Journal of Substance Abuse Treatment, Vol. 17, No. 4, pp. 305–312, 1999 Copyright © 1999 Elsevier Science Inc. Printed in the USA. All rights reserved. 0740-5472/99 $–see front matter PII S0740-5472(99)00010-0 ARTICLE The Value of Acupuncture Detoxification Programs in a Substance Abuse Treatment System Michael Shwartz, phd,* Richard Saitz, md, mph,† Kevin Mulvey, phd,‡ and Patrick Brannigan, ba§ *School of Management, Boston University, Boston, MA †Clinical Addiction Research and Education Unit, Section of General Internal Medicine, Department of Medicine, Boston University School of Medicine and Boston Medical Center, Boston, MA ‡Office of Research, Health Assessment and Data Systems, Boston Public Health Commission, Boston, MA §Boston University School of Public Health, Boston, MA Abstract – Our purpose is to compare baseline characteristics and detoxification readmission rates of clients treated at outpatient acupuncture programs and at short-term residential programs, two options available to persons seeking substance abuse detoxification. This was a retrospective cohort study using data on clients discharged from publicly funded detoxification programs in Boston between January 1993 and September 1994. Multivariate models were used to examine the effect on 6-month detoxification readmission rates of treatment at residential detoxification programs (used by 6,907 clients) versus at outpatient acupuncture programs (used by 1,104 clients) after adjusting for baseline differences. Acupuncture clients were less likely to be readmitted for detoxification within 6 months (odds ratio [OR] 0.71, 95% confidence interval [CI] 0.53–0.95). Similar results were found when the analysis was performed on a subsample of clients that were relatively similar in terms of baseline characteristics (OR 0.61, 95% CI 0.39–0.94). We determined that acupuncture detoxification programs are a useful component of a substance abuse treatment system. © 1999 Elsevier Science Inc. All rights reserved. Keywords – acupuncture; detoxification; substance abuse treatment. INTRODUCTION that met in 1991 concluded that “A review of available data indicates no clear evidence that acupuncture is effective compared to placebo or to existing treatments in the detoxification, primary rehabilitation, or relapse prevention of opiate or chemical dependence” (McLellan, Grossman, Blaine, & Haverkos, 1993, p. 575). Despite this, “ear acupuncture for substance abuse is used on 5,000 patients daily in literally hundreds of different settings in the United States and Europe” (Smith, 1994, p. 587), perhaps reflecting the fact that acupuncture has not “been proven ineffective in the treatment of substance dependence” (McLellan, Grossman, Blaine, & Haverkos, 1995, p. 141). There have been several small randomized trials comparing outcomes of those exposed to sham versus real The role of acupuncture in substance abuse treatment is controversial. A National Institute on Drug Abuse panel We would like to thank the Bureau of Substance Abuse Services, Massachusetts Department of Public Health, for use of their MIS data collection instruments and data on out-of-Boston utilization by Boston clients, and Terry Courtney, MPH, LAC, for helping us better understand the nature of acupuncture treatment. Supported in part by the Center for Substance Abuse Treatment, Cooperative Agreement 1-U88-T100022 Requests for reprints should be addressed to Michael Shwartz, School of Management, Boston University, 595 Commonwealth Avenue, Boston MA, 02215. Received August 17, 1998; Accepted October 20, 1998. 305 306 manual auricular acupuncture, most of which have suggested some value from acupuncture (Bullock, Culliton, & Olander, 1989; Bullock, Umen, Culliton, & Olander, 1987; Lipton, Brewington, & Smith, 1994; Washburn et al., 1993), though not all (Worner, Zeller, Schwarz, Zwas, & Lyon, 1992). Studies such as these attempt to isolate the value of acupuncture when added to other program components. We are not aware of any studies that have compared entire programs that include acupuncture as “an adjunct or a complement” to other program activities (the way in which acupuncture is viewed by its supporters; Brumbaugh, 1993) to alternative treatment modalities. The need for well-designed randomized controlled trials of acupuncture programs is indisputable. Nevertheless, the report by the Quantitative Methods Working Group convened by the National Institutes of Health (NIH) in support of the NIH Office of Alternative Medicine (Levin et al., 1997), while recognizing that “for alternative therapies to become accepted, they must endure the same degree of scientific scrutiny as conventional therapies” (p. 1092), also notes that “clinical trials are not the only game in town” (p. 1087) and that “in fact, new medical knowledge most often is obtained through observational means. . . . This is typically an important first step in understanding an effect on a particular outcome of a previously unconsidered factor” (p. 1089). Our focus in this article is on the benefits of outpatient acupuncture detoxification (detox) programs as a component of a substance abuse treatment delivery system, where, for those seeking detox, it is an alternative to traditional short-term residential detox. Observational studies are particularly useful for examining alternatives in a field setting. This is the design used in our study of clients seeking detox through the publicly funded substance abuse treatment system. We compare characteristics and 6-month readmission rates for detox of those who were initially treated in an outpatient acupuncture program to those initially treated in a short-term residential detox program. Outpatient acupuncture programs usually combine acute detox with longer-term maintenance and motivational counseling over several months; short-term inpatient residential detox programs, which last around a week or less, medically treat short-term withdrawal symptoms and then refer clients to further inpatient, outpatient, or self-help treatment. These are very different treatment modalities. However, as part of a substance abuse treatment system in which clients seeking treatment present at intake sites for assessment and referral, outpatient acupuncture programs are an alternative to short-term residential detox programs. At the point of entry, a decision must be made—is the client referred to an acupuncture program or a residential detox program? These alternatives at the time of entry into the system provide the rationale for our comparison of clients treated at acupuncture programs to those treated at shortterm residential detox programs. M. Shwartz et al. It is important to emphasize at the outset that the purpose of our analysis is not to isolate the value of the technique of acupuncture per se, but to compare outpatient detox programs that include acupuncture to short-term residential programs, the detox modality most commonly used in the publicly funded treatment system. METHODS Sample Boston was one of the eight Target Cities demonstration projects funded in 1990 by the Center for Substance Abuse Treatment (CSAT), Substance Abuse and Mental Health Services Administration (SAMHSA) to design and implement enhancements to the publicly funded substance abuse treatment system. (By publicly funded, we mean programs receiving federal or state funds, in our case which flow through the Massachusetts Bureau of Substance Abuse Services [BSAS]. These funds are used to finance costs for clients without other means to pay for treatment.) One of the main enhancements of what came to be called the Boston Office of Treatment Improvement (BOTI) was the establishment of three central intake units, where clients were assessed and referred to appropriate treatment programs. For clients presenting at the central intake units and in need of detox, alternatives were referral to a residential detox program or an outpatient acupuncture program. The Boston publicly funded treatment system has four free-standing short-term residential detox programs, with an average length of inpatient stay of 1 week, and three outpatient acupuncture detox programs, with an average duration of treatment of about 4 months. The residential detox programs offer medical detox, individual and group counseling, education, introduction to self-help groups, and referrals to appropriate postdetox inpatient or outpatient treatment, which may last up to a year for clients entering these programs. The acupuncture programs follow the National Acupuncture Detoxification Association (NADA) guidelines (Brumbaugh, 1994). During the first 2 weeks, where the focus is on acute detox, standard acupuncture with auricular points is given daily. Following acute detox, there is a maintenance period in which acupuncture is given two to three times per week for several months. During this period, body points may be added, depending on client signs and symptoms. An important component of the programs is motivational counseling, provided either individually or in group settings. As part of BOTI, a city-wide management information system (MIS) was established that allowed tracking of clients as they moved through the treatment system. By merging Boston data with data from the BSAS MIS, we were able to track admissions to all Massachusetts programs. We closed the analytical database created Value of Acupuncture Detox from the MIS at the end of March, 1995. In order to have 6 months of postdischarge follow-up data on each client, the population for this study was defined as cases that had been discharged from a residential detox program or an outpatient acupuncture program during the period January 1993 through September 1994. (Though one does not usually think of “discharge” from outpatient treatment, in fact a formal discharge form is required by the State for clients treated at publicly funded programs. “Discharge” does not imply clients “completed” treatment, only that they were no longer formally in the treatment program.) Each client was classified as a residential case or an acupuncture case based on the modality of their first discharge during the study period. The admission to this first modality is referred to as the index admission. Data The BOTI MIS collects admission and discharge information on each client, using the same standardized forms as the Massachusetts BSAS (Camp, Krakow, McCarty, & Argeriou, 1992). Variables recorded at admission that were used as covariates in multivariate models included: gender, race/ethnicity, age, education, employment status, yearly income, insurance, living situation (with child, with other adult but no child, alone), residence (street/shelter, institution/boarding house, private residence), prior mental health treatment (yes/no), primary drug (alcohol, cocaine, crack, heroin, and marijuana), and prior substance abuse treatment history (for different modalities, coded yes/no based on lifetime use). We created two additional variables for each client from the MIS admission data: (1) number of admissions to each modality in the year preceding each index admission (determined from the MIS by linking admission records on the same client and coded as 0, 1, $2); and (2) a drug severity score. As described in Shwartz, Mulvey, Woods, Brannigan, & Plough (1997), the drug severity score was developed by assigning points to responses to the following three questions on the admission form: age at first use, time of last use, and frequency of last use. The score was calculated by summing up points assigned to responses for each major drug used (alcohol, cocaine, crack, heroin, and marijuana). Outcome Measures Our outcome measure was whether clients were readmitted for detox, either residential or acupuncture, during the first 6-month period at which they were at risk for relapse. Clients in residential treatment are not at risk of relapse until they have been discharged from the program. Hence, for the residential detox programs, we measured 6-month readmission from the time of discharge from the program. Acupuncture is an outpatient program, and cli- 307 ents are at constant risk of relapse. Hence, for acupuncture programs, we measured 6-month detox readmission from admission to the program. Statistical Analysis Most baseline variables and the outcome variables were categorical. Chi-square tests were used to examine the statistical significance of differences in baseline variables between acupuncture and residential cases and to analyze differences in readmission rates. When the baseline variables were continuous (e.g., the drug severity score), t-tests were used. To adjust for differences in baseline characteristics that impacted outcome, we used a stepwise procedure to build a multivariate logistic regression model predicting the log odds of detox readmission as a function of statistically significant baseline variables (at a .05 level of significance). An indicator variable, coded as one if the case was discharged from acupuncture and zero if from residential detox (the reference group) was included in the model. Our main interest was in the odds ratio (OR) associated with this indicator variable. Also, we constructed separate stepwise logistic regression models stratifying on clients’ recent detox history, a variable that turned out to be strongly related to our outcome measure, and models stratifying on primary drug. We report the c statistic, a standard summary measure of model performance when predicting a dichotomous outcome variable (Ash & Shwartz, 1997; Harrell, Lee, Califf, Pryor, & Rosati, 1984). Among all possible pairs of clients in which one was readmitted to detox within 6 months and one was not, the c statistic is equivalent to the proportion of pairs in which the predicted probability of readmission was higher for the client who was readmitted than for the client who was not. As will become apparent, acupuncture clients differed from residential detox clients in terms of many of the covariates, giving rise to the concern that multivariate models might not be able to adequately adjust for differences. To examine this possibility, we identified a subsample of acupuncture and residential detox clients that were similar in terms of baseline characteristics and reran the multivariate analysis on this subsample. To identify the subsample, we used the “propensity score” approach (Rosenbaum & Rubin, 1984), as follows: (a) stepwise logistic regression was used to build a model to predict the probability that a case was treated with acupuncture based on client baseline characteristics; (b) the predicted probabilities of acupuncture for those clients receiving acupuncture were ranked and then divided into 20 categories, such that an equal number of cases fell into each “predicted probability of acupuncture” category; (c) for each of the “predicted probability of acupuncture” categories, the number of residential detox cases whose “predicted probability of acupuncture” fell in that category was determined; and 308 M. Shwartz et al. (d) a matched sample of residential detox cases (matched on the “predicted probability of acupuncture”) was selected by random sampling within each predicted-probability category of a number of residential detox cases equal to the number of acupuncture cases in that category (or, in the higher predicted probability categories where there were not sufficient residential detox cases, we randomly sampled from the group of acupuncture cases a number equal to available residential detox cases). The result of this process was a subsample of acupuncture and residential detox cases that were similar in terms of their predicted probability of receiving acupuncture. A conditional logistic regression model (Breslow & Day, 1980) (used to reflect the fact we sampled cases by strata of propensity score) was used to predict detox readmission for this subsample of cases. Independent variables in this analysis were the indicator variable for acupuncture use and the variables for recent detox history (zero, one, or two, or more residential or acupuncture detox admissions in the preceding year). RESULTS Baseline Characteristics Of 8,011 clients discharged from detox during the study period, 6,907 (86%) had their first detox discharge from a residential program and 1,104 (14%) from an acupuncture program. Table 1 compares baseline characteristics of clients discharged from the two treatment modalities. Among the notable differences, acupuncture clients were more likely to have graduated college (13% vs. 4%), to be employed (57% vs. 13%), to have private insurance (15% vs. 3%), to be living with a child or adult (76% vs. 55%), and to have had prior mental health treatment (28% vs. 12%). They were less likely to live in a shelter (3% vs. 30%). Acupuncture clients’ drug usage profile differed somewhat from residential detox clients, consisting of a lower percentage of clients whose primary drug was alcohol or heroin, and a somewhat larger percentage whose primary drug was crack or marijuana. The TABLE 1 Percentage of Residential Detox and Acupuncture Clients with the Indicated Characteristic Characteristic Gender (female) Race/ethnicity Black Hispanic White Education High school graduate College graduate Employment (unemployed) Insurance Uninsured Medicaid Private insurance Lives with Child Adult (no child) In shelter In institution Prior mental health treatment Primary drug Alcohol Cocaine Crack Heroin Marijuana Substance abuse admissions in the last year Residential detox None 1 $2 Short-term residential (none) Long-term residential (none) Outpatient (none) Acupuncture (none) % Residential Detox (n 5 6,907) % Acupuncture (n 5 1,104) 29.1 33.0 .01 45.7 11.7 40.9 46.2 9.7 43.0 .74 .05 .19 55.9 3.9 86.8 59.4 13.4 43.2 .03 ,.01 ,.01 65.4 28.2 3.0 52.3 21.2 15.4 ,.01 ,.01 ,.01 9.5 45.5 30.3 4.0 12.3 19.3 56.5 2.9 6.1 27.8 ,.01 ,.01 ,.01 ,.01 ,.01 42.3 16.2 15.9 24.6 0.7 32.4 16.6 20.2 19.0 8.4 ,.01 .78 ,.01 ,.01 ,.01 56.7 20.2 23.1 76.2 80.5 80.6 95.9 81.0 12.1 7.0 94.8 93.5 54.3 90.1 ,.01 ,.01 ,.01 ,.01 ,.01 ,.01 ,.01 p Value of Acupuncture Detox 309 average drug severity score was similar in the two groups, 27.6 for acupuncture clients and 27.4 for residential detox clients (p 5 .65). There were large differences in prior use of the substance abuse treatment system. In the year preceding their index admission, acupuncture clients were much less likely than residential detox clients to have had admissions to residential programs: 19% versus 43% to residential detox, 5% versus 24% to short-term (30-day) treatment, 6% versus 19% to long-term treatment. However, they were more likely to have had an outpatient (nonacupuncture) treatment admission in the last year (46% vs. 19%). Outcomes Eighteen percent of acupuncture clients were readmitted to detox within 6 months, much lower than the 36% of residential detox clients who were readmitted within 6 months of discharge. However, after adjusting for differences in baseline characteristics, the odds that an acupuncture client was readmitted to detox in 6 months were 0.71 of the odds that a residential detox client was readmitted (p 5 .02, 95% Confidence Interval [CI] 5 0.53–0.95). The c statistic for the model was 0.96. In addition to the “prior detox treatment in the last year” variables, the other statistically significant variables to enter the model were the following: black (OR 5 0.76), high school graduate only (OR 5 1.27), drug severity score (OR 5 1.01), admission following intake at a central intake site (OR 5 1.28) and one admission to a short-term (30-day) residential treatment program in the last year (OR 5 0.79). However, number of detox admissions in the year preceding the index admission were the only really important variables in the model: one detox admission in the last year, OR 5 412; two or more detox admissions, OR 5 1723. In a model including only the recent detox history variables (zero, one, or two, or more detox ad- missions in the year preceding the index admission), the c statistic was 0.95, indicating the extent to which recent detox history was the main factor accounting for detox readmission rates. The reason for the very high OR associated with recent detox admissions is illustrated in Table 2. Clients without a detox admission in the last year were very unlikely to be readmitted for detox in 6 months, whatever their mode of treatment; clients with two or more detox admissions in the last year were very likely to be readmitted, whatever their mode of treatment. Table 2 also shows results when stratifying on recent detox history. Acupuncture appears particularly beneficial for those clients with two or more detox admissions in the year preceding their index admission. There is weak evidence (p 5 0.15) that residential treatment might be the preferred modality for those with a single detox admission in the preceding year. As shown in Table 3, acupuncture appears particularly effective for clients whose primary substance was alcohol. Acupuncture and residential adjusted readmission rates (to take into account differences in statistically significant baseline variables) were similar for the other drugs. Based on results of the stepwise logistic regression model developed to predict the probability of acupuncture, the average predicted probability of receiving acupuncture among clients actually getting acupuncture was .50; the average predicted probability of receiving acupuncture among clients in residential detox was .07. These differences reflect baseline differences between the two groups of clients. The c statistic associated with this model was 0.90. We were able to match 740 acupuncture clients with suitable residential detox clients (in the higher “predicted probability of acupuncture” categories there were not sufficient residential detox cases). The subsamples selected through stratified sampling based on propensity score category were quite similar in terms of the covariates. For example, listed by percent- TABLE 2 Percentage of Clients Readmitted to Residential Detox or Acupuncture Within 6 Months and Odds Ratio Associated with Acupuncture as a Function of Number of Detox Admissions in the Year Preceding Index Admission Residential Detox Admissions in Year Preceding Index Admission No acupuncture admissions and no residential detox admissions 1 residential detox admission $2 residential detox admissions No residential detox admissions and 1 acupuncture admission a Acupuncture Detox Multivariate Model Odds Ratio n % Readmitted n 3,781 1,326 1,518 0.0 65.6 89.4 821 113 61 0.0 (.64)a 72.6 (.13) 78.7 (.01) –b 1.37 (0.89–2.12)c 0.47 (0.25–0.88) 124 78.2 69 40.6 (,.01) – % Readmitted Numbers in parentheses are the p values for a test of the null hypothesis that readmission rates of residential detox and acupuncture clients are similar. b Too few readmission cases to develop a model. c Odds ratio (acupuncture clients compared to the reference group of residential detox clients) and 95% confidence interval for the odds ratio. 310 M. Shwartz et al. TABLE 3 Percentage of Clients Readmitted to Residential Detox or Acupuncture Within 6 Months and the Multivariate Model Odds Ratio Associated with Acupuncture Readmission as a Function of Primary Drug Residential Detox Primary drug Alcohol Cocaine Crack Heroin Acupuncture n % Readmitted n % Readmitted Multivariate Model Odds Ratioa 2,919 1,122 1,099 1,699 37.3 31.0 28.8 40.6 358 183 223 210 10.9 19.7 21.5 31.4 0.53 (0.29–0.98) 1.03 (0.52–2.04) 0.97 (0.54–1.74) 1.11 (0.63–1.96) a Odds ratio associated with acupuncture admission (95% confidence interval). age of acupuncture clients and then percentage of residential detox clients, clients in the two subsamples were much more equivalent (compared to the differences in Table 1) in terms of having graduated college (7% in both cases), being employed (42% vs. 41%), having private insurance (6% vs. 9%), living with a child or adult (77% vs. 72%), living in a shelter (4% vs. 5%), and having had prior mental health treatment (21% in both cases). They were also more equivalent in terms of their recent use of the treatment system: in the preceding year: 26% versus 27% had detox admissions, 8% versus 9% had admissions to short-term residential treatment, 9% versus 11% had admissions to long-term residential treatment, and 37% versus 39% were in outpatient treatment. Finally, they were more similar in terms of primary drug: alcohol (30% vs. 35%), cocaine (18% in both cases), crack (23% in both cases), heroin (23% vs. 21%), and marijuana (3% in both cases). When the logistic regression model was run on the subsample of matched cases, the coefficient associated with acupuncture indicated that the odds of an acupuncture case being readmitted to detox in 6 months were 0.61 of the odds of a residential case being readmitted (95% CI 0.39–0.94). DISCUSSION Acupuncture appears as an effective modality for at least some clients seeking substance abuse detox. The 14% of clients that used acupuncture had lower 6-month detox readmission rates, even after taking into consideration ways in which they differed from clients admitted to residential detox programs. The propensity score analysis suggests that for at least another 10% or so of those in residential treatment (740/6,907), acupuncture would have been an effective alternative. Acupuncture appears particularly effective for those clients whose primary substance was alcohol and for those with two or more detox admissions in the year preceding their index admission. In terms of our drug severity score, acupuncture clients were similar to residential detox clients. However, it is important to note that our measure of drug severity was used only as a measure of relative severity among clients within the Boston system, and its external validity has not been analyzed. We did collect severity information using the Addiction Severity Index (ASI) (McLellan, Luborsky, O’Brien, & Woody, 1980; McLellan et al., 1988; McLellan et al., 1992) on a convenience sample of about 830 residential detox clients and 560 acupuncture clients at the central intake sites. The ASI measures severity along the following dimensions: medical, employment, alcohol, drugs, legal, family/social, and psychological/emotional. The only statistically significant difference in either severity scores (which are based on both objective client responses and subjective interviewer assessment) or composite scores (which are calculated solely from client responses), was the greater severity score for acupuncture clients along the psychological dimension. There were no differences in either alcohol severity or drug severity between the two groups, supporting the findings when our severity score was used. It is interesting to note that the group of acupuncture clients on whom the ASI was done had 6-month readmission rates similar to the group on whom the ASI was not done (p 5 .47). The group of residential detox clients on whom the ASI was done had poorer outcomes than those on whom the ASI was not completed (20% higher predicted probability of 6-month readmission, p , .001), perhaps suggesting the small sample of ASI completers among residential detox cases was a more severe group than the general population of residential detox cases. The extent to which outcomes depended on recent detox history is discouraging. Those clients who had not been admitted for detox within a year of their index admission were very unlikely to be readmitted to detox within 6 months, whatever their treatment. Those who had been admitted for detox in the last year had much higher readmission rates. Thus, it really matters less how the person is treated than what the client’s recent history is. This may reflect the particular population studied, an urban poor Value of Acupuncture Detox population utilizing the publicly funded treatment system, and the chronic, relapsing nature of the disease. It is interesting to note that once account is taken of the number of detox admissions in the last year, primary drug does not have a statistically significant impact on the likelihood of detox readmission. The effect of the different relapse rates associated with different primary substances is captured by recent relapse history. Though an acupuncture session is much less expensive than a day in a residential detox program, the longer duration of treatment reduces the potential economic benefit. We do not know “true” costs of acupuncture or residential treatment. However, based on Massachusetts reimbursement rates at the time of the study, an average residential detox stay is estimated to “cost” $850 to $900. The state pays $16.77 per unit of acupuncture service, which includes a visit for both acupuncture and group counseling. With an average duration of treatment of about 17 weeks, and assuming an average of 2.5 sessions per week after the 2 week daily visits for acute detox, the cost of acupuncture to the state is about $830, not much less than a week in residential treatment. However, a big advantage of acupuncture is that capacity can be expanded without adding inpatient services. An obvious limitation of this study is that clients were not randomized to modality. Rather, the modality to which clients were referred reflected both clinical assessment and patient preference. Discussions with central intake coordinators suggest that, for clients judged clinically suitable for acupuncture, it was necessary to educate the clients about acupuncture before they were willing to enter this modality. Thus, particularly for those without prior exposure to acupuncture, referral to acupuncture reflected to a large extent clinical judgment about appropriateness, followed by successful patient education about the potential value of acupuncture. The other major limitation of the study is with our outcome measure, readmission to detox. Though one would obviously like data on pre- and post-drug and alcohol usage, our measure is not unreasonable. For example, it has been shown that among homeless substance abusers, there were significant differences in ASI scores in five of the seven ASI dimensions among those readmitted for detox versus those who were not (Argeriou, McCarty, Mulvey, & Daley, 1994). This lends validity to the use of detox readmission as an outcome measure. Further, our outcome measure was useful in a study of the role of case management in substance abuse treatment (Shwartz, Baker, Mulvey, & Plough, 1997). And, readmission for substance abuse treatment has been used as an outcome measure in a study of the performance of inpatient substance abuse treatment programs in the Veterans Administration (Barnett & Swindle, 1997; Peterson, Swindle, Phibbs, Recine, & Moos, 1994). Although detox readmission is not synonymous with relapse, there is no a priori reason to expect that acupuncture versus residential detox clients might differentially seek detox 311 admission when they relapse. Thus, lack of data on actual relapse rates is not likely to seriously bias our results. As noted in the introduction, our study is of the value of outpatient detox programs that utilize acupuncture; it is not of the contribution that acupuncture makes to the outcomes associated with these programs. Though more focused observational studies and randomized trials are clearly needed, our study does suggests the value of outpatient acupuncture detox programs as a component of a substance abuse treatment system. This modality is particularly useful when residential detox beds are in short supply, for it allows some of the demand for detox to be met on an outpatient basis. REFERENCES Argeriou, M., McCarty, D., Mulvey, K., & Daley, M. (1994). Use of the Addiction Severity Index with homeless substance abusers. Journal of Substance Abuse Treatment, 11, 359–365. Ash, A.S., & Shwartz, M. (1997). Evaluating the performance of riskadjustment methods: Dichotomous outcomes. In L.I. Iezzoni (Ed.), Risk adjustment for measuring healthcare outcomes (2nd ed.) (427–469). Chicago: Health Administration Press. Barnett, P.G., & Swindle, R.W. (1997). Cost-effectiveness of inpatient substance abuse treatment. Health Service Research, 32, 615–629. Breslow, N.E., & Day, N.E. (1980). Statistical methods in cancer research: Vol. 1. The analysis of case-control studies. Lyon: International Agency for Research on Cancer. Brumbaugh, A.G. (1993). Acupuncture: New perspectives in chemical dependency treatment. Journal of Substance Abuse Treatment, 10, 35–43. Brumbaugh, A.G. (1994). Transformation and recovery: A guide for the design and development of acupuncture-based chemical dependency treatment programs. Santa Barbara, CA: Stillpoint Press. Bullock, M.L., Culliton, P.D., & Olander, R.T. (1989). Controlled trial of acupuncture for severe recidivistic alcoholism. The Lancet, 8, 1435–1439. Bullock, M.L., Umen, A.J., Culliton, P.D., & Olander, R.T. (1987). Acupuncture treatment of alcoholic recidivism: A pilot study. Alcoholism: Clinical and Experimental Research, 11, 292–295. Camp, J.M., Krakow, M., McCarty, D., & Argeriou, M. (1992). Substance abuse treatment management information systems: Balancing federal, state and service provider needs. The Journal of Mental Health Administration, 19, 5–20. Harrell, F.E., Jr., Lee, K.L., Califf, R.M., Pryor, D.B., & Rosati, R.A. (1984). Regression modeling strategies for improved prognostic prediction. Statistics in Medicine, 3, 143–152. Levin, J.S., Glass, T.A., Kushi, L.H., Schuck, J.R., Steele, L., & Jonas, W.B. (1997). Quantitative methods in research on complementary and alternative medicine: A methodological manifesto. Medical Care, 35, 1079–1094. Lipton, D.S., Brewington, V., & Smith, M. (1994). Acupuncture for crack-cocaine detoxification: Experimental evaluation for efficacy. Journal of Substance Abuse Treatment, 11, 205–215. McLellan, A.T., Grossman, D.S., Blaine, J.D., & Haverkos, H.W. (1993). Acupuncture treatment for drug abuse: A technical review. Journal of Substance Abuse Treatment, 10, 569–576. McLellan, A.T., Grossman, D.S., Blaine, J.D., & Haverkos, H.W. (1995). Letter to the editor: Dr. A. Thomas McLellan and colleagues’ response regarding evidence for effectiveness of acupuncture. Journal of Substance Abuse Treatment, 12, 141. McLellan, A.T., Kushner, H., Metzger, D., Rogers, P., Smith, I., Grissom, G., Pettinati, H., & Argeriou, M. (1992). The fifth edition of the Addiction Severity Index. Journal of Substance Abuse Treatment, 9, 199–213. 312 McLellan, A.T., Luborsky, L., Cacciola, J., Griffith, J., McGahan, P., & O’Brien, C.P. (1988). Guide to the Addiction Severity Index: Background, administration, and field testing results. Rockville, MD: Treatment Research Report, National Institute on Drug Abuse. McLellan, A.T., Luborsky, L., O’Brien, C.P., & Woody, G.E. (1980). An improved evaluation instrument for substance abuse patients: The Addiction Severity Index. Journal of Nervous and Mental Diseases, 168, 26–33. Peterson, K.A., Swindle, R.W., Phibbs, C.S., Recine, B., & Moos, R.H. (1994). Determinants of readmission following inpatient substance abuse treatment: A national study of VA programs. Medical Care, 32, 535–550. Rosenbaum, P.R., & Rubin, D.B. (1984). Reducing bias in observational studies using subclassification on the propensity score. Journal of the American Statistical Association, 79, 516–524. M. Shwartz et al. Shwartz, M., Baker, G., Mulvey, K.P., & Plough, A. (1997). Improving the publicly funded substance abuse treatment: The value of case management. American Journal of Public Health, 87, 1659–1664. Shwartz, M., Mulvey, K.P., Woods, D., Brannigan, P., & Plough, A. (1997). Length of stay as an outcome in an era of managed care: An empirical study. Journal of Substance Abuse Treatment, 14, 11–18. Smith, M. (1994). Letter to the editor. Journal of Substance Abuse Treatment, 11, 587. Washburn, A.M., Fullilove, R.E., Fullilove, M.T., Keenan, P.A., McGee, B., Morris, K.A., Sorensen, J.L., & Clark, W.W. (1993). Acupuncture heroin detoxification: A single-blind clinical study. Journal of Substance Abuse Treatment, 10, 345–351. Worner, T.M., Zeller, B., Schwarz, H., Zwas, F., & Lyon, D. (1992). Acupuncture fails to improve treatment outcome in alcoholics. Drug and Alcohol Dependence, 30, 169–173.