PsychNews International, Vol. 2, May-June 1997
Bait and Switch in Project MATCH
What NIAAA Research Actually Shows About Alcohol Treatment
Project MATCH was an extremely expensive clinical trial in which a volunteer, selected group of mainly alcohol dependent patients who had a small number of outpatient sessions of three different types of therapy with a highly trained and monitored staff reduced their drinking substantially over a year period. However, the primary purpose of the study was to show that alcoholics with different profiles will respond better to different sorts of treatment, so that overall outcomes can be improved by proper matching of alcoholics with treatment type. Project MATCH results disproved the matching hypothesis, which prominent clinical researchers had pushed for years (Allen & Kadden, 1995; IOM, 1990; Mattson, 1994; Miller, 1989).
In MATCH's aftermath, NIAAA director Enoch Gordis and other MATCH representatives touted the surprisingly good overall outcomes in the study, which they interpreted to show that alcoholism treatment as delivered in the U.S. (which in over 93 percent of centers is 12-step treatment; Roman & Blum, 1997) produces good results. This conclusion is unfounded. Indeed, a concurrent community study (NLAES) by the NIAAA revealed results of treatment as actually experienced in the U.S. to be shockingly bad, leading to outcomes worse than for those who went untreated for their alcohol dependence!
The NIAAA's willingness to interpret MATCH results to support ideas it could not properly test (in the absence of a nontreatment control group) is disturbing both because it is unscientific and because it seems to be politically motivated. Moreover, several revisionist commentators on Project MATCH (including the present author) have been attacked by MATCH investigators. MATCH and the NIAAA have embargoed alternative interpretations in order to control these results and spin them to their advantage and that of the alcoholism establishment.
The results that are consistent across both MATCH and NLAES are that (1) minimal or no treatment produces outcomes that are equal to/better than those from longer/standard treatments; (2) patient traits and initiative are far more important than treatment type or intensity for recovery; (3) reduced drinking is the most common outcome for alcohol-dependent individuals, an idea that has been verboten on the American alcoholism scene (Peele, 1992).
The Basic Project
Project MATCH Overview. MATCH was organized by the National Institute on Alcohol Abuse and Alcoholism to assess the impact of three different types of alcohol treatment (Coping Skills, Motivational Enhancement, and 12 Step) with special reference to patient characteristics that might predict better outcomes for each treatment for different types of alcoholics. Treatment was on an outpatient basis (for a purely outpatient group and an aftercare to hospital treatment group). Subjects were treated for 3 months and outcomes were measured at three-month points until a year after treatment. The two primary outcome measures were days abstinent and drinks per drinking day (Project MATCH Research Group, 1997).
Project Staff and Resources. The cost of Project MATCH was originally placed at $27 million. However, John Allen, an NIAAA administrator, has since put the cost at $35 million (Jeffrey Schaler, private communication, May, 1997). The training manuals created by MATCH include a foreword by Enoch Gordis announcing: "Researchers in Project MATCH are among the most senior and experienced treatment scientists in the field." Overall, MATCH involved up to 130 clinical professionals in its treatment, administration, and analysis components.
Subjects. The study began with 1,726 subjects, almost all of whom (>95%) measured alcohol dependent. Potential subjects were assessed for alcohol abuse/alcohol dependence according to DSM-III-R, but were excluded if they had concurrent drug dependencies (even though "combined alcohol and drug abuse show up as the most frequent problem at admission to substance abuse treatment"; SAMHSA, 1997). Initially, 4,481 subjects were identified. At initial screening, 459 declined to participate because of the "inconvenience" of treatment (which the Project MATCH Research Team, 1997, p. 11, asserted was "attributable to logistical considerations rather than... motivation"). Other potential subjects were excluded for "failure to complete the assessment battery; residential instability; legal or probation problems, etc.," but no figure is given for how many fell into each category.
Thus, subjects were volunteers with better than average prognosticators (e.g., residential stability; not also abusing drugs). But the study report did not compare patients who were treated to those who were eliminated from the study or to alcoholics in general. Nonetheless, the Team contended, "nor is there reason to believe that the recruitment procedures failed to provide a broad range of clients typically seen in these types of clinical settings" (p. 11).
About 10 years ago at the conference "Evaluating Recovery Outcomes" (University of California, 1988, p. 23), MATCH principal investigator Richard Longabaugh commented on a study at a private treatment center [Edgehill Newport] which found that "66 percent of the patients in the program were continuously sober at a one-year follow-up. However, the program report was limited to treatment of socially-stable patients... [along with other filters]. Was this population representative of the population they were treating?" Longabaugh contended that since the research did not analyze excluded subjects, "skeptics like myself" had to conclude it was worthless as a comparative assessment of treatment.
Treatment Delivery. Treatment was delivered individually on an outpatient basis. Delivery involved 12 weekly sessions, except for motivational enhancement, which was four sessions. Subjects on average completed only two-thirds of their sessions (which seems like 3 sessions for motivational enhancement and 8 sessions for the other two therapies). Treatment protocols and standards for therapist certification were developed; all sessions were videotaped and 25 percent were monitored by supervisors. The care and quality control of Project MATCH therapy differs substantially from standard treatment provided for alcoholism patients in the U.S. Jon Morgenstern, of the Rutgers Center of Alcohol Studies, has observed alcohol counselors in regular treatment settings and reports their standard of therapy is extremely poor (personal communication, early 1997, at a PBS production meeting).
Client Measures. Treatment lasted 12 weeks with three-month assessments up to 15 months. Ten primary client characteristics were measured and reported (e.g., conceptual level, motivation, psychiatric severity, gender). Additional secondary characteristics not yet reported were also measured, including DSM diagnostic categories, self-efficacy, alcohol dependence and seven other traits and "several higher order and global matching hypotheses."
Drinking Results. Subjects substantially reduced drinking intensity and frequency of drinking across all treatments, but relatively few did not drink at all. Overall, drinking fell from an average of 25 days to 6 days per month and amount consumed on drinking occasions was also reduced. About 20 percent of the outpatient group and slightly over a third (35%) of the aftercare group abstained throughout the year follow-up. However, only 40% of aftercare and 46% of outpatient subjects drank heavily for three straight days during the year follow-up. In other words, a third of outpatient alcoholics drank without bingeing compared with a fifth who abstained.
Treatment and Interaction Effects. There were no significant differences in drinking days overall or in any month or any other main effects among the treatment groups. Sixteen hypothesized interactions between the ten client characteristics and three treatment types were conceived. In the aftercare group, no significant impact on drinking outcomes was found for any of these interactions. In the outpatient group, less psychiatrically severe subjects had 4 more abstinent days per month on average in 12-Step than in Coping Skills treatment.
Thus, 64 interactions were tested -- 16 (patient x treatment interactions) x 2 (outpatient v aftercare) x 2 outcomes (drinking days v # drinks on drinking days) -- and one of 64 proved significant, an interaction involving two of the three treatments that was not explicitly hypothesized. When large numbers of hypotheses are tested, some will be significant by chance alone. Thus, the null hypothesis that matching has no significance for outcomes was supported by this study. There were significant client characteristic effects, however -- for example, among outpatients, higher motivation led to significantly better drinking outcomes, while in both the outpatient and aftercare groups greater social support for drinking (presumably excessive) predicted worse outcomes.
- "The largest, statistically most powerful, psychotherapy trial ever conducted," and also by far the most expensive, produced little new information.
- Research limitations -- including a highly selected client population and unusually high level of quality control of treatment -- make generalization of these results dubious. "The overall effect of being part of Project MATCH, with extensive assessment, attractive treatments, and aggressive follow-up may have minimized naturally occurring variability among treatment modalities and may, in part, account for the favorable treatment outcomes" (Project MATCH Research Team, 1997, p. 24).
- The principal result of this study was that modest treatment contact produced substantial and enduring drinking reductions among an alcohol dependent population. Motivational enhancement involving 4 or fewer sessions produced results as good as treatments involving several times as many sessions.
- While no treatment characteristics were significant, and virtually no treatment-patient matches, psychological and contextual traits of alcoholics were significant for outcomes.
- Treatment outcomes were conceived primarily as successful reduction in drinking, which is utterly at variance with treatment goals as usually conceived in the U.S., where 99 percent of programs urge abstinence (Roman & Blum, 1997).
The Larger Context and the Aftermath of Project MATCH
A June, 1996 presentation of MATCH results and a December, 1996 MATCH press conference were highly publicized (cf. Leary, 1996; Peele, 1996). MATCH principals and NIAAA director Gordis emphasized the remarkable success of alcoholism treatment. According to Gordis: "The good news is that treatment works. All three treatments evaluated in Project MATCH produced excellent overall outcomes" (Bower, 1997, p. 62). MATCH's lead researcher, Thomas Babor, of the University of Connecticut, declared "participation in any of the MATCH treatments would be associated with marked positive change" (Alcoholism & Drug Abuse Week, 1997).
The effectiveness of treatment is heavily promoted by government agencies in conjunction with the treatment industry in the U.S. "Treatment Works! Month" is celebrated on the Internet (www.health.org/csat/) and other media: "Designed by SAMHSA (Substance Abuse and Mental Health Services Administration)/CSAT (Center for Substance Abuse Treatment) with the cooperation of the National Association of Alcoholism and Drug Abuse Counselors (NAADAC), these promotional materials will help educate people throughout your state, county, city or community about the true value of treatment and the fact that it really works." If treatment works in the U.S., then 12-step therapy must be highly successful. As NIAAA MATCH coordinator Margaret Mattson concluded: "The results indicate that the Twelve Step model, which is the most widely practiced treatment in the U.S., is beneficial" (Mattson, 1997).
But clinical research has not found that all treatments are equally effective. Tolling decades of controlled research on alcohol treatment, Miller and his colleagues rated brief interventions, social skills training, and motivational interviewing highest in effectiveness, while general alcoholism counseling and education lectures were at the bottom. Rated at the bottom of the list of treatments without sufficient evaluations to be reliably tested was Alcoholics Anonymous. Yet alcohol treatment in America comprises primarily these disproven elements -- "The negative correlation between scientific evidence and application in standard practice remains striking, and could hardly be larger if one intentionally constructed treatment programs from those approaches with the least evidence of efficacy" (Miller et al., 1995, p. 33).
Another NIAAA research project provided an actual comparison between treated and untreated alcoholics, unlike MATCH. This is the National Longitudinal Alcohol Epidemiologic Survey (NLAES) of subjects in the general population who were assessed as alcohol dependent, and who either received or did not receive (the large majority) treatment. Its results were:
|Outcome categories||Treated (n=1,233)||Untreated (n=3,309)|
|< 5 years since onset dependence|
|drinking w/o abuse/dependence||19%||41%|
|20+ years since onset dependence|
|drinking w/o abuse/dependence||24%||60%|
|Source: Dawson (1996)|
Thus, a study designed and conducted by the NIAAA to determine actual experiences of American alcoholics yields a picture far different from the optimistic results projected by NIAAA and Project MATCH officials and the Treatment Works! coalition: i.e., those entering alcohol treatment actually fare worse than untreated alcoholics.
Although all subjects in NLAES were alcohol dependent, treated subjects had somewhat more severe symptoms than untreated subjects. But the goal of government agencies and private treatment providers of getting untreated alcoholics into treatment finds no justification in these NIAAA data. The most obvious results of NLAES are (1) alcohol dependence diminishes over time in both treated and untreated groups, and such "maturing out" dwarfs any benefit of treatment; (2) while treatment encourages abstinence, it suppresses the moderation of alcohol dependence that occurs naturally in the majority of cases, so that more treated alcoholics continue to drink abusively. The results of both NLAES and MATCH indicate the abstinence fixation that characterizes American alcoholism treatment is deeply dysfunctional.
Spinning Project MATCH
The NIAAA faced in NLAES and MATCH a combination of embarrassing results: the most expensive treatment study ever conducted was a dud, while NIAAA research shows that alcohol treatment in the U.S. doesn't work well. MATCH researchers and the NIAAA acknowledged that "These findings challenge the notion that patient-treatment matching is necessary in alcoholism treatment" (Gordis quoted in Leary, 1996). The NIAAA's Mattson (1997), a major "matching" backer, declared, ".... based on the main findings of Project MATCH, matching seems to add little to improving treatment...."
But when outsiders commented on MATCH results, MATCH researchers screamed murder. The first announcement of results in Washington DC in June, 1996 made clear the virtual absence of significant matching relationships, while Gordis said the results showed alcoholism treatment worked well. When psychologist Jeffrey Schaler gave his own interpretation of MATCH results on Internet lists, Ronald Kadden, writing as Chair of the Project MATCH Steering Committee, objected to Schaler's discussion (Schaler, 1996). This quasi- governmental communication charged that Schaler falsely stated that Mattson had confirmed Schaler's summary of MATCH results. The NIAAA later conceded that she had indeed done so.
In my own case, a MATCH researcher, Richard Longabaugh, of Brown, wrote on the APA addiction list server (this quote, which is fairly incomprehensible, is exact):
I am a Project MATCH P.I., who has followed much of the correspondences on this list server regarding Project MATCH results. Contary [sic] to Stenton [sic] Peele's conclusion, the results of Project MATCH are far from concluded. There are at least two more papers (one under review, and one in preparation) focusing on the primary match of findings. That should be digested. Only then do I think we will have a sufficient view of the finding's (sic) to evaluate MATCH's contribution. To the alcohol treatment knowledge base, I'd say publications of these results are about a year away. On the other hand, discussion is 'livelier' without the facts. (APA Division 50 list server, April 16, 1997)
The NIAAA's and other "insider" researchers' reactions to outside critiques is predictable in terms of how bureaucracies behave. But it is nonetheless startling to see researchers claim that others cannot interpret their data, even as NIAAA administrators and MATCH researchers do exactly that publicly! According to Longabaugh, for instance, non-MATCH investigators should not respond to interviews and published comments by Gordis, Babor, and Mattson until at least another year, despite the highly detailed, 22-journal-page research article the Team has published.
The MATCH situation calls to mind the Dead Sea Scrolls, to which "access [was] ... tightly controlled by a small circle of scholars, who ... built their careers on translating and analyzing the 2,000-year-old religious works" ("Resurrection of the Dead Sea Scrolls," 1991). Here, a government agency and its representatives are trying to suppress free analysis and interpretation of federally funded research. Another case of suppression of free inquiry and exchange involving alcohol was recently revealed in regard to the NIH's refusal in 1972 to allow Harvard anthropologist Carl Selzer to publish data revealing that moderate drinkers had less heart disease than abstainers (Altman, 1997; Selzer, 1997). This cannot be tolerated by either science or a free society.
Alcoholism & Drug Abuse Week. (1997, January 6). P 1.
Allen, J.P., & Madden, R.M. (1995). Matching clients to alcohol treatments. In R.K. Hester and W.R. Miller (Eds.), Handbook of alcoholism treatment approaches (2nd Ed., pp. 278-291). Boston: Allyn and Bacon.
Altman, L.K. (1997, April 29). Experts see bias in drug data. New York Times, pp. C1, 8.
Bower, B. (1997, January 25). Alcoholics synonymous: Heavy drinkers of all stripes may get comparable help from a variety of therapies. Science News, 151, 62-63.
Dawson, D.A. (1996). Correlates of past-year status among treated and untreated persons with former alcohol dependence: United States, 1992. Alcoholism: Clinical and Experimental Research, 20, 771-779.
Institute of Medicine (IOM). (1990). Broadening the base of treatment for alcohol problems. Washington, DC: National Academy Press.
Leary , W.E. (1996, December 18). Responses of alcoholics to therapies seem similar. New York Times, p. A17.
Mattson, M.E. (1994, September 22). Patient- treatment matching. Alcohol Health & Research World, 18, 287-295.
Mattson, M.E. (1997, March). Treatment can even work without triage: Initial results from Project MATCH. EPIKRISIS, 8(3), 2-3.
Miller, W.R. (1989). Matching individuals with interventions. In R.K. Hester and W.R. Miller (Eds.), Handbook of alcoholism treatment approaches (pp. 261- 271). Elmsford, NY: Pergamon.
Miller, W.R., Brown, J.M., Simpson, T.L., Handmaker, N.S., Bien, T.H., Luckie, L.F., Montgomery, H.A., Hester, R.K., and Tonigan, J.S. (1995). What works?: A methodological analysis of the alcohol treatment outcome literature. In R.K. Hester and W.R. Miller (Eds.), Handbook of alcoholism treatment approaches (2nd Ed., pp. 12-44). Boston: Allyn and Bacon.
Peele, S. (1992). Alcoholism, politics, and bureaucracy: The consensus against controlled drinking in America. Addictive Behaviors, 17, 49-62.
Peele, S. (1996, September/October), Recovering from an all-or-nothing approach to alcohol. Psychology Today, pp. 35-43; 68-70.
Project MATCH Research Group. (1997). Matching alcoholism treatments to client heterogeneity: Project MATCH posttreatment drinking outcomes. JSA, 58, 7-29.
Resurrection of the Dead Sea Scrolls. (1991, September 8). New York Times, 4, p. 7.
Roman, P.M., and Blum, T.C. (1997). National treatment center study. Athens, GA: Institute of Behavioral Research, University of Georgia.
SAMHSA (1997, February). National admissions to substance abuse treatment services: The treatment episode data set (TEDS) 1992-1995 (Advance Report No. 12). Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Studies.
Schaler, J. (1996, August/September). Selling water by the river: The Project MATCH cover-up . PsychNews International, 1(5).
Selzer, C.C. (1997). Second thoughts: "Conflicts of interest" and "political science." Journal of Clinical Epidemiology, 50, 627-629.
University of California. (1988, February). Evaluating recovery outcomes: A national conference. San Diego: UCSD Extension.