of  methadone maintenance programs in California.  They found that 55% of the terminated client group became re-addicted to heroin compared with a 31% rate for a control group who remained in treatment.  The arrest and incarceration rates were approximately double for the terminated group relative to those for the control group.  McGlothlin and Anglin concluded that the social cost increase more than offset the cost of eliminating the programs without taking into effect the detrimental effects experienced by the clients.  This natural experience suggests that methadone maintenance programs produce net benefits (i.e., B/C > 1.0).

(2)   Tabbush (1986) examined the cost-effectiveness of publicly-funded drug abuse treatment programs in California.  He analyzed drug use and costs one year before admission to treatment and one year after treatment.  His analysis included social costs (drug-related crime and crime enforcement activities, as well as publicly-borne medical costs due to morbidity)  and private costs (labor productivity and reduced life span).  Residential drug-free treatment for heroin addicts produced the highest benefit-cost ratio (26.3).  The benefit-cost ratio for outpatient drug-free was also high (24.7).  The benefit-cost ratio for methadone maintenance was 13.8.  Even detoxification programs produced ratios of 9.7 for residential and 7.4 for outpatient.  As French (1995) has noted these estimates are upwardly biased.  First, the study included no control group and covered a short pre- and post-period (1 year) so the “ramp-up” effect and “regression-to-the-mean problem and “spontaneous” recovery problems are strong.  Second, the validity of the results depends heavily on unverified self-reports of illicit drug use and criminal behavior.  Third, he frequently double-counted costs such as income used to purchase drugs and the value of stolen property.  Finally, key measures of impact such as costs to victims, criminal justice system costs, medical care costs, and lost productivity were estimated from national surveys.

(3)   Hubbard et al. (1989) using TOPS data compared the costs of drug abuse treatment with the dollar benefits (avoided costs) of lower criminal activity.  Components of the social cost of drug-related crime included victim costs, criminal justice system costs, and crime-career productivity costs.  They compared the average cost of treatment per episode with the reduction in crime-related costs in the year following treatment.  The benefit/cost ratio was calculated for two impact categories and three modalities as shown in the table below.

Crime Reduction Benefit/Cost Ratios

According to French (1995), this study suffers from the same methodological problems as the Tabbush (1986) study.

(4)   Anglin, Speckart, Booth, and Ryan (1989)

Once again Anglin and colleagues looked at the costs and benefits of closing several publicly funded methadone clinics in California.  They used a two-and-a-half year follow-up and an adequate control group.  They reported major adverse consequences for clients unable or unwilling to transfer to private treatment programs including higher crime and dealing rates, more contact with the criminal justice system, and higher rates of illicit drug use compared to the clients that did transfer.  They concluded “the savings resulting from a reduction of MM program costs were nearly offset by increased direct costs for incarceration, legal supervision and other government funded drug treatment.  Indirect costs were not assessed” (p. 307).  These results suggest a B/C = 1.0 from the taxpayers point of view.

(5)   Deschenes et al. (1991) reviewed the natural history of narcotics addiction among a sample of Chicano and White methadone maintenance patients in Southern California, interviewed between 1978 and 1980.  They examined the extent of criminal behavior over a 15-year period and assessed the social costs of criminology and narcotics addiction.  Analysis indicated that criminal activity and related social and economic costs were at their highest during periods of addiction.  Overall, the average social cost per addict per year was $20,000.  This estimate shows the potential for effective drug abuse prevention and treatment to result in very high benefits (French 1995).

(6)   French et al. (1990) and French and Zarkin (1992) found that drug treatment improved the employment and earnings of drug abusers.  Following treatment the percent of drug abusers employed rose from 31 percent in the pre-treatment period to almost 45 percent.  Similarly, average personal earnings for employed patients rose from $6,158 during the year before treatment to $7,120 during the year after treatment.  French and Zarkin (1992) showed that increasing time spent in methadone treatment by just 10-percent increases legal earnings by 1.5 percent and decreases illegal earnings by 3.2 percent.  A 10-percent increase in time spent in residential programs increases legal earnings 2.4 percent and decreases illegal earnings 4.1 percent.  They computed the dollar benefit of additional time in treatment that accrue through higher income from a legitimate job or business and lower illegal earnings and compared it with the average annual cost of treatment.  As French (1995) notes, their evidence shows that treatment cannot be justified financially looking only at employment outcomes, but factoring in other treatment outcomes would increase the benefit-cost ratio.

(7)   French et al. (1996) proposed a theoretically rigorous methodology based on the quality-adjusted life year (QALY) concept (see Chapter 2) for estimating the health-related costs of drug abuse.  The methodology is used to estimate the potential dollar value of avoiding adverse health consequences as a result of successful drug-abuse interventions.  Specifically the authors estimate the dollar value of avoiding a number of diseases known to be related to drug abuse for a white male aged 32 years.  Just the morbidity costs for a severe case are estimated to be:  acute Hepatitis B ($7,469), Symptomatic AIDS ($114,798), hypertension ($74,513), Bacterial pneumonia ($57,528), Sexually transmitted diseases ($1,100), and Tuberculosis ($96,005).  To calculate the full cost of these diseases one needs to add the utility loss from premature death caused by these diseases.  These estimated values represent the potential benefits of successful drug-abuse interventions.

(8)   Rajkumar and French (1997) developed a method to estimate the tangible (e.g., criminal justice system) and intangible (e.g., pain and suffering of crime victims) benefits of avoided crime that can result from successful substance abuse intervention.  They demonstrate that the intangible benefits can be quite large, even though some areas are not investigated.  This finding is important because most evaluation studies ignore intangible benefits because they are difficult to calculate.  They used their method to estimate the pre- and post-treatment costs of criminal activity for a sample of 2420 drug abusers.  They concluded the plausible range of estimates for cost reduction when intangible costs are included is $10,918 to $23,234 per client compared to $7193 per client if only tangible costs are computed.  Using the lower estimate ($10,918) and the average annual “slot” cost of outpatient methadone ($2828), residential ($8,920), and outpatient drug-free ($2908) treatment at TOPS programs (Hubbard et al. 1989), they calculated the following benefit/cost ratios:  outpatient methadone (3.86), residential (1.22), and outpatient drug-free (3.75).  This led them to conclude “despite the limitations with the crude comparison of costs and benefits, drug abuse treatment clearly has the potential to return significant net benefit to society in the form of avoided criminal activity” (p. 323).  Because this study had no control group and used a short pre-period (one year) and short follow-up period (one year), its estimates of benefits and benefit/cost ratios are upwardly biased because of the “ramp up” and “regression-to-the-mean” effect and the likelihood of “spontaneous” recovery.  In addition, the use of self-reported criminal activity is also problematic.

(9)   Barnet and Swindle (1997) used a program administrator’s survey and cost and discharge data for 38,863 patients treated in 98 Veterans Affairs treatment programs to identify the characteristics of cost-effective inpatient substance abuse treatment programs.  A treatment was defined as successful if the patient was not readmitted for psychiatric or substance abuse care within six months after treatment.  They used random-effects regression to find the effect of program and patient characteristics on cost and readmission rates.   Treatment was more expensive when the program was smaller or had a longer intended length of treatment or a higher ratio of staff to patients.  Readmission was less likely when the program was smaller or had longer intended length of treatment.  The staff to patient ratio had no significant effect on readmissions.  They also reported that patients with a history of prior treatment were more likely to be readmitted but their subsequent stays were less costly.  The average treatment cost was $3,754 with a 75.0% chance of being effective, which implied a cost-effectiveness ratio of $5,007 per treatment success.  Finally, a 28-day treatment program was $860 more costly and 3.3% more effective than a 21-day program, an incremental cost-effectiveness of $26,450 per treatment success.  On the basis of these findings they concluded a 21-day limit should be placed on treatment programs.  Consolidation of small programs would reduce cost, but would also reduce access to treatment.  Lastly, a reduction in the staff to patient ratio would increase the cost-effectiveness of the most intensively staffed programs.

(10)    Barrett (1999) uses life-years of survival as the measure of treatment success.  This

           measure  is  widely  used  in  the economic evaluations of health care interventions

but has not previously been applied to substance abuse treatments.  He applied this concept to calculate the cost-effectiveness of methadone maintenance treatments.  He combined literature estimates of the effect of methadone treatment on the rate of mortality associated with opiate addiction with data on the average cost and duration of treatment.  Barrett used a two-state Markov model to estimate the incremental effect of methadone on the life span and treatment of a cohort of 25-year old heroin users.  He found that providing opiate addicts with access to methadone maintenance has an incremental cost-effectiveness ratio of $5,915 per life-year gained. Barrett says this is well below the $50,000 per quality-adjusted life year of survival, which is considered sufficiently cost-effective to justify intervention adoptions by the U.S. health care system.  Barnett’s analysis shows that standard methods of cost-effectiveness developed for medical care interventions can also be applied to substance abuse treatment.

(11) Flynn (1999) examined the costs and benefits of treatment for cocaine addiction in

long-term residential (LTR) and outpatient drug-free (ODF) programs.  Before, during, and after treatment interviews were conducted with a DATOS national sample of 502 clients in 10 U.S. cities.  Overall, reductions in costs of crime to society during and after treatment substantially surpass the cost of treatment in both LTR and ODF.  The average net economic benefit differed between the two modalities.  The average net economic benefit from an episode of LTR treatment was $10,344.  Net benefits from an episode of ODF treatment was only $795.  However, the types of clients and the costs of treatment also were different in each modality.  Flynn (1999) used Rajkumar and French’s (1997) crime costs (tangible and intangible) to estimate the cost of crime to society, net benefits, and cost-benefit ratios.  The highest crime group was LTR with per client crime costs of $20,743 before and $4,605 after treatment.  ODF client crime costs were $3,494 before and $2,503 after treatment.  Because the costs of treatment were much higher in LTR, Flynn et al. reported the ratios of total benefits to cost of treatment were 1.94 for LTR and 1.56 for ODF.  These estimates are likely biased upward because the authors had no control group and used short (1 year) pre- and post-treatment periods.  “Ramp-up” and “regression-to-the-mean” effects and “spontaneous” recovery effects are likely to be included in the estimated treatment effects on crime reduction.

6.6  Recent Evaluations of State Drug-Abuse Treatment Programs

      There have been several recent evaluations of State drug-abuse treatment programs that are particularly germane to this study. 

(1)   Gerstein et al. (1994) evaluated recovery services under the California Drug and Alcohol Treatment Assessment (CALDATA).  The purpose of CALDATA was to study:  (1) the effects of treatment on participant behavior; (2) the costs of treatment; and (3) the economic value of treatment to society.  The effects of treatment are measured by the differences in behavior and experience reported by clients in the year before and 15 months after treatment.  The costs of treatment were calculated form the financial records of the providers.  The economic value of treatment was based largely on the costs avoided due to reductions in the burden of crime and illness, as well as a careful review of shifts in income sources.  CALDATA drew random samples of clients in four types of treatment programs in California: 

·        Residential programs

·        Residential social model programs

·        Outpatient programs

·        Outpatient methadone

      The economic benefits of treatment were calculated two ways:  benefits to taxpaying citizens and benefits to the total society.  The major difference is that taxpaying citizens benefit when there is less theft and other crime and when the state makes fewer drug-related disability payments and other welfare-type transfers.  However, these transfers of income and property are considered economically neutral to the total society, since one person’s loss equals another’s gain.

                  Their analysis revealed the following general effects of treatment.

·        The level of criminal activity declined by 67%.

·        Alcohol/Drug use declined by 40%

·        Health Care hospitalizations declined 33%.

·        Treatment did not have a positive effect on employment, but the lengths of stay in treatment did have a positive effect on employment.

·        Disability and Medi-Cal spending increased from 16% to 50%.  They reported that treatment increased the eligibility to receive disability payments.

      The overall average cost per episode of treatment per client was $1,425, but it varied by modality:  Residential ($4,405); Social Model ($2,712); Outpatient ($990); and methadone discharge ($404).  The four modalities also treated different types of clients.  The per person economic costs for the year before treatment to society were:  Residential ($35,845), Social Model ($34,667); Outpatient ($23,154), Methadone Discharged ($38,401); and Methadone Continuing ($33,625).

       The benefit/cost ratios to taxpaying citizens and to society for the five types of treatments are shown in the table below.

This study received national attention with its conclusion that for every dollar spent in treatment, seven dollars were saved in avoided costs to taxpayers in the first year after treatment.

      This estimate and the benefit-cost ratios reported in the table are undoubtedly upward biased for the following reasons.  The study used a pre-post design and had no comparison group.  The observation time period was brief:  pre-period one year and post-period 15 months.  Therefore the treatment effects include the “ramp-up” and “regression-to-the-mean” effects and the “spontaneous” recovery effects.  On the other hand, the benefit measures are not comprehensive; they present only crime, health, or productivity but seldom all three.  Finally, the benefits are based on subjective self-reported data from clients who were required to remember the occurrence of events up to 36 months after they occurred.

(2)   Mauser et al. (1994) evaluates the economic impact of Wisconsin’s Treatment Alternative Program (TAP) modeled after the national Treatment Alternatives to Street Crime (TASC) program.  The study used the pre/post research design with no control group.  The pre-period was one year and the follow-up period was 20 months.  There was a problem with the follow-up sample.  Of the 259 clients admitted into TAR, 112 (or 43%) agreed to participate in the outcome study.  Only 25 clients completed the follow-up interview on which the benefit-cost estimates are based.  Mauser et al. reported that the benefit-cost ratios, which only included benefits in terms of reductions in criminal justice system costs, was 1.80 for all outcome study participants (N=76) and 2.58 for the 25 participants who completed the follow-up interview.  When benefits from changes in health care use and changes in productivity effects are added, the benefit cost ratios decline to 1.41 for the 25 participants who completed the follow-up interview.  The small follow-up sample raises doubts about the study.  Also, with no control group and relatively short pre- and post-periods, we worry about “ramp-up” and “regression-to-the-mean” problems and the “spontaneous” recovery problem.

(3)   Finigan (1996) examined societal outcomes and cost savings of drug and alcohol treatment in the State of Oregon.  This study was designed to overcome some of the methodological limitations of past studies of the benefits and costs of drug and/or alcohol treatment.  The study used a matched (in terms of age, gender, race, drug use, and severity of drug abuse) control group of clients who received little or no treatment for comparisons with the experimental group of treatment completers.  Second, the study used a two-year prior period and a three-year post-period.  These two aspects should significantly reduce the problems of the “ramp-up” and “regression-to-the-mean” problem and the “spontaneous” recovery effects.  Third, the study relied on the use of existing state agency data bases rather than self-report data for maximum objectivity. 

           The study was based on a random sample of 1125 clients for fiscal year 1991-1992.  Three service modalities were reviewed:  outpatient, residential, and methadone.  The study considered societal outcomes and cost savings in arrests and convictions, incarceration, employment and earnings, food stamp assistance, and medical costs.

      Finigan’s analysis revealed the following treatment effects for completers compared to the control group in the pre/post periods.

·        Arrests were 33% lower and convictions were 34% lower for treatment completers

·        Incarceration rates were 70% lower for treatment completers

·        Treatment completers received 65% higher wages

·        Treatment completers had 67% less use of food stamps

·        Open child welfare cases were 50% lower for completers

·        Medical expenses were substantially lower for completers.  For example, treatment completers had 53% fewer emergency room visits.

He also reported that the effects of treatments varied quite substantially between the three modalities.  However, he made no attempt to adjust for the differences in the types of patients served, so it is not clear what the “true” effects of treatment are at these modalities.

      He estimated that the total amount of benefits (i.e., avoided costs) to taxpayers for the two and a half year period were $83,147,187.  This can be broken down into the following categories:  criminal justice ($21,222,945), public assistance ($3,222,963), victim ($23,480,512), and theft ($35,220,767).  The estimated cost of providing treatment to the 1991-1992 treatment completers was $14,879,128.  Thus, the estimated benefit-cost ratio to taxpayer citizens is 5.60 (=$83,147,187 ¸ $14,879,128).  On this basis, Finigan concludes that every dollar taxpayers spent on those who completed treatment in 1991-1992 produced $5.60 of avoided costs savings to the taxpayer.  He maintains this is a conservative estimate because additional savings presumably accrued from those clients who received a good amount of treatment but who did not complete treatment.

      The cost-savings to government are much smaller because they do not include victim and theft costs.  The costs avoided by state and local governments are approximately $24,450,000.  From governments’ point of view the benefit-cost ratio is 1.64 (=$24,450,000 ¸ $14,879,128).

      Although methodologically stronger than earlier studies, Finigan’s estimate benefit-cost ratios are still probably upwardly biased.  This is because treatment completers and early dropouts are likely to differ in important ways.  The motivations and desires to change their behavior probably differ.  If completers are more highly motivated than dropouts, this effect is included in the estimated treatment effects and biases them upward to an unknown extent.   

(4)   Wickizer and Longhi (1997) assessed the economic benefits and costs associated with substance abuse treatments provided to indigent clients through Washington State’s Alcoholism and Drug Addiction Treatment and Support Act (ADATSA) Program.  This study utilized a quasi-experimental design.  The experimental group included clients who completed treatments or completed the primary phase of treatments. The control group consisted of clients who were judged eligible for treatment, but who either did not begin treatment or who failed to complete the initial phase of treatment.  

      Cost savings were analyzed for a 12-month period following treatment.  Medicaid costs included all inpatient and outpatient medical expenses, exclusive of Medicaid payments made for substance abuse treatments.  Public assistance costs included all welfare payments made to clients during the 12-month follow-up period, including AFDC payments as well as payment for general assistance unemployment (GAU).  Chemical treatment reentry costs represented expenditures for the following services:  detox, inpatient treatment, and outpatient treatment.

      The study used multiple regression analysis to estimate the cost savings of treatment.  This statistical technique allowed the analysis to estimate the effect of treatment on cost outcomes, controlling for the effects of age, age at first use, severity of dependence, mental health problems, gender, education, ethnic group and prior admissions.  In this way, it was possible to derive estimates for both the treated and comparison groups of adjusted costs, e.g., adjusted Medicaid costs or adjusted public assistance payments.  The difference between adjusted costs for the treatment and comparison groups provides an estimate of the cost savings.

      For the average treated client, the estimated savings for the three cost outcome areas were as follows:  Medicaid ($647), Public Assistance (-$89), and treatment reentry ($134).  Therefore the overall cost-savings was $692 per treated client.  For the average client, the (incremental) cost of treatment was $1,779 compared to a benefit of $692, so that the benefit-cost ratio is 0.38.  This implies that for every $1 invested in the average client’s treatment, 38 cents was recouped in the form of reduced Medicaid, public assistance and treatment reentry costs during the 12-month period following treatment.

      Wickizer and Longhi’s relative small benefit-cost ratio reflects their methodology, particularly the use of multiple regression analysis, and the omission of criminal justice cost savings which are generally quite substantial.  Also, their analysis is limited to indigent clients.      

(5)   Estee and Nordlund (2001) examined the medical cost savings from providing chemical dependency (alcohol and drug abuse) treatments to recipients of Supplemental Security Income (SSI) in the State of Washington.  They used a quasi-experimental research design.  The experimental group consisted of those who received treatment and the control group consisted of a sample that needed treatment but did not get it.

      The average monthly medical and AOD treatment costs were $540 lower for those who got treatment than Medicaid costs for those who appeared to need treatment but did not get it.  These differences were based on regression equations that controlled for the effect of age, race, gender, and prior monthly medical costs.  The average annual medical cost-savings was $6,480 per client.  Since the annual cost of treatment was $2,956 per client the estimated benefit-cost ratio is 2.19.  The authors reported that the cost-savings were significantly greater for the following subgroups of SSI clients:  (1) those aged 45 years or older ($931), (2) those not arrested for drug-or-alcohol-related offenses ($739), and those who only received Medicaid ($704).  Because of the research design and use of multiple regression analysis, we have more confidence in their estimated benefit-cost ratio than many of the other studies.

      These studies highlight the importance of methodology, selection of potential cost-saving outcomes, inclusion of modalities and clients, and analysts’ perspective in the calculation of benefit-cost ratios for drug abuse treatments.  Estimated benefit-cost ratios range from 26.3 to 0.38.  There is a clear need for developing a standard methodology for conducting benefit-cost studies of drug abuse so that researchers can generate consistent and comparable estimates for policy purposes (French 1995 and NIDA 1999).  It will take years before this happens.  In the interim, greater attention should be accorded the well-designed studies that include a control group, longer pre- and post-periods, and use multivariate techniques that control for the effects of confounding factors.

6.7  Summary and Conclusions on Drug Abuse Treatments

      We learned earlier that drug-dependency imposes very high costs on state governments and society.  This chapter surveys the literature on evaluations of drug abuse treatments to determine the amount such treatments can reduce the societal costs of drug-dependency.  The evaluation of drug abuse treatment is very complex because of the multitude of treatments provided, the variation in the treatment population, and the multiple goals and outcomes that programs pursue.

      Historically, there were five major treatment modalities:  (1) methadone maintenance, (2) outpatient drug-free, (3) long-term residential or therapeutic community (TC), (4) short-term residential, and (5) detoxification programs.   Many new programs have evolved over time.  Treatment programs are diverse; they offer a wide array of all or some of the following treatments:  drug counseling, drug education, pharmacotherapy interventions, psychotherapy services, vocational and skills training, urine testing, relapse prevention training, and social and community support services.

      The same methodological problems that plagued evaluations of alcoholism treatments make it difficult to evaluate studies of drug abuse treatments.  These include:  (1) the lack of standardized treatment protocols, (2) the lack of standard outcome measures, (3) the lack of control for patient variation, (4) the failure to report or lack of consistency in reporting treatment costs, (5) the failure to use the “ideal” research design where alcoholics are randomly assigned to experimental treatment groups and to a no-treatment control group, (6) the so-called “ramp-up effect” just prior to treatments leads to a potential “regression-to-the-mean problem,” (7) follow-up analysis is plagued by censored samples, reliance on self-report data and varying post-treatment periods most of which are too short to capture the long run effects of treatment and to minimize the effects of “regression-to-the-mean” and “spontaneous” recovery, (8) the high relapse rate among successfully treated alcoholics makes it difficult to measure both the short-run and long-run effects of single episodes of treatment, (9) there is no way to separate the effects of spontaneous recovery from treatment effects in the absence of the “ideal” control group, and (10) some studies use multiple regression analysis to separate treatment effects from confounding factors, but most do not.

      The literature on the effectiveness of drug abuse treatments has expanded rapidly over the past 30 years.  Three large national studies have been subsidized by the NIDA:  DARP, TOPS, and DATOS.  The evaluation studies of treatment effectiveness flowing from these national projects are methodologically stronger than earlier studies in terms of sample size, research design, and statistical analysis, but they still suffer from a number of defects, such as a simple pre/post research design and lack of an adequate control group.

      Despite the methodological weaknesses in the individual studies, a consensus has been reached in the literature that drug abuse treatments are effective.  That is, some types of treatments work for some drug-abusers.  The literature also suggests that some treatment modalities are more effective than others.  However, the literature also reveals that treatment outcomes vary across patient groups and that patient groups differ across treatment modalities.  In addition, we learned that the length of treatment is positively correlated with favorable outcomes.  Since the length of treatment also varies across modalities, it is impossible at present to determine the relative effectiveness of treatment modalities.

      We are only beginning to study the effectiveness of individual treatments and program characteristics.  Until individual treatments are related to outcomes at the patient level, we will continue to be unable to match different types of patients to specific modalities and treatments to achieve the most beneficial results.  Finally, we need to recognize that addiction is a chronic disease and that successfully treated patients frequently relapse and reenter treatment programs.  We need to shift from evaluating single episodes of treatment to evaluating the lifetime effects of drug abuse treatments as is done with evaluations of most medical care interventions.

      There have been only a relatively small number of published cost-effectiveness evaluations of drug abuse treatments despite policy makers need for such evaluations.  In general, this is because the costs and dollar benefits are difficult to conceptualize and even harder to estimate.  Cost-effectiveness evaluations are quite expensive and policy makers are unwilling to pay for them.

      Most studies only report the average cost of treatment per patient for the program as a whole; the so-called “slot” cost.  The “slot” cost is generally based on accounting records rather than the opportunity cost of resources used.  Most studies ignore patient and donor time costs as well as the value of donated capital.  Therefore, the estimated “slot” costs underestimate the true costs of treatment.  “Slot” costs are not accurate for individual patients because they receive different amounts of treatments.  It is not surprising that length of treatments is positively correlated with favorable outcomes.  Clients who complete the program consume more resources than those that dropout early and they should have more favorable outcomes.  The few studies that have examined the cost of drug abuse treatment report that the methods used to measure cost are not consistent and that is difficult to make comparisons across studies.  There is a great need for a standardized method of measuring costs.

      Outcomes vary from program to program and there is no agreement on how to measure or to value them (i.e., monetize them).  Outcome data can be subjective based on patient follow-up interviews or it can be objective based on financial and accounting records.  The outcomes and monetized benefits included in a cost-effectiveness evaluation depend on the analyst’s perspective.  Most of the benefits of drug abuse treatments come in the form of avoided costs.  The question is, whose benefits or avoided costs are relevant to the study? Many benefits such as reductions in intangible costs are ignored because they are difficult to measure.  The same could be said for “quality-of-life” effects on patients and their families.  Most cost-benefit studies include only the relevant benefits that are easy to measure in their analysis.  There is an obvious need for a standardized method of measuring benefits.

      Until a standard method is developed for calculating treatment costs and measuring outcomes, CEA studies will have little value.  The cost-effectiveness of a program or treatment needs to be compared with the cost-effectiveness of another program measured in the same way to have value.  At present, only a small number of CEA studies of drug abuse treatments have been published.

      Cost-benefit studies can stand-alone because both costs and benefits are measured in dollars.  CBAs can be used to determine if we are getting value for the money spent.  If there was a standard for measuring costs and benefits, we could learn something about the relative efficiencies of programs by comparing their benefit-cost ratios.  Since we do not have such standards, comparisons of benefit-cost ratios should be made cautiously.

      Our review of ten individual evaluations of drug abuse treatment leads to the following conclusions.  As expected, the calculated benefit-cost ratios ranged from 26.3 for residential drug-free treatment for heroin addicts (Tabbuah 1986) to less than 1.0 based on increased employment and earnings (Frank et al. 1990 and French and Zarkin (1992).  The estimated values for benefit/cost ratios are quite sensitive to methodology, the analyst’s perspective, the benefits analyzed and the type of modality.  Despite the methodological weaknesses of the studies, it must be concluded that in general drug abuse treatments generate benefit-cost ratios exceeding 1.0.  The cost-savings from reductions in criminal activity are the greatest and the benefits in employment and earnings appear to be the lowest.  The gains in medical care cost savings fall somewhere in between.  When you add up all of the benefits from the society’s point of view, it is clear the benefit-cost ratio exceeds 1.0.  This finding is supported by the three natural experiment studies that investigated the effects of the involuntary closing of drug abuse treatment programs.

      There appears to be some variation in benefit-cost ratios across treatment modalities.  However, because modalities differ in terms of the clients they serve, we cannot be sure the B/C ratio differences are due to treatments and not patients.

      The recent contributions by French et al. (1996) on the use of QALYs and Rajkumar and French (1997) on measuring intangible costs and Barrett (1999) on life-years gained are notable contributions to the drug abuse treatment literature and need to be pursued.

      Our review of five recent evaluations of State drug abuse treatment programs also was enlightening.  The estimated B/C ratios ranged from 12.58 for methadone discharge patients from the citizen taxpayer’s point of view (Geratein et al. 1994) to 0.38 for indigent patients in Washington state’s ADATSA Program from the government’s point of view (Wickizer and Longhi 1997).  The B/C estimates are highly sensitive to the study methodology, the selection of benefit categories, the types of clients and modalities, and the analyst’s viewpoint.  Generally speaking, crime cost-savings are the greatest and employment and earnings cost-savings are the smallest (e.g. sometimes negative).  Medical care and public assistance cost-savings fall somewhere in between.

      Despite their individual weaknesses, the five studies collectively provide strong support for the argument that state drug abuse treatment programs are efficient (i.e., their B/C > 1.0).  The evaluations that use a control group and multiple regression analysis tend to produce smaller benefit-cost ratios (Finigan 1996, Wickizer and Longhi, 1997 and Estee and Nordlund 2001).  The most methodologically sound study by Finigan (1996) reported a B/C = 5.6 from the citizen taxpayers point of view and a B/C = 1.64 from the government’s point of view.  His citizen taxpayer’s estimate of B/C was based on the following types of cost savings:  criminal justice, public assistance, victim, and theft.  His government’s estimate of B/C was based exclusively on criminal justice and public assistance cost savings.

      Other studies have indicated that medical care cost savings are important (Tabbush 1986, Gerstein et al. 1984, Mauser t al. 1994 and Estee and Nordlund (2001).  The most methodologically sound study of the benefits and costs of medical care cost savings is that of Estee and Nordlund (2001).  They estimated a B/C = 2.19 from the society point of view based exclusively on medical care cost savings.  If we add this estimate to Finigan’s we find a B/C = 7.79 (=5.60 + 2.19) from the society point of view.  We do not know how much of the medical care cost savings would accrue to government.  If we assume all of it and we add Estee and Nordlund’s (2001) B/C = 2.19 to Finigan’s (1996) B/C = 1.64, we obtain a total B/C = 3.83 from the government’s point of view.