The Effectiveness of Telephone-Based Continuing Care for Alcohol and Cocaine Dependence: 24-Month Outcomes

Context  Telephone-based disease management protocols have shown promise in improving outcomes in a number of medical and psychiatric disorders, but this approach to continuing care has received little study in alcohol- and drug-dependent individuals.

Objective  To compare telephone-based continuing care with 2 more intensive face-to-face continuing care interventions.

Design  A randomized 3-group clinical trial with a 2-year follow-up.

Setting  Two outpatient substance abuse treatment programs, one community-based and the other at a Veterans Affairs medical center facility.

Patients  Alcohol- and/or cocaine-dependent patients (N = 359) who had completed 4-week intensive outpatient programs.

Interventions  Three 12-week continuing care treatments: weekly telephone-based monitoring and brief counseling contacts combined with weekly supportive group sessions in the first 4 weeks (TEL), twice-weekly cognitive-behavioral relapse prevention (RP), and twice-weekly standard group counseling (STND).

Main Outcome Measures  Percentage of days abstinent from alcohol and cocaine, total abstinence from alcohol and cocaine, negative consequences of substance use, cocaine urine toxicological results, and γ-glutamyltransferase.

Results  Participants in TEL had higher rates of total abstinence over the follow-up than those in STND (P<.05). In alcohol-dependent participants, 24-month γ-glutamyltransferase levels were lower in TEL than in RP (P = .005). In cocaine-dependent participants, there was a significant group × time interaction (P = .03) in which the rate of cocaine-positive urine samples increased more rapidly in RP as compared with TEL. On percentage of days abstinent or negative consequences of substance use, TEL did not differ from RP or STND. Participants with high scores on a composite risk indicator, based on co-occurring alcohol and cocaine dependence and poor progress toward achieving intensive outpatient program goals, had better total abstinence outcomes up to 21 months if they received STND rather than TEL, whereas those with lower scores had higher abstinence rates in TEL than in STND (P = .04).

Conclusions  Telephone-based continuing care appears to be an effective form of step-down treatment for most patients with alcohol and cocaine dependence who complete an initial stabilization treatment, compared with more intensive face-to-face interventions. However, high-risk patients may have better outcomes if they first receive group counseling continuing care after completing intensive outpatient programs.

Alcohol dependence and drug dependence are now thought to be chronic disorders for many patients who receive formal treatment.^1,2 Because of the relapsing nature of these disorders, some form of continuing care—also referred to as aftercare—is usually recommended following completion of an initial phase of treatment.³ Despite the potential benefits of continuing care, many substance-dependent individuals either do not attend any continuing care or stop attending after a relatively small number of sessions.^4-6

Continuing care protocols that are more flexible and less burdensome than traditional face-to-face counseling sessions may be more attractive to patients and might therefore promote better adherence and improved disease management over time. One such intervention is telephone-based continuing care. The telephone is considered to have a viable therapeutic role in the monitoring and treatment of a number of disorders, including depression, obsessive-compulsive disorder, congestive heart failure, and chronic pulmonary disease.^7-14 With addictions, telephone counseling has been effective in smoking-cessation programs.^15,16 However, few studies have been done with patients with alcohol or drug use disorders, and results have been mixed.^17-19

This study examined the relative effectiveness of telephone-based continuing care for patients with alcohol and/or cocaine dependence who had completed a 4-week intensive outpatient program (IOP). Participants were randomized to a condition that consisted primarily of brief telephone counseling contacts combined with several group support sessions early in the protocol (TEL) or to 1 of 2 comparison conditions: individualized relapse prevention (RP) or standard 12-step–oriented group counseling (STND). Relapse prevention was selected as a comparison condition because it has shown efficacy in numerous research studies.^20-24 Group counseling was selected as a second comparison condition because it is by far the most common approach to continuing care in addiction treatment programs. Therefore, the comparison conditions represented the state of the art in the clinical research field and treatment as usual, respectively. Each continuing care condition was 12 weeks in length, and participants were observed for a total of 24 months.

The primary research question was whether either RP or STND would produce better substance use outcomes than TEL. The 2 face-to-face conditions were not directly compared, as results from several recent studies^24-28 suggested that they would yield similar substance use outcomes. In analyses of data from the first year of the follow-up,²⁹ the TEL condition did not differ from either RP or STND on percentage of days abstinent, total abstinence, frequency of cocaine use, or cocaine urine toxicological results. However, moderator analyses indicated that patients with high scores on a composite risk indicator measure—which included current dependence on both alcohol and cocaine and failure to achieve therapeutic goals while in the IOP program—had better total abstinence outcomes in STND than in TEL. Conversely, patients with lower scores on this measure had better total abstinence outcomes in TEL than in STND.³⁰

Although these results from the first year of the follow-up provided initial evidence of the equivalence of the telephone-based continuing care intervention with the more intensive interventions, the possibility remained that longer-term outcomes would not be as positive. In this article, data from the entire 2-year follow-up are presented to evaluate longer-term outcomes in the TEL condition, relative to RP and STND. Moderator analyses with the composite risk indicator were also conducted to determine whether the interaction effects in the first year would persist into the second year of the follow-up. The outcomes that were examined were percentage of days of abstinence, total abstinence, negative consequences of substance use, and cocaine urine toxicological results. A biological measure of heavy alcohol use obtained at the end of the follow-up—γ-glutamyltransferase (GGT)—was also examined. Finally, the treatment conditions were compared on self-help participation and additional episodes of inpatient and outpatient substance abuse treatment in the second year of the follow-up.

The participants were 359 adults with DSM-IV³¹ cocaine or alcohol dependence at the time of entrance to treatment who had completed an IOP, were willing to participate in research, and did not have a psychiatric or medical condition that precluded outpatient treatment. Other inclusion and exclusion criteria are described elsewhere.²⁹ The participants averaged 41.9 years of age, 12.4 years of education, 8.5 years of regular cocaine use, 18.6 years of regular alcohol use, 2.8 prior drug abuse treatments, and 3.5 prior alcohol treatments. The majority of participants were male (82.6%), African American (77.1%), and not currently married (82.6%). Participants were using alcohol or cocaine on 2 of every 3 days during the 4-month period that preceded the IOP. Data on participants in each treatment condition are presented in Table 1.

Patients participated in an IOP at 1 of 2 sites: a clinical research unit located at a community-based outpatient program or the Philadelphia Veterans Affairs Medical Center. Both programs offered 9 to 10 hours of group therapy per week and were 4 weeks in length. To graduate from the IOP, patients had to attend regularly and achieve a week of urine toxicological examination–confirmed abstinence by the end of the program. Following graduation from the IOP, patients were randomly assigned to 1 of the following 3 treatment conditions, each of which is described in more detail elsewhere.^29,32

Patients received 2 group therapy sessions per week, which were a combination of addiction counseling and 12-step recovery practices. This condition is typical of aftercare as it is delivered in many outpatient clinics.

Patients received 1 individual RP session, from a protocol developed by Annis and Davis,³³ and 1 group session per week. A treatment manual and a series of structured modules were used to guide within-session activities as well as between-session homework assignments.³⁴

Patients had 1 initial individual face-to-face session to orient them to the protocol and thereafter completed one 15-minute telephone call per week. A workbook was used to focus and structure these contacts (R. Morrison, MA, RN, J. R. McKay, PhD, unpublished data, 1997). The telephone sessions began with a brief review of progress toward 1 to 2 goals identified in the prior session. Plans for achieving goals over the next week were then discussed, along with any other pressing issues. During the first 4 weeks of treatment, patients were also offered a weekly support group. The main purpose of this group was to ease the transition from clinic-based to telephone-based treatment. The therapists had the option to retain a patient in the group beyond the 4-week period if the patient was at high risk for relapse. Therefore, although the majority of contacts in this condition were made via the telephone, it was a multimodal intervention that also included some face-to-face therapeutic contacts.

The 10 therapists who provided the continuing care treatments were all experienced substance abuse counselors. Each condition was delivered by a minimum of 3 therapists at each site. The individual sessions of RP and TEL therapists were audiotaped to monitor adherence to the protocol as described in the manuals. Individual supervision was provided weekly by the study clinical coordinator, who listened to a majority of the audiotapes, and 1 group supervision session was also held each week. Further information on the therapists and adherence is provided elsewhere.²⁹

Participants at the community site (n = 188) were recruited through local treatment programs and social service agencies. These individuals became eligible for the study if they completed the IOP program, met inclusion criteria, and were able to provide cocaine-free urine tests at the end of the IOP. At the Veterans Affairs site, patients in the final week of the 4-week IOP were screened for study eligibility. Patients who met inclusion criteria (n = 175) signed informed consent forms and were randomized at the end of the week. The study was approved by University of Pennsylvania and Philadelphia Veterans Affairs Medical Center institutional review boards.

There were approximately 1500 telephone inquiries at the community site from prospective participants. Individuals who appeared to meet eligibility criteria were invited for in-person screening, and of these, approximately 25% kept their appointments. The vast majority of individuals who attended these sessions were deemed eligible to participate and entered the IOP. Analyses presented elsewhere²⁹ indicated that IOP graduates were more likely to be living in halfway houses and less likely to have used cocaine within 2 days of intake than those who did not complete IOP.

At the Veterans Affairs site, 417 patients with alcohol or cocaine use disorders were screened, and of these, 175 (42.0%) were eligible to participate and were enrolled in the study. The most common reasons for failure to enroll in the study were declined participation (96 of 417; 23.0%), psychiatrically unstable (19; 4.6%), outside of the age range (18; 4.1%), referred to a geriatric group (17; 4.1%), intravenous heroin use in the prior year (17; 4.1%), and failure to come to the baseline assessment (14; 3.4%).

Urn randomization procedures were used at each site.³⁵ The balancing factors are described elsewhere.²⁹ Because several of the balancing factors had highly skewed distributions, the randomization did not produce equal sample sizes across the 3 conditions (STND, 122; RP, 135; and TEL, 102).

Baseline assessments were started in the last week of IOP, prior to randomization. All patients who completed the first portion of the baseline assessment (ie, prerandomization) were considered to be in the study. The follow-up assessments were conducted at 3, 6, 9, 12, 18, and 24 months postbaseline. The follow-up rates for self-report data were 90% or greater in the first year, 89% at 18 months, and 86% at 24 months. The treatment conditions did not differ on follow-up rates for self-report or urine toxicological data. Assessments were conducted by highly trained and experienced research staff.²⁹

The Structured Clinical Interview for the DSM-IV was used to assess DSM-IV Axis I disorders.³⁶

The Addiction Severity Index³⁷ was used to gather information on medical, employment, drug use, alcohol use, legal, family/social, and psychiatric problem severity. The Addiction Severity Index has demonstrated good internal consistency, test-retest, and interrater reliabilities in different groups of substance abusers.^38,39

Prior work by our group and others indicated that patients who were dependent on both alcohol and cocaine and those who made little progress toward achieving the primary treatment goals of the IOP were likely to have particularly poor substance use outcomes³⁰ and might therefore be inappropriate for telephone-based continuing care. To test this hypothesis, we developed a composite risk indicator measure that was composed of the following items: (1) current dependence at entrance to the IOP on both alcohol and cocaine vs dependence on only 1 substance, (2) any cocaine use during the IOP, (3) any alcohol use during the IOP, (4) attendance at fewer than 12 self-help meetings during the IOP,⁴⁰ (5) score on the social support measure at the end of the IOP that was below the median in the study sample,⁴¹ (6) abstinence goal at the end of the IOP that was less stringent than total abstinence,⁴² and (7) self-efficacy score at the end of the IOP that was less than 80%.⁴³ The items were all derived from assessment instruments with good reliability and predictive validity. Each item was coded as either 1 (indicating higher risk) or 0. A total score was obtained by summing the scores on all 7 items (mean ± SD, 2.50 ± 1.36).³⁰ Psychometric analyses indicated the 7 items were adequately described by a single factor model. The procedures used to develop this measure, including rationales for the cut points that were used with the continuous measures, are described in detail elsewhere.³⁰

Time-line follow-back⁴⁴ techniques were used to gather self-reports of alcohol and cocaine use during the 6 months preceding continuing care and the 24-month follow-up period. In studies with alcoholics and drug addicts, there has been good agreement between time-line follow-back data and collateral and biological data.^45-47 The primary outcome measures derived from the time-line follow-back data were percentage of days of abstinence from both alcohol and cocaine and a dichotomous variable representing complete abstinence from both alcohol and cocaine. Both measures were computed for each 3-month segment of the 24-month follow-up.

The Inventory of Drug Use Consequences⁴⁸ was used to gather data on negative consequences of alcohol or drug use. The total score was used in the analyses. This measure has good to excellent test-retest reliability.⁴⁸

Urine samples were obtained at each follow-up and tested for the cocaine metabolite benzoylecgonine using either the Enzyme Multiple Immunoassay Test system or fluorescence polarization immunoassay analysis (with quantitative output converted to a dichotomous variable).

Blood samples obtained at baseline and 24 months were assayed for GGT. Higher scores are typically associated with heavy alcohol consumption.⁴⁹ Because procedures for collecting blood samples were not implemented until part way through the study and because of occasional difficulties in getting blood samples processed, GGT values from both baseline and the 24-month follow-up were available for 148 of 272 participants with alcohol dependence.

Days of inpatient and outpatient substance abuse treatment during follow-up were obtained from the time-line follow-back data. An 8-item self-report questionnaire was used to assess patients’ participation in self-help groups. The measure had good internal consistency, and scores predict subsequent alcohol and cocaine use.^40,50

Mixed-effect regressions (SAS PROC MIXED; SAS Institute, Cary, NC) were used for the analyses with percentage of days abstinent and negative consequences of substance use. The time-line follow-back data were collapsed into a precontinuing care baseline period (6 months, including the IOP) and eight 3-month follow-up periods. Generalized estimating equations were used to analyze the dichotomous total abstinence measure and cocaine urine toxicological results.

Preliminary analyses with each outcome measure indicated that there were significant main effects for current substance dependence diagnosis and site in each analysis (ie, outcomes worse in patients with both alcohol and cocaine dependence and in those at the Veterans Affairs site), but other potential covariates were not significant in any of the analyses.²⁹ Furthermore, none of the interactions between these variables and treatment approached significance. Therefore, the final models contained main effects for site, baseline percentage of days abstinent (or negative consequences), treatment condition contrasts, the composite risk indicator (which included dual vs single dependence), and time, as well as interactions between the treatment condition contrasts, composite risk indicator, and time. In these models, the time variable tested for changes over the course of the follow-up (ie, from 3 to 24 months).

Potential differences between treatment conditions in treatment services and self-help participation data in year 2 were analyzed with analyses of variance; treatment condition differences in GGT scores at 24 months were examined with analyses of covariance, which controlled for baseline GGT levels. Because the primary research question was whether the TEL condition would be less effective than the 2 face-to-face continuing care conditions, the outcome analyses focused on 2 planned contrasts: STND vs TEL and RP vs TEL.^29,51

Participants in the 3 treatment conditions were compared on 21 demographic, diagnostic, treatment, and problem-severity variables assessed at baseline. Only 1 significant difference was observed; at the Veterans Affairs site, Addiction Severity Index legal composite scores were higher in TEL than in the other 2 conditions. There were no significant differences at the community site.²⁹

The mean ± SD number of continuing care sessions (telephone, group, and individual) received by participants was 14.24 ± 7.90 in STND, 14.36 ± 8.47 in RP, and 10.94 ± 5.92 in TEL. Patients in the TEL condition received an average of 6 telephone contacts, 4 group sessions, and 1 individual orientation session. Notably, 35% of those in TEL attended more than 4 group sessions.²⁹ On average, patients in TEL received approximately half as much therapeutic contact during continuing care as those in RP or STND (428 minutes vs 861 and 845 minutes, respectively).

Results from the repeated-measures analyses with the 4 outcome measures are summarized in Table 2 and presented in Figure 1, Figure 2, Figure 3, Figure 4, and Figure 5, which are based on available data (ie, not adjusted for missing data points). Higher scores on the composite risk indicator measure predicted worse outcomes on all 3 self-report measures but not on cocaine urine toxicological results. Rates total abstinence decreased significantly across the follow-ups; however, the other 3 measures did not change during the follow-up.

Data on percentage of days abstinent from alcohol and cocaine (untransformed) within each continuing care condition are presented in Figure 1. Neither the STND vs TEL nor the RP vs TEL contrast was significant. There were also no significant 2- or 3-way interactions between the treatment contrasts, time, and the composite risk indicator.

The percentages of participants in each treatment condition who reported total abstinence from alcohol and cocaine within each 3-month segment of the follow-up are presented in Figure 2. Abstinence rates were significantly lower in STND than in TEL (z = −1.98, P = .047). The STND vs TEL contrast × composite risk indicator interaction was also significant (z = 2.04, P = .04). Patients with risk indicator scores of 4 or higher had higher abstinence rates in STND than in TEL through month 21 of the follow-up, whereas those with scores of 3 or lower had higher abstinence rates in TEL than in STND. In months 21 to 24, however, abstinence rates were higher in TEL than in STND, regardless of risk indicator score. The interaction effect that was evident through month 21 is illustrated in Figure 3, which presents data from one representative 3-month segment of the follow-up (ie, months 13-15). In the comparison of RP and TEL, there was a significant interaction of the treatment contrast, composite risk indicator, and time (z = −2.10, P = .04). This effect was due to fluctuations over time in the direction of the interaction between the RP vs TEL contrast and the composite risk indicator, and it was therefore not clinically meaningful.

Negative consequence scores (untransformed) are presented in Figure 4. Neither the STND vs TEL nor the RP vs TEL contrast was significant. There were also no significant 2- or 3-way interactions between the treatment condition contrasts, time, and the composite risk indicator in these models.

These analyses were done with the 268 patients who had current cocaine dependence at entrance to the IOP. Percentages of cocaine-positive urine samples within each treatment condition at each assessment point are presented in Figure 5. Neither the STND vs TEL nor the RP vs TEL contrast reached significance. However, the rate of cocaine-positive urine samples increased more rapidly over time in RP compared with TEL (z = 2.20, P = .03), and a similar trend was obtained in STND compared with TEL (z = 1.93, P = .053). The 2-way interactions between these treatment contrasts and the composite risk index were not significant, nor were the 3-way interactions that included time.

Finally, GGT data obtained at the 24-month follow-up were examined in patients with alcohol dependence. The analyses indicated that GGT levels (log transformed) were significantly lower in TEL (n = 45; mean ± SE, 3.47 ± 0.10) than in RP (n = 48; mean ± SE, 3.86 ± 0.09) (F_1,90 = 8.45, P = .005). γ-Glutamyltransferase levels were also lower in TEL than in STND (n = 55; mean ± SE, 3.69 ± 0.09), but the difference only reached the level of a trend (F_1,97 = 2.76, P = .10).

Participants had a mean ± SD of 3.99 ± 12.51 days of inpatient substance abuse treatment during months 13 to 18 and 4.51 ± 15.92 days during months 19 to 24. The treatment conditions did not differ at either time point (13-18 months: F_2,318 = 0.42, P = .66; 19-24 months: F_2,304 = 2.05, P = .13). Participants had a mean ± SD of 5.67 ± 14.23 days of outpatient substance abuse treatment in months 13 to 18 and 5.83 ± 14.11 days in months 19 to 24. Once again, there were no significant differences between the conditions (13-18 months: F_2,318 = 2.07, P = .13; 19-24 months: F_2,304 = 2.82, P = .06). The conditions did not differ on self-help involvement at either 18 months (F_2,291 = 0.73, P = .49) or 24 months (F_2,294 = 2.10, P = .13).

This article presents 24-month outcomes from a randomized study that compared a telephone-based continuing care intervention with 2 more intensive face-to-face interventions in substance-dependent patients who had completed an initial phase of intensive outpatient treatment. Data from the entire follow-up were used to address 2 concerns that were raised in the analyses of the first-year outcomes.^29,30 The first was that outcomes in the telephone-based continuing care intervention might deteriorate faster in the second year, relative to outcomes in the 2 more intensive face-to-face conditions. This was clearly not the case. There were no significant treatment condition × time interactions on the 3 self-report outcome measures, and rates of cocaine-positive urine samples actually increased more rapidly during the follow-up in RP than in TEL. The TEL condition also had significantly lower GGT values at 24 months than RP. Finally, when the whole follow-up was considered, TEL produced higher rates of abstinence than STND.

The second concern was that despite the overall good performance of the telephone-based continuing care intervention, there might be some patients for whom this intervention was inadequate, at least as a first step-down treatment. Based on prior research, we hypothesized that patients who were dependent on both alcohol and cocaine and who failed to achieve IOP goals might respond better to more intensive face-to-face treatment than to a telephone-based protocol. These risk factors were operationalized using existing measures with good psychometric properties and combined to form a composite risk indicator. The moderating effect of the composite risk indicator in the comparison of TEL vs STND was significant in the analyses to predict total abstinence and persisted through month 21 of the follow-up. Patients with low to moderate risk scores (eg, 0-3) tended to have better outcomes in TEL than in STND, whereas patients with higher scores tended to have better outcomes in STND than in TEL. It is noteworthy that only 20% of the sample had composite risk indicator scores of 4 or higher.

The study had a number of strengths that increase confidence in the validity and generalizability of the results, including a large sample, 2 sites, manualized treatments, assessment instruments with good psychometric properties, corroboration of self-reports of alcohol and cocaine use by biological tests, and an excellent follow-up rate. Data on other treatment services received and self-help participation during the follow-up were also gathered to rule out these possible confounding factors. Finally, the participants were typical of patients in publicly funded, outpatient substance abuse treatment programs.^52,53

With regard to limitations, it is important to stress that the results of the study do not in any way suggest that telephone-based interventions will be more effective as an initial approach to the treatment of substance dependence than intensive outpatient treatment, relapse prevention, or group counseling. In that regard, cognitive-behavioral treatments have demonstrated efficacy as an initial treatment in numerous studies,^20,22-24 although there are some studies that find no difference between cognitive-behavioral treatments and other active treatments.^25-28 Another limitation of the study is that it did not include a control condition that provided no continuing care. Therefore, the effectiveness of the telephone-based continuing care intervention relative to no continuing care has yet to be established.

About half of the patients who started treatment in the IOPs did not complete those programs and were therefore ineligible for the continuing care study. Data suggested that IOP graduates were more likely to be in an institutional living arrangement, such as a sober house, than IOP dropouts and were less likely to have used alcohol or drugs in the 2 days preceding entrance into an IOP.²⁹ Therefore, the participants in the study should be considered as better-prognosis patients who had achieved some initial success in the IOPs. The telephone-based protocol actually included a combination of individual, group, and telephone sessions. Moreover, the number of group sessions was flexible, so that patients who were having difficulties that threatened their continued recovery could remain in the group beyond the 4-week cutoff. In this way, the telephone condition functioned as an adaptive or flexible intervention.⁵⁴

Similar results were generally obtained with the self-report and corroborating data. However, we also repeated the analyses with the total abstinence outcome measure, recoding patients who reported no use but had a cocaine-positive urine sample as nonabstinent at that follow-up. Although there were only a small number of such recodings, the rate was lowest in TEL, followed by RP and STND. These analyses produced results that did not differ from those of the original analyses reported here. Finally, we did not adjust α levels in the data analyses for the number of tests that were performed. If a somewhat more stringent α level had been adopted (eg, P<.01), the only result that would have been significant was lower GGT levels in TEL relative to RP.

In conclusion, our findings indicate that for most substance-dependent patients who complete an initial stabilization phase of outpatient treatment, telephone-based monitoring and brief counseling appears to be as effective a form of step-down continuing care as more intensive face-to-face treatments. Although formal economic evaluations have not been completed, it is likely that this condition will prove to be cost-effective as well. The mechanisms of action within the telephone condition are not yet known; however, it may work well for patients who have achieved initial stabilization because it is highly focused, more convenient than face-to-face continuing care, and does not interfere as much with other responsibilities, such as employment and child care.

The telephone may be a particularly effective method for delivering continuing care protocols to patients who have limited access to transportation, or to those with family or work-related responsibilities that preclude regular visits to substance abuse treatment clinics after initial stabilization has been achieved. In addition, telephone-based continuing care interventions with durations of longer than 3 months may lead to further improvements in the management of alcohol and drug use disorders.⁵⁵ In this regard, a recent study showed that regular contacts to assess recovery status provided over a 2-year period led to faster re-entry into treatment for patients who had relapsed and less need for further treatment at 24 months, compared with those in the control condition.⁵⁶ Furthermore, other continuing care interventions that have yielded significant effects in controlled studies have provided 12 months or more of treatment.⁶ Further research is needed to determine whether extended telephone monitoring, with the flexibility to increase the frequency or intensity of treatment when warranted through stepped-care protocols,⁵⁷ might be an effective and economical method for further improving the management of substance-dependent individuals over longer periods of time.

Correspondence: James R. McKay, PhD, Treatment Research Center, University of Pennsylvania, 3900 Chestnut St, Philadelphia, PA 19104 (mckay_j@mail.trc.upenn.edu).

Submitted for Publication: May 7, 2004; final revision received August 2, 2004; accepted August 19, 2004.

Funding/Support: This study was supported by grants R01-DA10262 and K02-DA000361 from the National Institute on Drug Abuse, Bethesda, Md, and grant R01-AA10341 from the National Institute on Alcohol Abuse and Alcoholism, Bethesda. Additional support was provided by the Medical Research Service of the Department of Veterans Affairs, Philadelphia, Pa. We thank Sara Ratichek, MA, and Janelle Koppenhaver, MA, for collection of research data; Rebecca Morrison, MA, RN, for coordination and supervision of continuing care treatment; and Jesse Chittams, MA, and Megan Long Beauchamp, MS, for assistance with data analyses.