WK3CH20

Chapter 20 Linking Organizational Factors to Substance Abuse Treatment Outcomes: Multilevel Correlates of Treatment Effectiveness

Toorjo Ghose While substance abuse treatment is a vast and complex industry in the United States, few studies have examined the characteristics of this industry or its institutional processes (Kimberly & McLellan, 2006; Roman, Ducharme, & Knudsen, 2006). Reflecting the dearth of scholarship on organizational processes, research on substance abuse treatment effectiveness has focused exclusively on individual-level correlates of treatment outcomes (Etheridge, Hubbard, Anderson, Craddock, & Flynn, 1997; Hubbard et al., 1989; Schildhaus, Gerstein, Dugoni, Brittingham, & Cerbone, 2000; Sells, 1975; Simpson & Sells, 1982). With a few exceptions (Heinrich & Fournier, 2004; Hser, Anglin, & Fletcher, 1998; Sosin, 2002), organizational correlates of posttreatment substance use are unexamined. Orwin, Ellis, Williams, and Maranda (2000) point out that substance abuse treatment practice and policy will be better informed by understanding the link between organizational variables and treatment outcomes. Scholars have speculated that program policies, the quality of therapeutic staff, and the breadth of services provided by a facility substantially influence outcomes (Anglin & Hser, 1990; Ball & Ross, 1991; McGlothlin & Anglin, 1981). Consequently, researchers have emphasized the need to use multilevel models in examining program-level and individual-level processes in substance abuse treatment (Broome, Simpson, & Joe, 1999; Heinrich & Lynn, 2002; Hser, Joshi, & Anglin, 1999; Orwin & Ellis, 2000). This study answers the need to expand the universe of correlates of posttreatment use to include organizational variables, as well as factors associated with the organizational field of substance abuse treatment. It uses data from the Alcohol and Drug Services Study (ADSS) Survey, 1996–1999, a national study on substance abuse treatment conducted by the U.S. Department of Health and Human Services, Substance Abuse and Mental Health Services Administration (U.S. DHHS SAMHSA, 2002). Individual-level and organizational factors are simultaneously examined in a multilevel analysis of posttreatment illicit drug use. Addressing Research Gaps The lack of research on the link between organizational processes and individual-level treatment outcomes can be attributed to limitations of data sets (Heinrich & Fournier, 2004) and analytical methods (Yoo & Brooks, 2005). This research seeks to address both shortcomings. It uses data from the ADSS survey, which contains rich information on clients and their posttreatment behaviors, as well as the treatment facilities they attended. ADSS is the latest in a series of three national studies on substance abuse treatment clients and facilities. It builds on the previous two, the Drug Services Research Study conducted in 1990 and the Services Research Outcome Study, 1989–1995, by using a more complete sampling frame, incorporating an enhanced sampling design and including more detailed measures of facilities, treatment services, and clients in treatment (Office of Applied Studies [OAS], 2005). The data set allows an examination of a more comprehensive set of organizational and individual-level variables than have been undertaken in previous studies. Addressing the dearth of analytical methods to examine the multiple levels of treatment, this study uses hierarchical, logistical models that simultaneously examine individual-level and organizational influences on posttreatment substance use. These multilevel models account for the variance within facilities, as well as between them, resulting in more precise estimations of variance and accurate inferences (Roudenbush & Bryk, 2002).   Author's Note: I would like to thank the U.S. Substance Abuse and Mental Health Services Administration for making the ADSS data publicly available. I would like to thank Drs. Yeheskel “Zeke” Hasenfeld, Kim Blankenship, Rob Schilling, Oscar Grusky, Jeannette Ickovics, Todd Franke, and the CIRA NIMH fellows at Yale University for their helpful feedback. A version of this paper has been published in 2008 in the Journal of Substance Abuse Treatment, 38, 49–62. Background Conceptual Framework Scholars have argued that substance abuse service provision and treatment outcomes are affected by factors in the external political and economic environment of a facility, by internal program-level variables, and by client characteristics (Etheridge & Hubbard, 2000; Ghose, 2007; Heinrich & Fournier, 2004). Drawing on this work, this article uses a multilevel conceptual framework comprising external and internal organizational as well as individual-level factors to examine posttreatment substance use. It is assumed that factors at each of these levels influence the treatment process, thus ultimately affecting posttreatment outcomes. In identifying correlates of post-treatment use, factors examined include (a) those external to the organization that influence treatment decisions, (b) aspects of the internal program-level treatment technology, and (c) individual-level treatment characteristics. The External Organizational Field of Substance Abuse Treatment Pfeffer and Salancik (1978) argue that an organization's policies, norms, and practices are influenced by external agencies that it depends on for resources. Moreover, institutional theorists point out that organizational practices are shaped by an external community of agencies that an organization seeks legitimacy in (DiMaggio & Powell, 1983; Meyer & Rowan, 1977). Treatment-related decisions and processes are influenced by three types of actors in a treatment facility's external environment: funders, parental organizations, and regulatory agencies. Managed Care and the Private Sector. Managed-care organizations have significantly increased their role as a funding resource for substance abuse treatment facilities, through the years (Alexander, Lemak, & Campbell, 2003). Scholars have found that increased managed-care regulation decreases treatment intensity (Lemak & Alexander, 2001a, 2001b), reduces the number of services (Corcoran & Vandiver, 1996; Gold, Hurley, Lake, Ensor, & Berenson, 1995; Iglehart, 1996; Lo Sasso & Lyons, 2004; Olmstead, White, & Sindelar, 2004), limits autonomy of the provider (Alexander & Lemak, 1997; Mechanic, Schlesinger, & McAlpine, 1995; Schlesinger, Dorward, & Epstein, 1996; Schwartz & Wetzler, 1998), fails to increase technical efficiency in service provision (Alexander, Wheeler, Nahra, & Lemak, 1998), and increases relapse rates (Sosin, 2005). Scholars have found that service provision and treatment are adversely affected when a facility depends on the private sector rather than the government for its resources. For example, compared with public nonprofit facilities, private for-profit facilities were less likely to provide social and medical services (Friedmann, Durkin, Lemon, & D'Aunno, 2003). Given the constraints imposed on services and treatment by managed-care regulation and for-profit ownership, it is proposed that they will adversely affect posttreatment outcomes. Accreditation. Monitoring agencies have been found to increase the quality of treatment provided in treatment centers. Facilities accredited by the Joint Commission on the Accreditation of Healthcare Organizations (JCAHO) were more likely to provide primary care and mental health services (D'Aunno, 2006; Friedmann, Alexander, & D'Aunno, 1999), as well as physical exams and routine medical care (Durkin, 2002). Parent Organizations. An important element of an organization's external environment is its relationship with its parent facility. D'Aunno, Vaughn, and McElroy (1999) note that parent facilities tend to provide more financial and resource support to substance abuse facilities that adopt their methods, technologies, and ideologies. Support from a parent facility has been found to be associated with greater service provision to specialized populations (Ghose, 2007), whereas a lack of links with parental entities was associated with the provision of fewer services (Lee, Reif, Ritter, Levine, & Horgan, 2001), and in many cases, dissolution of the facility (Johnson & Roman, 2002). Given the supportive role JCAHO and parental organizations play in the provision of services, it is proposed that they will have beneficial effects on posttreatment outcomes. Hypotheses. It is hypothesized that elements of the external institutional environment will significantly influence treatment outcomes. Specifically, it is proposed that 1. managed-care regulation and for-profit ownership will increase the risk of posttreatment substance use, while JCAHO accreditation and parental organizational support will reduce the risk of resumption of drug use. Internal Program-Level Factors The modes of treatment, types of services, skill levels of service providers, and the distribution of workload constitute the treatment technology of addiction treatment programs. Treatment technology is therefore manifested in treatment modality, counselor expertise, scope of services provided, and caseload size. Modality. Scholars have found that compared with outpatient facilities, residential facilities were more effective in reducing relapse (Hubbard et al., 1989) as well as social problems and psychiatric symptoms (Guydish, Wardegar, Sorenson, Clarke, & Acampora, 1998). Hser, Evans, Huang, and Anglin (2001) found that in residential treatment, caseload size decreased the chances of retention and programmatic focus increased it, while in outpatient drug-free programs, group therapy focus decreased the odds of retention. Given the differential effects of factors in the two modalities, this research hypothesizes that in addition to a direct effect of modality on posttreatment use, modality will interact with other program factors in influencing outcomes. Degreed Counselors. D'Aunno et al. (1999) note that with the blending together of the community mental health and substance abuse treatment sectors, facilities are now influenced by two types of distinct ideologies: the mental health perspective and the Alcoholics Anonymous (AA) 12-step orientation to treatment. The mental health orientation is embedded in a psychological perspective with degreed professionals providing psychosocial treatment, whereas the AA perspective places emphasis on the role of recovering, often with nondegreed counselors providing the majority of treatment. The distinction in the type of counselors hired by a facility can have implications for client outcomes. Hser, Evans, et al. (2001) found that a greater proportion of counselors who were themselves recovering from addiction resulted in lower retention rates within methadone maintenance programs. It is proposed that a greater proportion of degreed counselors will result in better treatment outcomes for clients. Scope of Services. The breadth of services in a facility is a measure of its ability to provide comprehensive wrap-around care. Some researchers have found that a broader array of services were associated with longer stays in treatment and better treatment outcomes (Hoffman, Caudill, Koman, & Lucky, 1996; McLellan, Woody, & Metzger, 1996). Thus, treatment outcomes should be better in organizations that provide more services. Caseload and Facility Size. Staff time spent with clients can be expected to diminish with higher ratios of clients to staff members (D'Aunno et al., 1999), leading to lower rates of retention in treatment (Hser, Joshi, Maglione, Chou, & Anglin, 2001), decreased use of HIV testing and counseling services (D'Aunno & Vaughn, 1995; D'Aunno et al., 1999), as well as lower rates of medical services (Durkin, 2002; Friedmann et al., 1999) and employment counseling (Durkin, 2002). Larger facilities were associated with greater administrative and clerical hours worked by clinical staff members and thus reduced clinical productivity (Lemak, Alexander, & Campbell, 2003). Given their effect on time and attention spent with clients, this paper argues that higher caseloads and bigger facilities (i.e., greater likelihood of administrative requirements) will lead to poorer treatment outcomes. Hypotheses. It is proposed that the aspects of the treatment technology reviewed above will influence posttreatment outcomes. Specifically, it is hypothesized that 2. a greater proportion of professional counselors, lower client to staff ratio, smaller facility size, and a higher number of services offered will reduce posttreatment risk of substance use. Moreover, it is proposed that in addition to a direct effect on risk of use, treatment modality will modify the effects of other program-level factors. Individual-Level Correlates Length of Stay in Treatment. Several national studies have found that the time spent in treatment is the most consistent protective factor for clients in treatment. The Drug Abuse Reporting Program (DARP) 1969-1974 and the Service Research Outcomes Study (SROS) 1990 found that the length of treatment was negatively associated with the use of any drugs (Schildhaus et al., 2000; Simpson, 1981), while heroin use dropped with longer treatment stays in the Treatment Outcome Prospective Study (TOPS; Hubbard et al., 1989) and the Drug Abuse Treatment Outcome Study (DATOS) 1991-1993 (Simpson, Joe, & Broome, 2002). Interactions of Organizational Factors With Time Spent in Treatment. Simpson (2004) notes that length of treatment effects constitutes the cumulative effect of patient-level, therapeutic, and environmental factors. Attempting to unpack the effects of length of stay in treatment by examining its interaction with other variables, Simpson et al. (2002) found that longer treatment episodes benefited clients with problems of higher severity more than those with lower severity. In the only study examining the interaction of treatment length with organizational characteristics, a 5-year follow-up of DATOS clients found that length of stay reduced posttreatment use among residential but not outpatient clients (Hubbard, Craddock, & Anderson, 2003). It is proposed that in addition to its direct effect in preventing posttreatment substance abuse, length of stay in treatment will interact with organizational-level factors to influence treatment outcomes. Specifically, it is hypothesized that the organizational-level factors reviewed above, in influencing treatment processes, will modify the effect of length of stay in treatment on post-treatment substance use. Treatment Completion. While length of stay in treatment has often been used as a proxy measure of treatment completion (Wickizer et al., 1994), the two are related but different concepts (Sayre et al., 2002). Messina, Wish, and Nemes (2000) found that treatment completion significantly reduced drug use and increased employment. Individual-Level Controls. Several client-level factors have been found to be associated with treatment outcomes and need to be controlled for in an examination of posttreatment substance use. Drug Use. Several studies have found that regular use of drugs before treatment raised the risk of posttreatment substance use (Cushman, 1977; Hser et al., 1998; Hubbard et al., 1989; McLellan, Luborsky, & O'Brien, 1986; Schildhaus et al., 2000; Simpson et al., 2002). Client Demographics. Some studies indicate that compared with whites, African Americans entering treatment had lower retention rates (Agosti, Nunes, & Ocepeck-Welikson, 1996; McCaul, Svikis, & Moore, 2001; Milligan, Nich, & Carrol, 2004), more severe levels of addiction (Bernstein et al., 2005), and were more likely to use posttreatment (Schildhaus et al., 2000). Hispanics on the other hand were more likely to stay abstinent from any drug use, posttreatment (Bernstein et al., 2005). Researchers have found that women were more likely to be involved in treatment (Hser, Huang, Teruya, & Anglin, 2004) and less likely to use posttreatment (Hser et al., 2004; Hubbard et al., 1989). Age has been found to be correlated with illicit drug use (OAS, 2006), with older clients recording a greater reduction in posttreatment substance use (Brennan, Nichol, & Moos, 2003; Satre, Mertens, Arean, & Weisner, 2003). Criminal and Employment History. Pre-treatment criminal history was negatively associated with treatment completion (Knight, Logan, & Simpson, 2001) and positively associated with posttreatment substance use (Hubbard et al., 1989; Messina et al., 2000). Researchers have found that employment is associated with higher rates of treatment retention (Vendetti et al., 2002) and lower rates of posttreatment use (Leukefeld, McDonald, Staton, & Mateyoke-Scrivner, 2004; Platt, 1995; Wolkstein & Spiller, 1998). Mental Illness. Risk of posttreatment substance use was found to be higher for clients who suffered from antisocial personality (Grella, Joshi, & Hser, 2003; Hesselbrock, 1991; Woody, McLellan, Luborsky, & O'Brien, 1995) and depression (McKay et al., 2002; Ross, Glaser, & Germanson, 1988; Roy, Dejong, Lamparski, George, & Linnoila, 1991). Hypotheses. Given the individual-level treatment-related correlates of posttreatment use identified in the literature reviewed above, it is hypothesized that 3a. after controlling for pretreatment drug-use history, criminal activity, mental health, and client demographics, the time spent in treatment and treatment completion lower posttreatment risk of use. 3b. Moreover, it is proposed that the effect of length of stay will be modified by program-level factors, such as treatment modality, proportion of professional counselors, client-to-staff ratio, facility size, and scope of services offered as well as external organizational factors, such as managed-care funding, parent organizational support, for-profit ownership, and JCAHO accreditation (Figure 20.1). Methods Data and Sample ADSS 1996–1999, commissioned by the U.S. DHHS SAMHSA (2002), was conducted by the Schneider Institute for Health Policy, Brandeis University. It is a national longitudinal survey of substance abuse treatment facilities and clients. The sample for the study was constructed using a multistage sampling strategy. In Phase 1, substance abuse treatment facilities were selected using Stratified Random Sampling (SRS) procedures. Program-level data were collected from facility directors in 1996. Directors were mailed the questionnaire and requested to complete it in advance of a follow-up telephone interview using the same instrument. In Phase 2, SRS procedures were used to select a nationally representative subsample of 155 residential and outpatient nonmethadone treatment facilities. Records of 2,670 clients, who were eligible to be followed up with, were abstracted from these facilities to gather information on clients and their treatment. In Phase 3, 1,184 of these clients (48.5% response rate) from 128 facilities were interviewed in 1999 (for a full discussion of the sample, see also SAMHSA, 2003). Adjusting for Attrition Bias. Longitudinal studies with substance users usually suffer from significant attrition. Through 3 years, the attrition rate for this sample was 51.5%, which is similar to other large national longitudinal samples (Simpson et al., 2002). The advantage of a multistage sample such as the ADSS is that information on nonresponders collected at prior stages can be used to adjust weights. Using raking procedures that incorporate Phase 1 facility-level, Phase 2 abstract-level, and Phase 3 management-level data on nonresponsive clients, weights were adjusted in the ADSS to ensure a nationally representative sample (see also Ritter et al., 2003, for details). Figure 20.1 Conceptual Framework Variables and Measures Dependent Variable. Several past studies examining posttreatment outcomes use the dichotomous measure of any use post-treatment versus staying abstinent (Gossop, Stewart, Browne, & Marsden, 2002; Hser et al., 1999; Hubbard, Flyn, Craddock, & Fletcher, 2001; Walton, Castro, & Barrington, 1994). McLellan, Lewis, O'Brien, and Kleber (2000) point out that addiction is a chronic medical condition and question the appropriateness of measuring treatment success in terms of a “cure,” such as complete cessation of use. While this study uses a measure of “any use posttreatment” as the dependent variable, it seeks to make the measure more appropriate by interpreting it strictly as the risk of use, thereby highlighting the fact that it is a measure of vulnerability rather than “failure.” It is operationalized as the posttreatment risk of using any of the following drugs in the period from the end of treatment in 1996 to the 3-year follow-up interview in 1999: crack, cocaine, heroin, marijuana, amphetamines, hallucinogens, nonprescribed methadone, opiates, sedatives, tranquilizers, or nonmedical use of over-the-counter drugs. Organizational-Level Variables. All organizational data are taken from Phase 1 of the study. Regulation by managed-care organizations is operationalized as proportion of managed-care funding, calculated by dividing the amount of private managed-care funding by total revenue. Accreditation was operationalized as being accredited by the JCAHO. A dichotomous variable measured whether or not a facility operated for-profit and was an outpatient facility (with the reference category being residential). Facility size was calculated as the total number of staff members and clients in a program. Support by a parent organization was measured as an additive scale summing the number of the following dimensions of support provided by a parent organization: (1) personnel, (2) funding, (3) treatment protocols, (4) pricing of services, (5) waiting lists, (6) client intakes and assessments, and (7) quality assurance and utilization review. The presence of professional degreed counselors was operationalized as the proportion of counselors with doctorates. Number of services offered was the total number of services offered out of a menu of 18 services. Caseload was operationalized as the number of clients per full-time staff member. Treatment-Level Variables. Clinical information is taken from client records abstracted in Phase 2 of the study. Completion of treatment is operationalized as a dichotomous measure, while time spent in treatment is measured as a continuous variable. Individual-Level Controls. Data on client history and characteristics are taken from the Phase 3 follow-up interview. Pretreatment drug use is operationalized as the use of cocaine, heroin, marijuana, or intravenous drugs prior to treatment. Being African American, Hispanic, employed, or female, as well as the diagnosis of a mental illness and arrest for an illegal offense while in treatment are operationalized as dichotomous measures. Age is a continuous measure. All program-level variables are grand-mean centered, which controls for between-program differences in the centered variables (Roudenbush & Bryk, 2002, p. 142). Dichotomous variables were not centered, allowing regression coefficients to represent the difference in the odds of post-treatment use between the dichotomized groups. Models The Hierarchical Linear Modeling (HLM 6) software was used to analyze correlates of posttreatment use. The multilevel logistic models use the following three equations: Equation 1 regresses the likelihood of posttreatment substance use on individual-level predictors. Yij represents the log odds of posttreatment use, while βij represents the effect of i individual-level correlates within j organizations. Equation 2 represents the association between organizational correlates and the risk of use. It models the direct effect of organizational factors on the likelihood of posttreatment use. γ00 represents the mean likelihood of posttreatment use across all j organizations. U0j is the residual in the mean likelihood of use between the organizations. γ0j is the association between odds of posttreatment use and organizational-level correlates. Equation 3 models the association between organizational factors and the individual-level slopes. It models the extent to which organizational factors modify the association between individual-level factors (in this case, length of stay) and the likelihood of posttreatment use. γio represents the mean value of the slope of length of stay across all j organizations. Uij is the residual in the slope of length of stay between the organizations. γij is the effect of organizational predictors on the slope of length of stay. Five models were used to test the hypotheses of this research. The first model was an unconditional model with no independent variables that allowed the intercept to vary randomly across organizations. The second, a random slope and intercept model, included all hypothesized individual-level correlates and allowed the intercept as well as the slope of the length of stay to vary randomly across organizations in order to examine the between-organization variance in the odds of posttreatment use and the length of stay slope. In the third and fourth models, all hypothesized correlates at the individual and organizational level were included. The third model fixed the slope and the intercept, while the fourth model allowed the intercept to vary randomly across organizations. This allowed for a comparison of results between a true multilevel model and one that is essentially non-hierarchical. Finally, the fifth model retained all significant correlates. Results Table 20.1 presents the means of the variables included in the model. Among clients followed up, 47% reported that they had used illicit drugs, posttreatment. Table 20.2 presents the results of the unconditional random intercept model. The mean odds of posttreatment substance use is 0.86 and varies significantly across facilities (U0 = 0.361, p < .01). The next model examined the extent to which odds of posttreatment use and the effects of length of stay varied across facilities after all individual-level correlates were included. Table 20.2 presents the results of this random intercept and slope model. The residuals of the intercept (U0) and slope (U1) are significantly greater than 0 (p < .001 for both), thus indicating that the odds of posttreatment use and effect of length of stay fluctuate significantly from one facility to the next. These results provide support for the hypothesis that organizational context plays a significant role in clients using substances posttreatment and in modifying the effect of length of stay on posttreatment substance use. The variance of the odds of posttreatment substance use in this model (variance component = 0.227) is 37% less than the variance in the unconditional model, indicating that this model is better specified, with more of the variance accounted for, than the unconditional model. Table 20.3 presents the results of two models. Model 1 regressed posttreatment use onto all the correlates, without accounting for variance between facilities. In other words, this is a single-level regression model where organizational- and individual-level correlates are treated similarly. Model 2 allowed the intercept (or the mean odds of posttreatment use) to vary across organizations. It is a true multilevel model and takes into account the variance between and within organizations. All individual-level and organizational correlates were simultaneously entered into both models. The length of stay slope was regressed on each organizational factor sequentially, and variables with significant associations were retained in the model. Table 20.1 Means and Proportions Note: DOC, drug of choice; IV, intravenous; JCAHO, Joint Commission on the Accreditation of Healthcare Organizations; SEM, standard error of mean. Comparing Models 1 and 2. A comparison of the standard errors in Models 1 and 2 indicates that the errors in Model 1 are uniformly lower. This follows from the fact that unlike Model 2, this model does not account for the variance between organizations. Underestimating the standard error leads to inflated t values and incorrect inferences in Model 1. Thus, several factors in Model 1 emerge as significantly associated with the odds of use, although they are not significant correlates in Model 2. Being employed, Hispanic, as well as attending facilities with higher levels of parent monitoring are all significantly associated with posttreatment use in Model 1, but not in Model 2. Conclusions drawn from Model 1 would lead to a Type 1 error where the null hypothesis (that there is no significant association) is rejected incorrectly. Table 20.2 Unconditional and Random Intercept and Slope Models Results from Model 2 indicate that controlling for client demographics and pre-treatment client history, length of stay in treatment was significantly associated with the odds of posttreatment use. At the program level, managed care and JCAHO accreditation were associated with use, while parental facility monitoring, caseload size, managed-care funding, and proportion of counselors with a doctorate degree modified the effect of treatment episode length. Moreover, treatment modality significantly modified the effects of managed care and JCAHO accreditation. To build the most parsimonious model with the highest power, only significant correlates from Model 2 were retained in the final model. The results are presented in Table 20.4. The residual of the intercept continued to be significantly greater than 0 (χ2 = 184.08, p < .001), while the residual of the slope of length of stay was not (χ2 = 144.45, p = .09). Consequently, the intercept was allowed to vary randomly, while the slope was fixed. The inclusion of organizational variables explains the 27% more of the variance in the dependent variable than the model that included only individual-level variables and the 54% more of the variance than the unconditional model. With more of the variance in the odds of use accounted for, the estimate of the mean odds of posttreatment use (OR = 0.52, 95% CI = 0.39, 0.73) is more precise than the unconditional model and is significant at the .01 level. Thus, inclusion of organizational factors leads to a better specified model, indicating that they play a salient role in explaining posttreatment substance use. The relative strength of association in this model is indicated by the semistandardized coefficient. Managed care emerged as the strongest correlate, followed by age, preference for cocaine, JCAHO accreditation, preference for marijuana, commission of an arrestable offence, prior intravenous drug use, time spent in treatment, preference for heroin, and treatment completion, in that order. Risk factors included behaviors and personal characteristics. Intravenous drug use multiplied the risk of posttreatment use by more than a factor of two; committing an arrestable offence during treatment increased risk almost three times; while a preference for marijuana, cocaine, and heroin increased risk by factors of approximately 4.5, 3.8, and 3.6, respectively (all significant at the .01 level). Older clients were less likely to use posttreatment, with every standard unit increase in age reducing the risk by 47%(p < .01). Table 20.3 Multilevel Correlates of Posttreatment Substance Use Note: JCAHO, Joint Commission on the Accreditation of Healthcare Organizations; MCO, managed-care organization. a. Controlling for drug of choice, drug-use and criminal history, age, and being black and female. Table 20.4 Significant Correlates of Posttreatment Substance Use Note: DOC, drug of choice; IV, intravenous; JCAHO, Joint Commission on the Accreditation of Healthcare Organizations. a. Adjusted odds. b. Interaction effects reported after controlling for main effects of interacting variables. *p < .05; **p = < .01. Measures of the treatment process were found to be significantly associated with posttreatment use. Providing support for previous research findings, a standard unit increase in time spent in treatment decreased the risk by 27%, while completing treatment reduced it by 36% (both significant at the .01 level). Managed care emerged as the strongest direct correlate of posttreatment use. A standard unit increase in proportion of managed-care funding more than doubled the risk of posttreatment use (p < .01). Clients who were treated at a facility accredited by JCAHO were 46% less likely to use posttreatment compared with clients from non-JCAHO facilities. The effects of managed care and JCAHO accreditation were modified by treatment modality. Managed-care funding was less of a risk for clients in outpatient facilities—perhaps because there was less intervention at that level. For every standard unit increase in managed-care funding, the odds of posttreatment use for residential clients was 30% more than for the average outpatient client. On the other hand, JCAHO accreditation had less of a beneficial effect in outpatient programs. Whereas accredited residential facilities reduced the risk of posttreatment use by 65% compared with nonaccredited facilities, accreditation reduced the risk of use by 15% in outpatient facilities. Finally, organizational factors significantly modified the effect of length of stay in treatment. The ameliorative effect of longer length of stay was significantly lowered by some program-level factors. For example, a 10% increase in the level of parental organizational monitoring, caseload size, proportion of managed-care funding, and proportion of counselors with doctorates reduced the beneficial effect of length of stay by 7.5%, 2.5%, 5.7% (all significant at the .01 level), and 4.4% (p < .05), respectively. The results are consistent with the hypothesized effects of external organizational factors, such as managed-care funding and JCAHO accreditation, as well as individual-level factors, such as treatment length and treatment completion on posttreatment use. Moreover, the findings support the hypothesized interactions between modality and organizational variables, such as managed-care funding and JCAHO accreditation. Among the hypothesized organizational modifiers of the effects of length of stay, managed care, parent organizational monitoring, caseload size, and proportion of degreed counselors were significantly associated with the slope of length of stay. However, the reduction in the benefits of length of stay associated with higher proportions of counselors with a doctorate degree is contrary to the hypothesis that the presence of degreed counselors leads to better treatment outcomes. Moreover, contrary to the hypotheses, for-profit ownership and most aspects of the treatment technology at the program level had no direct effects on the risk of use. Discussion The findings (1) add to our understanding of how organizational factors, especially those originating in the external institutional environment, influence posttreatment use, (2) shed new light on how the same length of stay in treatment can have differential benefits in facilities with different organizational characteristics, and (3) highlight how hierarchical methods, by taking variance between organizations into account, avoid errors of inference, even for individual-level associations. The Influence of Organizational-Level Factors The significant variance of the intercept and slope across organizations, as well as results of the multilevel analysis, indicate that organizational factors in treatment design and delivery play a salient role in treatment outcomes. Specifically, external organizational factors appear to significantly influence posttreatment risk of use. The Salience of the External Institutional Environment. Monitoring and regulation by external institutions directly affect posttreatment risk of use and interact with other predictors. Managed-care regulation (as measured by greater levels of managed-care funding) increases the risk of use, while JCAHO accreditation decreases it. Moreover, the ameliorative effect of greater length of stay in treatment is reduced significantly when those treatment episodes are regulated and monitored through managed-care processes and/or by a parent organization. Results of this research indicate that substance abuse treatment facilities respond to three sources of institutional pressure in their external environment: (1) managed care, (2) parent organizations, and (3) accreditation agencies. The Influence of Managed Care. Researchers have found that the procedural algorithms used by managed-care organizations to approve treatment entry and duration, along with decreasing dollars per case or episode, have reduced services, treatment sessions, and completion rates (Alexander & Lemak, 1997; Durkin, 2002; Sederer & Bennett, 1996; Sosin, 2002). DiMaggio and Powell (1983) refer to these powerful institutional rules, which are enforced through the threat of sanctions, as comprising an “iron cage” that severely restricts the autonomy of a facility. Results of this research also suggest that the loss of autonomy associated with managed-care control may have adverse results on a client's ability to stay sober. The Influence of Parental Organization. Research shows that health care organizations do in fact respond to the demands of the economic environment by vertically integrating into networks under the umbrella of parental organizations (Longest, Rakich, & Darr, 2005; Walsh, 1998). Longest et al. (2005) note that in an integrated delivery system of this kind, each facility provides specialized care and services with the aim of providing seamless, wrap-around services to clients. However, such specialization can have negative outcomes for clients, as suggested by the results of this research. Specifically, vertical integration, as manifested through parent organizational administration, might undermine the effects of time in treatment insofar as facilities that provide a specialized (and therefore narrow) spectrum of care may exhaust their menu of treatment options for clients who stay longer. Moreover, heightened monitoring by parental entities may compromise an organization's ability to respond to its clients' needs over time. The Influence of JCAHO Accreditation. This is the first study to establish a direct association between accreditation and treatment outcome: Risk of posttreatment use is reduced for clients being treated in JCAHO-accredited facilities. Scholars have found that JCAHO accreditation, which is based on a triennial evaluation of facilities (JCAHO, 2006), is correlated with increased staff productivity (Lemak et al., 2003) and a greater breadth of service provision (Durkin, 2002; Friedmann et al., 1999). These organizational outcomes of accreditation translate into substantial benefits for clients, given that accreditation is associated with a significant reduction in the risk of posttreatment use. DiMaggio and Powell (1983) note that organizational processes become uniform and standardized in a given occupation as a result of processes that are coercive, normative, and mimetic. Whereas managed care and parental organizations influence facilities through coercive processes, JCAHO's vector of influence cannot follow the same route, as accreditation is voluntary in the field of substance abuse treatment. With only a minority of facilities (31%) choosing to become accredited, JCAHO accreditation has yet to become a norm in the substance abuse treatment field. The results of this research indicate that establishing accreditation as a professional standard would loosen some of the bars of the iron cage imposed by managed-care facilities and parent organizations and would lead to beneficial outcomes for clients. Interactions With Modality. JCAHO accreditation was more beneficial for clients in residential programs than it was for clients in outpatient facilities, while managed-care funding was more detrimental for those residential programs. These results indicate that on the policy front, bringing residential facilities under the purview of JCAHO accreditation and limiting the role of managed-care regulation might have benefits for clients and might be a more feasible step than doing so for the larger universe of treatment facilities. However, it is not known what effects such a policy would have on the costs of care. Merging the Public and Private Sectors. Gronbjerg (1993, 1998) notes that privatization efforts and the penetration of managed care have blurred the line between the public and private economies in the human services industry. The contradictory mandates introduced by the merger of economies have implications for treatment outcomes. The profit maximization mandate of managed care, for instance, has detrimental effects on posttreatment use, whereas JCAHO, charged with the public mandate to ensure quality services, ensures beneficial outcomes. The results indicate that client outcomes might be compromised if the private sector continues to take over the management of substance abuse treatment. Treatment-Level Factors The Modification of the Benefits of Extended Treatment Duration. Length of stay in treatment has been found to be one of the most consistent protective factors in posttreatment outcomes (Simpson, 2004). The results of this research also indicate that time spent in treatment is an effective deterrent to posttreatment substance use. However, its ameliorative effect appears to be modified by organizational factors. For example, there appears to be less benefit from longer treatment episode length in the presence of increased managed-care regulation and/or parental organizational administration. The mechanism by which this might occur is unknown, but it is possible that increased treatment dosage loses its effect when treatment regimens are rendered inflexible through external institutional control. Proportion of counselors with a doctorate and caseload size, both aspects of the treatment technology of facilities, also appear to modify the influence of length of stay in treatment. Higher caseloads seem to reduce the time that treatment providers can spend with clients (D'Aunno et al., 1999), thus reducing the beneficial effects of increased dosage, as manifested in longer treatment episodes. The interaction between length of stay and proportion of degreed counselors is in the opposite direction of the hypothesized association: The presence of counselors with doctorate degrees appears to reduce the benefits of length of stay. The finding is probably an artifact of anomalous data in this sample. It is possible, however, that the results indicate the differential manner in which professional and nonprofessional counselors work with clients. Counselors without professional counseling degrees may be more likely to be recovering individuals who favor a 12-step total abstinence perspective (D'Aunno et al., 1999). Prolonged treatment episodes, where more time is spent with clients, may increase the effectiveness of such modeling and might allow the 12-step ideology to be modeled effectively by recovering counselors and take root in clients. In contrast, professional counselors with doctorates tend to be embedded in a mental health perspective, delivering treatment within a more traditional, psychological framework where face- to-face contact is limited to sessions. Given this difference, time spent around counselors oriented to a 12-step abstinence-only approach might be more effective in keeping clients abstinent than time spent in a program with professional, degreed counselors. While retention of clients and expansion of the length of their treatment episodes have become major objectives for substance abuse treatment agencies, these results indicate that such decisions need to factor in organizational characteristics. Expanding treatment length may have diminishing value in the presence of increased levels of managed-care regulation and funding constraints; and/or greater administrative monitoring and control by parent organizations; and perhaps greater proportion of staff with doctorates. Treatment Completion. The results indicate that even after controlling for time spent in treatment, a favorable treatment discharge (completion) reduced risk of use, post-treatment. Researchers have found that self-efficacy is boosted by treatment, with abstainers reporting higher self-efficacy scores than users (Burling, Reilly, Molten, & Ziff, 1989). The role of treatment completion in posttreatment use points to the fact that after controlling for time in treatment, completion of treatment appears to decrease posttreatment use by possibly increasing self-efficacy among users. The Utility of Hierarchical Models The use of multilevel models to examine posttreatment use adds to the repertoire of analytical tools to examine posttreatment outcomes and has two major benefits. First, both program-level as well as individual-level correlates can be examined in one model. Second, the results indicate that the high level of intraclass correlation between subjects in treatment facilities has to be accounted for, even when examining individual-level correlates of posttreatment use. As a comparison of the two models in Table 20.3 indicates, if the differences between programs are not accounted for through hierarchical methods, the significance of individual-level associations is inflated, leading to Type I error. In a non-hierarchical model (Table 20.3, Model 1), employment and race (Hispanic) on the individual level and parental organizational monitoring at the program level emerge as significant correlates. However, the significance of these associations disappear in a multilevel model. Previous studies, especially those focusing exclusively on individual-level associations, may suffer from such errors of inference. Practice, Policy, and Future Research Implications The identification of correlates of post-treatment risk of use can inform treatment and policies that seek to lower the chronicity of addiction and possibly lengthen the time between treatment admissions. The results indicate that facilities may be best served by implementing feasible treatment plans that raise the likelihood of treatment completion and that retain clients for longer periods of time. However, efforts to lengthen treatment episode may have to take into account the constraints on a facility's autonomy exerted by external agencies, as well as the nature of a facility's treatment technology. More is not necessarily better in the case of facilities that need to adhere to managed care and parent organizational rules and are relatively understaffed. The same is true for facilities that depend on counselors with doctorates for treatment provision, if the counterintuitive association between length of stay in treatment and levels of professional expertise among counselors can be confirmed in future research and is not attributable to anomalous data in this research. Decision-making processes in facilities need to factor in organizational effects on treatment outcomes. The results strike a cautionary note on allowing managed-care funding to become a part of organizational budgets. Given the new evidence that this research presents on JCAHO accreditation's association with lower risk of posttreatment use, making accreditation mandatory might result in significant benefits for clients in treatment, especially in residential facilities. Given the significant differences between facilities, highlighted by the results of this research, future studies on treatment efficacy need to factor in the variance between treatment sites, even when examining individual-level correlates of outcomes of treatment being administered across facilities. The results of this research highlight the importance of collecting data on the individual and program level as well as using multilevel methods to examine the effects of substance abuse treatment. Limitations of the Study Attrition. Similar to the DARP and DATOS studies (Simpson et al., 2002), there was significant attrition at the 3-year follow-up. While results have been adjusted for nonresponse using data collected at prior stages, generalizations of the results should be made with caution. Despite the attrition (51.5%), the sample remains large at follow-up, ensuring robustness of the results. Datedness of the Data. Considerable changes have occurred in the substance abuse treatment landscape since the data for the study were collected. Managed care, for instance, has continued to expand its presence in substance abuse treatment facilities' budgets (Alexander & Lemak, 1997; D'Aunno, 2006; Drug and Alcohol Service Information System [DASIS], 2001), and it is possible that its adverse effects on treatment outcomes has intensified. Current studies need to replicate the ADSS study and examine the changes that have occurred since 1999. Conclusion This research highlights the link between macro- and microlevel processes in substance abuse treatment by examining program-level and individual-level correlates of posttreatment substance use. The findings highlight the role of the organization in influencing posttreatment use. The multilevel analytical framework provides the field of substance abuse research with a useful analytical tool to study treatment outcomes. The results of this research lay the foundation for further examination of multilevel predictors of posttreatment outcomes. References Agosti, V., Nunes, E., & Ocepeck-Welikson, K. (1996). Patient factors related to early attrition from an outpatient cocaine research clinic. American Journal of Drug and Alcohol Abuse, 22, 29–39. Alexander, J. A., & Lemak, C. H. (1997). The effects of managed-care administrative burden in outpatient substance abuse treatment organizations. Medical Care Research and Review, 54, 490–507. Alexander, J. A., Lemak, C. H., & Campbell, C. I. (2003). Changes in managed-care activity in outpatient substance abuse treatment organizations, 1995–2000. Journal of Behavioral Health Services Research, 30, 369–381. Alexander, J. A., Wheeler, J. A., Nahra, T. A., & Lemak, C. H. (1998). Managed care and technical efficiency in outpatient substance abuse treatment units. Journal of Behavioral Health Services and Research, 25, 377–396. Anglin, M. D., & Hser, Y. (1990). Treatment of drug abuse. In M. Tonry & J. Q. Wilson (Eds.), Drugs and crime (pp. 393–458). Chicago: University of Chicago Press. Ball, J. C., & Ross, A. (1991). The effectiveness of methadone maintenance treatment: Patients, programs, services and outcome. New York: Springer-Verlag. Bernstein, E., Bernstein, J., Tassiopoulos, K., Valentine, A., Heeren, T., Levenson, S., et al. (2005). Racial and ethnic diversity among a heroin and cocaine using population: Treatment system utilization. Journal of Addictive Diseases, 24 (4), 44–63. Brennan, P. L., Nichol, A. C., & Moos, R. H. (2003). Older and younger patients with substance use disorders: Outpatient mental health service use and functioning over a 12-month interval. Psychology of Addictive Behaviors, 17 (1), 42–48. Broome, K. M., Simpson, D. D., & Joe, G. W. (1999). Patient and program attributes related to treatment process indicators in DATOS. Drug and Alcohol Dependence, 7, 127–135. Burling, T. A., Reilly, P. M., Molten, J. O., & Ziff, D. (1989). Self-efficacy and relapse among inpatient drug and alcohol abusers: A predictor of outcome. Journal of Studies on Alcohol and Drugs, 50 (4), 354–360. Corcoran, K. J., & Vandiver, V. (1996). Maneuvering the maze of managed care: A manual for mental health practitioners. New York: Free Press. Cushman, P. (1977). Ten years of methadone maintenance treatment: Some clinical observations. American Journal of Drug and Alcohol Abuse, 4, 543–553. D'Aunno, T. (2006). The role of organization and management in substance abuse treatment: Review and roadmap. Journal of Substance Abuse Treatment, 31, 221–233. D'Aunno, T., & Vaughn, T. E. (1995). An organizational analysis of service patterns in outpatient drug abuse treatment units. Journal of Substance Abuse, 7, 27–42. D'Aunno, T., Vaughn, T. E., & McElroy, P. (1999). An institutional analysis of HIV prevention efforts by the nation's outpatient drug abuse treatment units. Journal of Health and Social Behavior, 40, 175–192. DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48, 147–160. Drug and Alcohol Service Information System. (2001). Growth of managed care in substance abuse treatment. Arlington, VA: Substance Abuse and Mental Health Services Administration. Retrieved June 30, 2006, from http://oas.samhsa.gov/2k1/managedcare/managedcare.htmDurkin, E. M. (2002). An organizational analysis of psychosocial and medical services in outpatient drug abuse treatment programs. Social Service Review, 76, 406–429. Etheridge, R. M., & Hubbard, R. L. (2000). Conceptualizing and assessing treatment structure and process in community-based drug dependency treatment programs. Substance Use and Misuse, 55, 1757–1795. Etheridge, R. M., Hubbard, R. L., Anderson, J., Craddock, S. G., & Flynn, P. M. (1997). Treatment structure and program services in the Drug Abuse Treatment Outcome Study. Psychology of Addictive Behaviors, 11, 244–260. Friedmann, P. D., Alexander, J. A., & D'Aunno, T. (1999). Organizational correlates of access to primary care and mental health services in drug abuse treatment units. Journal of Substance Abuse Treatment, 16, 71–80. Friedmann, P. D., Durkin, E., Lemon, S. C., & D'Aunno, T. A. (2003). Delivery of ancillary services in outpatient substance abuse treatment programs. Journal of Substance Abuse Treatment, 24, 81–88. Ghose, T. (2007). Organizational correlates of specialized substance abuse treatment provision to seropositive clients: A political economy perspective. Journal of HIV/AIDS & Social Services, 5, 41–59. Gold, M. R., Hurley, R., Lake, T., Ensor, T., & Berenson, R. (1995). A national survey of the arrangements managed-care plans make with physicians. Journal of Medicine, 333(25), 1678–1683. Gossop, M., Stewart, D., Browne, N., & Marsden, J. (2002). Factors associated with abstinence, lapse or relapse to heroin use after residential treatment: Protective effect of coping responses. Addiction, 97, 1259–1267. Grella, C. E., Joshi, V., & Hser, Y. (2003). Followup of cocaine-dependent men and women with antisocial personality disorder. Journal of Substance Abuse Treatment, 25, 155–164. Gronbjerg, K. A. (1993). Understanding nonprofit funding: Managing revenues in social services and community development organizations. San Francisco: Jossey-Bass. Gronbjerg, K. A. (1998). Markets, politics and charity: Nonprofits in the political economy. In W. W. Powell & E. S. Clemens (Eds.), Private action and the public good (pp. 137–149). New Haven, CT: Yale University Press. Guydish, J., Wardegar, D., Sorenson, J. L. Clarke, W., & Acampora, A. (1998). Drug abuse day treatment: A randomized clinical trial comparing day and residential treatment programs. Journal of Consulting and Clinical Psychology, 66, 280–289. Heinrich, C. J., & Fournier, E. (2004). Dimensions of publicness and performance in substance abuse treatment organizations. Journal of Policy Analysis and Management, 23, 49–63. Heinrich, C. J., & Lynn, L. E. (2002). Improving the organization, management, and outcomes of substance abuse treatment programs. American Journal of Drug and Alcohol Abuse, 28, 601–622. Hesselbrock, M. (1991). Gender comparison of antisocial personality disorder and depression in alchoholism. Journal of Substance Abuse, 3, 205–219. Hoffman, J. A., Caudill, B. D., Koman, J. J., & Lucky, J. W. (1996). Psychosocial treatments for cocaine abuse: 12-month treatment outcomes. Journal of Substance Abuse Treatment, 13, 3–11. Hser, Y., Anglin, M. D., & Fletcher, B. (1998). Comparative treatment effectiveness: Effects of program modality and client drug dependence history on drug use reduction. Journal of Substance Abuse Treatment, 15, 513–523. Hser, Y., Evans, E., Huang, D., & Anglin, M. D. (2001). Relationship between drug treatment services, retention and outcomes. Psychiatric Services, 55, 767–774. Hser, Y., Huang, Y., Teruya, C., & Anglin, M. D. (2004). Gender differences in treatment outcomes over a three-year period. Journal of Drug Issues, 34, 419–439. Hser, Y., Joshi, V., & Anglin, D. M. (1999). Predicting posttreatment cocaine abstinence for first-time admissions and treatment repeaters. American Journal of Public Health, 89, 666–671. Hser, Y., Joshi, V., Maglione, M., Chou, C., & Anglin, M. D. (2001). Effects of program and patient characteristics on retention of drug treatment patients. Evaluation and Program Planning, 24, 331–341. Hubbard, R. L., Craddock, S. G., & Anderson, J. (2003). Overview of 5-year followup outcomes in the drug abuse treatment outcome studies. Journal of Substance Abuse Treatment, 23, 125–134. Hubbard, R. L., Flyn, P. M., Craddock, S. G., & Fletcher, B. W. (2001). Relapse after drug abuse treatment. In M. Tims, C. G. Leukefeld, & J. J. Platt (Eds.), Relapse and recovery in addictions (pp. 109–121). New Haven, CT: Yale University Press. Hubbard, R. L., Marsden, M. E., Rachal, J. V., Harwood, H. J., Cavanaugh, E. R., & Ginzburg, H. M. (1989). Drug abuse treatment: A national study of effectiveness. Chapel Hill: University of North Carolina Press. Iglehart, J. K. (1996). Managed care and mental health. Health Policy Report, 334, 131–135. Johnson, J. A., & Roman, P. M. (2002). Predicting closure of private substance abuse treatment facilities. Journal of Behavioral Health Services and Research, 29, 115–125. Joint Commission on the Accreditation of Healthcare Organizations. (2006). Comprehensive accreditation manual for behavioral health care. Oak Brook, IL: Joint Commission Resources. Retrieved July 31, 2006, from www.joincommission.org Kimberly, J. R., & McLellan, T. A. (2006). The business of addiction treatment: A research agenda. Journal of Substance Abuse Treatment, 31, 213–219. Knight, D. K., Logan, S. M., & Simpson, D. D. (2001). Predictors of program completion for women in residential substance abuse treatment. American Journal of Drug and Alcohol Abuse, 27, 1–18. Lee, M. T., Reif, S., Ritter, G. A., Levine, H., & Horgan, C. M. (2001). Access to services in the substance abuse treatment system: Variations by facility characteristics. In M. Galanter (Ed.), Recent developments in alcoholism (Vol. 15, pp. 137–156). New York: Kluwer Academic. Lemak, C. L., & Alexander, J. A. (2001a). Managed care and drug treatment practices: A model of organizational response to external influence. Advances in Health Care Management, 2, 131–159. Lemak, C. L., & Alexander, J. A. (2001b). Managed care and outpatient substance abuse treatment intensity. Journal of Behavioral Health Services and Research, 28, 12–29. Lemak, C. H., Alexander, J. A., & Campbell, C. (2003). Administrative burden and its implications for outpatient substance abuse treatment organizations. Psychiatric Services, 54, 705–711. Leukefeld, C., McDonald, S. H., Staton, M., & Mateyoke-Scrivner, A. (2004). Employment, employment-related problems, and drug use at drug court entry. Substance Use and Misuse, 39, 2559–2579. Longest, B., Rakich, J. S., & Darr, K. (2005). Managing health service organizations. Baltimore: Health Professionals Press. Lo Sasso, A., & Lyons, J. S. (2004). The sensitivity of managed care treatment intensity to co-payment levels. Journal of Behavioral Health Services and Research, 31, 50–65. McCaul, M. E., Svikis, D., & Moore, R. D. (2001). Predictors of outpatient treatment retention: Patient versus substance use characteristics. Drug and Alcohol Dependence, 62, 9–17. McGlothlin, W. H., & Anglin, D. M. (1981). Long-term followup of clients of high- and low-dose methadone programs. Archives of General Psychiatry, 38 (9), 1055–1063. McKay, J. R., Pettinati, H. M., Morrison, R., Feeley, M., Mulvaney, F. D., & Gallop, R. (2002). Relation of depression diagnoses to 2-year outcomes in cocaine-dependent patients in a randomized continuing care study. Psychology of Addictive Behaviors, 16, 225–235. McLellan, T. A., Lewis, D., O'Brien, C., & Kleber, H. (2000). Drug dependence, a chronic mental illness: Implications for treatment, insurance, and outcomes evaluation. Journal of the American Medical Association, 284, 1689–1695. McLellan, T. A., Luborsky, L., & O'Brien, C. P. (1986). Alcohol and drug abuse treatment in three different populations: Is there improvement and is it predictable? American Journal of Drug and Alcohol Abuse, 12, 101–120. McLellan, T. A., Woody, G. E., & Metzger, D. (1996). Evaluating the effectiveness of addiction treatments: Reasonable expectations, appropriate comparisons. Milbank Quarterly, 74, 51–85. Mechanic, D., Schlesinger, M., & McAlpine, D. (1995). Management of mental health and substance abuse services: State of the art and early results. Milbank Quarterly, 73, 19–55. Messina, N., Wish, E., & Nemes, S. (2000). Predictors of treatment outcomes for men and women admitted to a therapeutic community. American Journal of Drug and Alcohol Abuse, 26 (2), 207–227. Meyer, J. W., & Rowan, B. (1977). Institutionalized organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83, 340–363. Milligan, C. O., Nich, C., & Carroll, K. M. (2004). Ethnic differences in substance abuse treatment retention, compliance, and outcome from two clinical trials. Psychiatric Services, 55, 167–173. Office of Applied Studies. (2005). Alcohol and drug services study. Retrieved July 31, 2006, from http://oas.samhsa.gov/adss.htm# Methods Office of Applied Studies. (2006). Prevalence of substance abuse among racial and ethnic subgroups in the U.S. Washington, DC: U.S. Department of Health and Human Services, Substance Abuse and Mental Health Service Administration. Olmstead, T., White, W. D., & Sindelar, J. (2004). The impact of managed care on substance abuse treatment services. Health Services Research, 39, 319–343. Orwin, R., & Ellis, B. (2000). The national treatment improvement evaluation study: Multilevel reanalysis of treatment outcomes. Fairfax, VA: Center for Substance Abuse Treatment, National Evaluation Data and Technical Assistance Center. Orwin, R. G., Ellis, B., Williams, V., & Maranda, M. (2000). Relationships between treatment components, client-level factors and positive treatment outcomes. Journal of Psychopathology and Behavioral Assessment, 22, 383–397. Pfeffer, J., & Salancik, G. R. (1978). The external control of organizations: A resource dependence perspective. New York: Harper &Row. Platt, J. J. (1995). Vocational rehabilitation of drug abusers. Psychological Bulletin, 117, 416–433. Ritter, G. A., Carusi, C., Krenzke, T., Legum, S. E., Mohadjer, L., Reif, S., et al. (2003). Phase III methodology: Client follow-up study. In Alcohol and drug services study (ADSS) methodology report: Phases I, II, and III. Rockville, MD: Substance Abuse and Mental Health Services Administration, Office of Applied Studies. Retrieved July 31, 2006, from www.oas.samhsa.gov/ADSS/ADSSmethods.pdf Roman, P. M., Ducharme, L. J., & Knudsen, H. K. (2006). Patterns of organization and management in private and public substance abuse treatment programs. Journal of Substance Abuse Treatment, 31, 235–243. Ross, H. E., Glaser, F. B., & Germanson, T. (1988). The prevalence of psychiatric disorders in patients with alcohol and other drug problems. Archives of General Psychiatry, 45, 1023–1031. Roudenbush, S. W., & Bryk, A. S. (2002). Hierarchical linear models: Applications and data analysis methods. Thousand Oaks, CA: Sage. Roy, A., DeJong, J., Lamparski, D., George, T., & Linnoila, M. (1991). Depression among alcoholics: Relationship to clinical and cerebrospinal fluid variables. Archives of Psychiatry, 48, 28–32. Satre, D. D., Mertens, J., Arean, P. A., & Weisner, C. (2003). Contrasting outcomes of older versus middle-aged and younger adult chemical dependency patients in a managed care program. Quarterly Journal of Studies on Alcohol, 64, 520–530. Sayre, S. L., Schmitz, J. M., Stotts, A. L., Averill, P. M., Rhoades, H. M., & Grabowski, J. J. (2002). Determining predictors of attrition in an outpatient substance abuse program. American Journal of Drug and Alcohol Abuse, 28, 55–72. Schildhaus, S., Gerstein, D., Dugoni, B., Brittingham, A., & Cerbone, F. (2000). Service research outcome study: Explanation of treatment effectiveness, using individual-level and programmatic variables. Substance Use and Misuse, 35, 1879–1910. Schlesinger, M., Dorward, R., & Epstein, S. (1996). Managed care constraints in psychiatrists' hospital practices: Bargaining power and professional autonomy. American Journal of Psychiatry, 153, 256–260. Schwartz, B. J., & Wetzler, S. (1998). A new approach to managed care: The provider-run organization. Psychiatric Quarterly, 69, 345–354. Sederer, L. I., & Bennett, M. J. (1996). Managed mental health care in the United States: A status report. Administration and Policy in Mental Health and Mental Health Services Research, 23, 289–307. Sells, S. B. (1975). The DARP research program and data system. American Journal of Drug and Alcohol Abuse, 2 (1), 1–14. Simpson, D. D. (1981). Treatment for drug abuse: Follow-up outcomes and length of time spent. Archives of General Psychiatry, 38 (8), 875–880. Simpson,