Adoption and implementation of new technologies in substance abuse treatment

Transcription

1 Journal of Substance Abuse Treatment 22 (2002) Regular article Adoption and implementation of new technologies in substance abuse treatment Paul M. Roman, Ph.D. a,b, *, J. Aaron Johnson, Ph.D. a a Institute for Behavioral Research, The University of Georgia, Athens, GA 30602, USA b Department of Sociology, The University of Georgia, Athens, GA 30602, USA Received 9 November 2001; received in revised form 20 February 2002; accepted 26 February 2002 Abstract In addition to clinical outcomes, understanding the adoption and implementation of new treatment interventions is essential. This analysis was designed to assess the predictive utility of organization-level features in understanding the adoption and implementation of new technologies in substance abuse treatment. Naltrexone, which was found to be in current use in 44.1% of a national sample of 400 private substance abuse treatment centers, was selected as an appropriate sample technology for study. Adoption of naltrexone is significantly related to both the treatment center s age and its administrative leadership. Naltrexone adoption is also significantly associated with the percentage of the center s caseload covered by managed care programs and by the percentage of relapsers represented in the caseload. The analysis was less successful in predicting naltrexone implementation for either primary alcohol dependence or primary opiate addiction. D 2002 Elsevier Science Inc. All rights reserved. Keywords: Naltrexone; Technology transfer; Substance abuse treatment; Organizational change 1. Introduction The transfer of new treatment techniques into the practice of substance abuse treatment is an issue of central concern as the twenty-first century begins. Many dimensions of the research to practice gap have been widely discussed (Lamb, Greenlick, & McCarty, 1998). These include the extent to which the institutionalized patterns of group therapy and 12-step principles can be successfully supplemented, altered and/or enhanced by the introduction of new treatments. Further is the resistance to adoption of new treatment practices among those who have been intensely socialized into the extant treatment techniques and feel both personally identified and strongly committed to these practices. A final dimension centers on the systems within which new forms of treatment and treatment practice must be introduced, namely the organizations within which substance abuse services are delivered. This latter focus is the topic of this article. * Corresponding author. 101 Barrow Hall, Institute for Behavioral Research, The University of Georgia, Athens, GA 30602, USA. Tel.: ; fax: address: proman@arches.uga.edu (P.M. Roman). This analysis examines the features of the organizational structure, as well as information about the center s leadership and caseload as possible predictors of the adoption and implementation of an innovation. Though a number of new technologies including both pharmaceutical and therapy-based interventions could have been selected, this article specifically examines the adoption and implementation of naltrexone in the treatment of substance abuse problems. Naltrexone s approval by the FDA in 1994 makes it a relatively new innovation, but one that has been available to clinicians long enough to expect at least some degree of diffusion. The analysis is focused not only on whether centers use naltrexone in treatment (adoption), but also on the extent to which naltrexone is used among potentially eligible treatment patients (implementation). Organizational structure, leadership and caseload characteristics are features of substance abuse treatment organizations that are readily identifiable without conducting extensive internal research within these organizations. Thus, an understanding of these factors may prove useful to external change agents charged with promoting technology transfer, and needing to readily define the differential receptivity of their targets without intensively examining each target organization. Obviously there are many other /02/$ see front matter D 2002 Elsevier Science Inc. All rights reserved. PII: S (02)

2 212 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) organizational factors affecting adoption and implementation of innovations other than the aforementioned domains. Simpson (2002, in this issue) outlines a comprehensive framework for a program change model. Simpson identifies four stages of transfer: exposure, adoption, implementation, and practice; and argues a number of factors affect these stages. These factors include individual background features and experiences that predispose different attitudes toward innovations among clinicians, the availability of sufficient institutional resources, and various features of the cultures and subcultures of treatment organizations. Likewise, there may be external influences from funding sources, competing centers and other organizations with which the center has a relationship. Though the presence of additional factors is acknowledged, the focus here is upon those surface characteristics that are easy to identify, and whose measurement is usually unambiguous. As a substance abuse treatment innovation, naltrexone is of particular interest for several reasons. First, its action as an antagonist in reducing the rewarding aspects of drug use represents a paradigm that is distinctive from methadone and related maintenance drugs. Second, it has a uniquely broad application on certain segments of the substancedependent population. Its initial use was associated with some success as an opiate antagonist, though there were evident problems with motivating patients to continue to use naltrexone (Terenius, 1998). Likewise, naltrexone was shown to have promise in treating alcohol dependence (Volpicelli, Alterman, Hayashida, & O Brien, 1992; O Malley, Jaffe, Chang, Schottenfeld, & Rounsaville, 1992), though its effectiveness in treating alcoholics in inpatient settings has been recently called into question (Knox & Donovan, 1999; Krystal, Cramer, Krol, & Rosenheck, 2001). Third, naltrexone is not necessarily a disruptive innovation that displaces existing treatment strategies. There is evidence its use can be integrated with existing treatment paradigms, with its effectiveness appearing greatest when combined with traditional group therapy, individual counseling and relapse prevention (Jaffe et al., 1996). Finally, while there is a need for a greater amount of research on long-term effects, the effectiveness of naltrexone as a treatment modality leading toward lasting recovery appears very promising (National Institute on Drug Abuse, 1999). A review of the extant literature on the diffusion of the use of naltrexone in treatment revealed only two published studies. Thomas et al. (2001) examined samples of physicians and nonphysicians counselors from Massachusetts, Tennessee, and Washington. The study found 60% of physicians in the practice of addiction medicine prescribed naltrexone at least occasionally, while 55% of nonphysician counselors had never recommended it. The study revealed that among the most important barriers to naltrexone use were lack of knowledge about it and beliefs about its cost. The study did however find there was a generalized receptivity to using medications in substance abuse treatment within both the physician and nonphysician samples. By contrast, Forman, Bovasso, and Woody s (2001) study of 317 substance abuse treatment clinicians found overwhelming support (82.1%) for increased use of 12-step programs, but only 39% favored the increased use of naltrexone. The approach in the present study is to examine organizational rather than individual adoption of naltrexone. Further, beyond the question of use or nonuse, we employ a measure of the implementation following adoption of naltrexone. Within substance abuse treatment organizations, three broad sets of characteristics are expected to be predictors of naltrexone adoption and implementation: center structure, center leadership, and caseload characteristics. Because of the paucity of literature on the subject, in the following paragraphs we outline a number of logical hypotheses. First are features of organizational structure. Considering profit status, we expected for-profit organizations to be more likely to adopt and implement naltrexone treatment because of the apparent lower amount of labor intensity associated with its use, and the greater pressure profit oriented organizations would be under to adopt new ideas in contrast to not-for-profit settings. Considering whether the treatment organization was based in a hospital or freestanding, we expected hospitalbased programs to be more likely to adopt and implement naltrexone. The reasoning here is the hospital setting may offer the medical expertise and an overall culture supportive of the use of medications in substance abuse treatment that would be less likely to be found outside hospital settings. A large body of research has examined the relationship between age and adoption behavior by individuals, but the overall results are inconclusive (Rogers, 1995, p. 269). Because organizational age is likely associated with a more distinctive organizational culture and clearly defined image in the community, we expected older centers would be less likely to adopt a new idea such as the use of naltrexone in substance abuse treatment. The size of a treatment organization was expected to positively relate to adoption and implementation of naltrexone treatment (Moch & Morse, 1977; Kimberly & Evanisko, 1981; Rogers, 1995). Size may be seen as a proxy indicator of organizations access to resources from the external environment. Further, such resource access may produce the organizational slack necessary to engage in experimental behaviors such as the adoption of innovations. The final feature of organizational structure considered the extent of physician time available to the treatment program. Since naltrexone is a medication, it was expected this variable would be positively related to the organization s adoption and implementation of naltrexone treatment. Looking at a dimension of organizational leadership, it was expected the length of time a center s chief administrator had been involved in substance abuse treatment would be negatively related to the organization s adoption and implementation of naltrexone treatment. Just as with

3 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) center age, an administrator s commitment to institutionalized ideas and practices likely increases with time. The administrator s degree was also expected to relate to the organization s adoption and implementation of naltrexone. Organizations employing administrators with clinical or medical degrees were expected to be more likely to report the adoption and implementation of naltrexone treatment. Similarly, the percentage of counselors with a master s degree or higher was expected to be positively related to the organization s adoption and implementation of naltrexone treatment. These expectations are linked with a longterm finding of higher levels of education and sophistication among the early adopters of innovations (Rogers, 1995). This analysis also considered characteristics of the center s caseload relative to the organization s adoption and implementation of naltrexone treatment. The extent to which the organization was filling its treatment capacity was expected to relate negatively to the organization s adoption and implementation of naltrexone treatment. The assumption here is the extent to which a center is operating at capacity is an indicator of the effectiveness of its treatment methods, and those centers attracting more patients are less likely to change their methods. Privately funded treatment organizations must attract patients in order to survive. The extent to which patients are drawn from a variety of referral sources varies across centers. We expected the greater the diversity of sources of referral, the less likely the organization s adoption and implementation of naltrexone treatment. The reasoning here is referral sources must be convinced of the legitimacy of the treatment organization s technology. The greater the number of such sources, the more difficult the process of educating the sources about the legitimacy of innovative treatment practices. We expected a positive association between the percentage of medical-based patient referrals and the organization s adoption and implementation of naltrexone treatment. If the treatment organization has a relatively strong relationship with physicians in terms of referrals, then there may be a better fit with the utilization of medications in substance abuse treatment that are isomorphic with many other forms of medical practice. Managed care organizations are charged with reducing costs associated with health care. The demands placed on private substance abuse treatment by managed care organizations have been particularly harsh (Roman, Johnson, & Blum, 2000). Given that medications are supposed to increase treatment efficiency and possibly reduce treatment costs, we expected the percentage of patients covered by managed-care style arrangements would be positively related to the organization s adoption and implementation of naltrexone treatment. We also expected the proportion of patients who had relapsed after having previously received formal treatment would be positively related to the organization s adoption and implementation of naltrexone treatment. The reasoning here is relapsers may be seen as representing the failure of traditional treatment methods, and when more of the caseload is made up of such patients, the organization may be more likely to try new treatment techniques. 2. Materials and methods The National Treatment Center Study (NTCS) is a nationally representative longitudinal study of private-sector alcohol and drug abuse treatment centers. From 1994 through 1999, the NTCS was designed to explore whether and how private sector centers diversified their services in the managed care era, and to determine which organizational features and management practices provided a competitive advantage. During this 5-year period, the administrators and clinical directors at participating centers completed two intensive on-site interviews in and 1998, and five short telephone interviews at 6-month intervals following both waves of on-site visits. In 2000, a third wave of on-site interviews began with a slightly different focus. Though we continue to follow the structure, organization, and management practices of these centers, a much greater emphasis is being placed on the adoption and implementation of innovative treatment strategies including both pharmaceutical and therapy-based treatments. The analyses included here are based on this third wave of data Sample The NTCS is based on a two-stage stratified random sample of 450 treatment centers representative of the private treatment sector in the US. The sample includes both forprofit and not for-profit centers. Because a focus of the study was on organizational competition for patients, nonprofit centers were eligible for inclusion only if they received less than half their annual funding from local, state or federal grant/contract sources. To be eligible for inclusion in our study s initial sampling frame, centers were required to offer treatment for alcoholism; in practice, nearly all of the centers in the sample treat both alcohol and drug problems. Centers were also required to offer a level of care at least equivalent to structured outpatient services as defined by American Society of Addiction Medicine standards. In effect, these requirements excluded counselors in private practice, halfway houses and transitional living facilities, and programs that offered exclusively methadone maintenance, court-ordered driver education (DUI) classes, or detoxification services. The 450 baseline interviews completed in represent 89% of all centers invited to participate in the study. These centers are located in 35 states, including urban, suburban, and rural areas. Both telephone followups and subsequent on-site interviews have maintained

4 214 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) response rates of 95% or higher among centers that remain open and eligible for the study. Since our baseline data collection, 97 of the original 450 participants have either closed completely, or no longer treat substance abuse disorders. A split panel design method was used in 1998 and again in 2000 to add random samples of 26 and 37 centers, respectively, to the existing respondents in order to maintain a sample size of Data collection Though on-site interviews did involve face-to-face interaction with human subjects, the information collected from these respondents was organization-level data not personal data. Thus, human subjects approval was granted from the University of Georgia Institutional Review Board in an expedited review with a waiver of consent. In lieu of a consent form, respondents received a statement of confidentiality describing the steps taken to protect the identity of their organization and assuring them the information they provide us about their organization will be kept strictly confidential. In order to insure the accuracy of the data, during on-site interviews, respondents were requested, but not required, to consult available records. In most cases, respondents consulted financial and census records in order to provide interviewers with accurate and up-to-date information. In some cases, centers did not keep records of the information we were requesting (e.g., the percent of relapsers treated at the center). In these instances, respondents were asked to give their best estimate based on their knowledge of the center s patient base. In contrast to publicly funded centers, most of the private centers are very small (median FTEs = 20). As such, the administrators and clinical directors in these centers tend to work very closely with clinicians. For this reason, we are confident that, in those cases in which estimates are necessary, the estimates are accurate Measures Three different dependent variables are used in these analyses. The first is a dichotomous variable based on the answer given when administrators were asked, Does this center currently use naltrexone. The remaining two are continuous variables measuring naltrexone utilization for two types of patients: those with primary diagnoses of alcohol dependence/abuse and those with primary diagnoses of opiate dependence/abuse. Three groups of independent variables are included in these models. (1) Center structure is comprised of five variables including: profit status (1 = for profit), hospital/ freestanding status (1 = hospital-based), center age (in years), the number of full-time equivalents (FTEs), and the average number of physician hours per patient, which is calculated by dividing the number of hours per week a physician is employed at the center by the 12-month average daily Table 1 Descriptive statistics for variables included in models (SD in parentheses) Mean Range Dependent variables Currently use naltrexone (1 = yes) 44.1% Yes % Primary alcoholism patients 13.2 (18.2) 0 to 90 Receiving naltrexone % Primary opiate-addicted patients 11.3 (19.4) 0 to 100 Receiving naltrexone Independent variables Profit status (1 = for profit) 31% For profit Hospital freestanding (1 = hospital) 58% Hospital-based Age (in years) 18.3 (11.4) 2 to 116 Size (# of FTEs) 40.2 (51.6) 1 to 337 Physician hours per patient 0.57 (0.79) 0 to 3 Administrator-years in field 17.8 (7.6) 1 to 41 Administrator-highest degree field Business 17.9% Clinical (Psych., MSW) 47.8% Medical 15.7% Other 18.6% % MS/MA counselors 53.8 (33.2) 0 to 100 % Capacity 66.0 (24.8) 0 to 100 Diversity of referrals 4.5 (2.0) 0 to 8 % Medical referrals 34.6 (27.8) 0 to 100 % HMO/ PPO patients 31.1 (25.0) 0 to 100 % Relapsers 49.2 (25.4) 0 to 100 census. (2) Center leadership includes the number of years the administrator has worked in the substance abuse treatment field, the administrator s degree field, and the percentage of counselors with a master s degree or higher. (3) Caseload characteristics includes five variables. Single-item measures include the percentage of patients covered by HMOs, PPOs, and other managed-care style arrangements, and the percentage of patients that have relapsed after having previously received some form of formal treatment. 1 The remaining three are constructed variables. The center s average percent capacity is measured by dividing the 12-month average daily census by the center s total capacity (beds + outpatient slots). The percentage of medical referrals is the sum of referrals from physicians in private practice, other health care providers such as mental health professionals or other treatment facilities, and, for those centers based within a hospital or treatment system, referrals from within the system. Finally, we consider the diversity of patient referrals. Referral diversity ranges from 0 8 and is measured by summing the number of legitimate referral sources from the following list: employee assistance programs, direct referrals from non-eap workplaces, legal system, social service agencies, physicians in private practice, other health care providers, clergy, and schools. A legitimate referral source is defined as one that provides 5% or more of a center s patients. Table 1 shows the descriptive characteristics (mean, range, and SD) for all variables. 1 Details of these patients including the location and content of prior treatment are unknown.

5 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) Statistical analyses The analysis tests the same model on three different dependent variables. In the case of adoption, the dichotomous dependent variable is coded 0 if the center does not use naltrexone and 1 if it does use naltrexone. Because of the dichotomous nature of this variable, logistic regression is the appropriate analytical method. The results of this model are summarized in Table 2. The remaining dependent variables, which measure implementation, are continuous variables and will therefore be tested using ordinary least squares regression (OLS). The purpose of these two models is to predict, among those centers using naltrexone, the average percentage of patients to whom it will be administered. The dependent variable in the first of these models looks at the distribution of naltrexone to patients with a primary diagnosis of alcoholism while the second model considers patients with a primary diagnosis of opiate dependence. The results of both these models are summarized in Table 3. Each model examines three sets of independent variables representing the center s structural, leadership, and caseload characteristics. With both the logistic regression and OLS models, one-tailed t-tests determine the significance of the variables in the model. Results of the logistic regression Table 2 Logistic regression of use of naltrexone on center structure, leadership and caseload characteristics Odds 90% C.I. for odds ratio Variable Coeff. t Ratio (Lower) (Upper) Center structure Profit status (1 = for profit) Hospital freestanding (1 = hospital) Age (in years).038* Size (# of FTEs) Physician hours per patient Leadership Admin.-Years.032* in Tx Field Admin.-degree field a Business Clinical Medical % MS/MA counselors.008* Caseload characteristics % Capacity Diversity of referrals %Medical referrals %HMO/PPO patients.016** % Relapsers.012* a Excluded category in administrator degree field is other degree. * p <.05. ** p <.01. Table 3 OLS regression of the percentage of primary alcoholism patients and the percentage of primary opiate-addicted patients receiving naltrexone on center structure, leadership and caseload characteristics (standardized coefficients, standard errors in parentheses) Beta (S.E.) Variable Primary alcoholism Primary opiate Center structure Profit status (3.964) (4.779) (1 = for profit) Hospital/freestanding (4.050) (4.922) (1 = hospital) Age (in years) (0.123) (0.149)** Size (# of FTEs) (0.032) (0.039) Physician hours per patient (2.037) (2.457) Leadership Admin.- # Years in Tx field.108 (.210).030 (.256) Admin.- Degree field a - Business.246 (6.061)*.278* (7.323) - Clinical.196 (4.358).120 (5.279) - Medical.344**(5.350).197 (6.451) % MS/MA counselors.038 (.050).008 (.061) Caseload characteristics % Capacity.161 (6.344).085 (7.684) Diversity of referrals.215* (0.791).092 (0.957) % Medical referrals.140 (0.060).069 (0.072) % HMO/PPO patients (0.059).068 (0.073) % relapsers.026 (0.066).091 (0.081) Intercept R a Excluded category in administrator degree field is other degree. * p <.05. ** p <.01. report the coefficient, t-value, and odds ratios. OLS regression results report the standardized coefficient, SE, and the coefficient of determination (r-square). 3. Results A bi-variate examination of the structural differences in centers adopting naltrexone (results not shown) indicated hospital-based centers and larger centers were more likely to use naltrexone. As the results of the logistic regression model in Table 2 show, however, these structural differences did not hold when controlling for other variables. In fact, none of our hypotheses regarding center structure were supported. Only one of the five structural variables had a statistically significant effect on naltrexone and it was opposite the hypothesized direction. Controlling for all other variables, older centers are significantly more likely to use naltrexone than younger centers. A 1-year increase in center age increases the likelihood of using naltrexone by 3.9%. Two of the three center leadership variables are significant predictors of naltrexone use. The number of years the administrator has worked in the treatment field is statistically significant, but the effect is opposite the hypothesized

6 216 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) direction. Instead, the administrator s tenure in the treatment field, increases the likelihood of naltrexone use. As predicted, the percentage of counselors with a master s degree or higher significantly increases the likelihood of using naltrexone. The administrator s educational background did not show a statistically significant relationship with adoption. Two of the five hypotheses regarding centers caseload characteristics received support. Results of the logistic regression indicate higher percentages of patients covered by HMOs, PPOs, or other managed care arrangements increase the likelihood of using naltrexone as does the percentage of relapsers treated by the center. Neither percentage operating capacity, referral source diversity, nor percentage of medical-based referrals affects centers likelihood of using naltrexone. The structure/leadership/caseload model demonstrates limited success at predicting adoption of naltrexone. An identical model was used to predict the extent of utilization or implementation among alcoholism or opiate-addicted patients. Results of the model predicting the percentage of primary alcoholism patients receiving naltrexone, is shown in the first column of Table 3. As predicted, centers employing administrators with medical degrees treated a higher percentage of patients with naltrexone than centers employing administrators with other degrees. The significant positive affect of administrators with business degrees as compared to other degrees was unexpected. The model also indicates referral source diversity is a significant predictor of naltrexone implementation, suggesting centers receiving referrals from a broader range of sources administer naltrexone to a larger percentage of alcohol dependent patients. This too was opposite the hypothesized direction. The model explains about 17% of the variance in the percentage of primary alcoholism patients receiving naltrexone. In the model predicting the percentage of primary opiateaddicted patients receiving naltrexone, shown in the second column of Table 3, both center age and having a business degree have significant effects. An increase in the age of the center will result in an increase in the percentage of opiateaddicted patients receiving naltrexone. Likewise, centers employing administrators with business degrees as compared to other degrees administer naltrexone to significantly larger percentages of opiate-addicted patients. Neither of these findings was predicted. This model explains about 21.5% of the variance in the dependent variable. 4. Discussion Within this sample of privately funded substance abuse treatment centers, 44.1% report current use, or adoption of naltrexone. This degree of adoption is relatively consistent with the limited research examining individual level attitudes toward naltrexone (Rychtarik, Connors, Dermen, & Stasiewicz, 2000; Thomas et al., 2001). Rychtarik et al. (2000) found among AA members 53% reported use of medication was either a good idea or might be a good idea. Likewise, Thomas et al. (2001) found about 60% of physicians working in the addiction treatment field had prescribed naltrexone. Compared to the degree of adoption, the degree of implementation within these centers is substantially lower, with a mean of 13.2% of primary alcohol dependent patients and 11.3% of primary opiate dependent patients receiving naltrexone treatment. Despite this low rate of implementation, the similarity in proportions of use across these two patient groups is notable. As adoption and implementation are underway within a population of organizations, it would be expected that implementation would lag well behind adoption. Three subsets of variables were examined to determine their relationship to the adoption and implementation of naltrexone: center structure, leadership, and caseload characteristics. At the bi-variate level, results suggest a center based in a hospital is more likely to adopt naltrexone. This relationship does not hold, however, when placed in a multivariate logistic regression model. From the standpoint of center structure, what does emerge in the adoption model is the impact of organizational age. Though in the opposite direction of what was predicted, older centers show a greater likelihood of adoption. This relationship was also present in the implementation model for primary opiate-addicted patients. Again, older centers demonstrated a greater rate of implementation. None of the other structural characteristics were significant predictors of either adoption or implementation of naltrexone. These three structural variables: (placement of the center within a hospital, organizational size, and the availability of physician time) were strongly expected to influence adoption and implementation through providing critical facilitating mechanisms. This appears to indicate adoption of naltrexone does not necessarily demand an expanded resource base, and such adoption can be integrated into routine treatment activities. In terms of center leadership, both the administrator s tenure in the substance abuse treatment field, and the proportion of degreed counselors had a significant positive effect on adoption. As with center age, the positive effect of administrator tenure on adoption was opposite the predicted direction. If, however, it can be assumed centers that are older and more firmly established and administrators with long-term service are those looked to as opinion leaders, this finding may bode very well for the continuing adoption of naltrexone. Though the administrator s degree background did not affect adoption, it did have a significant effect on the degree of implementation. Specifically, having an administrator with a business degree had a significant positive effect on implementation of naltrexone for both primary alcoholism patients and primary opiate-addicted patients. Having an administrator with a medical degree also had a significant positive effect on implementation of naltrexone for primary alcoholism patients. This may be interpreted as indicating that when naltrexone s use is perceived as both practical (from a management perspec-

7 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) tive) and medically efficacious, it is more likely to be used in treating more patients. Two features of caseload characteristics were significant predictors of adoption: the percentage of managed care patients and the percentage of relapsers within the patient population. These findings point toward two different dimensions of the use of naltrexone, namely its apparent link with the limitations on treatment resulting from managed care, and its perceived utility in treating patients who have not responded to other regimens. It should be noted the level of the organization s performance, as reflected in its capacity utilization, was not significantly linked, nor was the proportion of referrals from medical sources. The diversity of referral sources was positively related to naltrexone implementation with primary alcoholism patients but not with primary opiate-addicted patients. Because diversity of referral sources is an indicator of the scope of the treatment center s effort to attract patients, this finding may be interpreted as being consistent with the positive association between implementation and the administrator s business background. Though in the opposite direction of what was hypothesized, it seems reasonable a center relying on a wide diversity of sources for its patients would need to provide a wide diversity of treatment options. Though the model tested here was more successful in predicting adoption of naltrexone than the implementation of naltrexone for either primary alcoholism or primary opiate addiction, the variables that emerged as statistically significant were logical, and were consistent around the idea that a greater use of naltrexone is a sound business practice. Because less than half of the centers offer naltrexone, statistical power becomes an issue in the OLS models of implementation. This study, however, will continue to track the patterns of adoption and implementation of innovations within this sample of centers. Should adoption and implementation of naltrexone increase over time, future analyses will be able to discern a clearer picture of the predictors of implementation. This analysis was designed to assess the predictive utility of structural, leadership, and caseload features in understanding the adoption and implementation of innovations in substance abuse treatment. Using naltrexone as an example innovation, these models found adoption is related to both the age of the treatment center and the length of time its administrator has served in that capacity. Further, adoption is linked with higher levels of education among the counseling staff of the treatment program. In terms of caseload characteristics, adoption is significantly associated with the extent to which the center s caseload is covered by managed care programs and by the percentage of patients in the center s caseload that have previously received some form of formal treatment and have relapsed. These findings are encouraging about the possible opinion leadership exerted by more experienced centers and administrators, and by the apparent multiple utilities perceived to be associated with naltrexone use reflected in the caseload findings. It was, however, somewhat surprising that we did not find support for several hypotheses that are based on substantial theoretical as well as common sense. Many earlier adoption studies have found size (often as an indicator of resource munificence) to predict adoption, which was not the case here. Likewise, two indicators of support for medication use, hospital base and physician time available, did not emerge as significant predictors. Finally, since it would appear that medication treatment may reduce labor costs in terms of counselor time and thus be attractive to for-profit centers, this variable also was not shown to be a significant predictor. This pattern of findings might suggest that unlike many organizations that are constantly changing their technology and adopting new ideas, such experience has not been the norm in treatment programs. The association of seniority status among center administrators and centers themselves suggests opinion leadership is important in this particular adoption process, with centers looking toward the voice of experience for leadership. The hypotheses associated with center caseload that did not receive support are also consistent with this pattern of findings wherein adoption does not appear to be economically driven. Our notion that level of occupancy and the diversity of referrals were indicators of the center s need for cost-cutting and improved efficiency, and the linkage of such efficiency with naltrexone adoption, was not supported. Likewise, consistent with the nonsupport for the hypotheses linking adoption to a hospital base and to amount of available physician time, the hypothesis that the level of medically based referrals would predict adoption was not supported. Though naltrexone shows a pattern of fairly widespread adoption within private substance abuse treatment centers, the degree of implementation remains quite low. OLS regression models show implementation of naltrexone treatment for primary alcoholism patients is associated with the administrator s degree field and the diversity of the center s referral sources, while implementation for primary opiateaddicted patients is associated with the age of the center and the administrator s degree field. As mentioned earlier, this model represents those factors easily attainable and largely unambiguous. Future research might focus on applying this structure/leadership/caseload model to other treatment interventions such as motivational enhancement therapy. Though it would require individuallevel data collection from clinical staff, Simpson s (2002) more comprehensive program change model should also be empirically tested in a large sample of substance abuse treatment centers. Acknowledgments The authors gratefully acknowledge the grant support of the National Institute on Drug Abuse (Grant No.

8 218 P.M. Roman, J.A. Johnson / Journal of Substance Abuse Treatment 22 (2002) R01DA13110) and the National Institute on Alcohol Abuse and Alcoholism (Grant No. R01AA10130). References Forman, R. F., Bovasso, G., & Woody, G. (2001). Staff beliefs about addiction treatment. Journal of Substance Abuse Treatment, 21, 1 9. Jaffe, A. J., Rounsaville, B., Chang, G., Schottenfeld, R. S., Meyer, R. E., & O Malley, S. (1996). Naltrexone, relapse prevention, and supportive therapy with alcoholics: an analysis of patient treatment matching. Journal of Consulting and Clinical Psychology, 64, Kimberly, J. R., & Evanisko, M. A. (1981). Organizational innovation: the influence of individual, organizational and contextual factors on hospital adoption of technological and administrative innovations. Academy of Management Journal, 24, Knox, P. C., & Donovan, D. M. (1999). Using naltrexone in inpatient alcoholism treatment. Journal of Psychoactive Drugs, 31, Krystal, J. H., Cramer, J. A., Krol, W. F., & Rosenheck, R. A. (2001). Naltrexone in the treatment of alcohol dependence. New England Journal of Medicine, 345, Lamb, S., Greenlick M. R., McCarty D. (Eds.) (1998). Bridging the gap between practice and research: forging partnerships with communitybased drug and alcohol treatment. Washington, DC: National Academy Press. Moch, M. K., & Morse, E. V. (1977). Size, centralization and organizational adoption of innovations. American Sociological Review, 42, National Institute on Drug Abuse. (1999). Principles of drug addiction treatment: a research based guide. (NIH Publication No ). Washington, DC: Government Printing Office. O Malley, S. S., Jaffe, A., Chang, G., Witte, G., Schottenfeld, R. S., & Rounsaville, B. J. (1992). Naltrexone in the treatment of alcohol dependence: preliminary findings. In C. A. Naranjo & E. M. Sellers (Eds.), Novel pharmacological interventions for alcoholism ( pp ). New York: Springer- Verlag. Rogers, E. M. (1995). Diffusion of innovations (4th ed.). New York: The Free Press. Roman, P. M., Johnson, J. A., & Blum, T. C. (2000). The transformation of private substance abuse treatment: results of a national study. In Judith Levy (Ed.), Advances in medical sociology, (Vol 7, pp ). Greenwich, CT: JAI Press. Rychtarik, R. G., Connors, G. J., Dermen, K. H., & Stasiewicz, P. R. (2000). Alcoholics Anonymous and the use of medications to prevent relapse: an anonymous survey of member attitudes. Journal of Studies on Alcohol, 61, Simpson, D. D. (2002). A conceptual framework for transferring research to practice. Journal of Substance Abuse Treatment, 22 (4), Terenius, L. (1998). Rational treatment of addiction. Current Opinion in Chemical Biology, 2, Thomas, C., Wallack, S., Swift, R., Bishop, C., McCarty, D., & Simoni- Wastila, L. (2001). Adoption of naltrexone in alcoholism treatment. Journal of Addictive Diseases, 20, 180. Volpicelli, J. R., Alterman, A. I., Hayashida, M., & O Brien, C. P. (1992). Naltrexone in the treatment of alcohol dependence. Archives of General Psychiatry, 49,