Substance abuse treatment entry, retention and effectiveness: out-of-treatment opiate injection drug users

ELSEVIER

Drug and Alcohol Dependence 42 (1996) 11-20

Substance abuse treatment entry, retention and effectiveness: out-of-treatment opiate injection drug users’ Robert E. Booth”,*, Thomas J. Crowleya, Yiming Zhangb “Addiction

Research and Treatment Services, Department of Psychiatry, School of Medicine, University of Colorado Health Sciences Center, 1643 Boulder Street, Denver, CO 80211, USA “Department of Preventive Medicine and Biometrics, School of Medicine, University of Colorado Health Sciences Center, Denver, CO 80211, USA

Received 2 November 1995; accepted 24 April 1996

Abstract This study was designed to assess the effect of client characteristics and community interventions on treatment entry and retention, and to evaluate the relative effectiveness of treatment, compared to other interventions, in reducing drug use and crime

among out-of-treatment opiate injectors. Subjects (N = 2973) from 15 cities were randomly assigned to: standard intervention (SI)-HIV testing and counseling; or enhanced intervention (EI)-SI plus additional educational sessions stressing responsible drug use. EI in some cities included staff assistancewith treatment admission (i.e. ‘active’ referral). All locations provided intervention

by community outreach workers. Factors positively associated with treatment entry included: prior treatment, intervention by community workers, assignment to the EI, not injecting cocaine, injecting opiates, and fewer program interventions received. Sites where the EI included active referral achieved significantly higher treatment entry rates than sites where the EI did not. Findings supported the efficacy of treatment over other interventions in reducing drug use and arrests, the addition of staff assistance to facilitate clients’ entry into treatment, and the involvement of community outreach workers in achieving treatment entry. Keywords:

HIV; IDU; Drug use; Treatment

1. Introduction Drug use is associated with the human immunodefi-

ciency virus (HIV), the etiological agent for the acquired immunodeficiency syndrome (AIDS) (Friedland and Klein, 1987, Des Jarlais et al., 1989, Centers for Disease Control and Prevention, 1990). In the United States, injection drug users (IDUs) are the second largest risk group and the primary source for heterosexual transmission of HIV (Centers for Disease Control and Prevention, 1993). Among females, drug injection

* Corresponding author. Tel: + 303 433 4669; fax: + 303 458 6112. ’ This study was supported by the National Institute on Drug Abuse, grant DA-06912.

is the greatest risk factor for infection, followed by heterosexual contact with an infected male (Guinan and

Hardy, 1987, Campbell, 1990). The majority (56.8%) of females infected through heterosexual transmission report having had sex with an IDU (Centers for Disease Control and Prevention, 1993). These risks also play a key role in perinatal transmission:

lowing information

disseminated in the mass media and

by word-of-mouth (Kleinman et al., 1987, Selwin et al., 1987, Des Jarlais et al., 1989). Investigators assessing interventions

0376-8716/96/$15.000 1996 Elsevier Science Ireland Ltd. All rights reserved SO376-8716(96)01257-4

PII

through June 1993,

56.2% of US pediatric AIDS casesoccurred in children whose mothers were IDUs or the sexual partner of an IDU (Centers for Disease Control and Prevention, 1993). IDUs began to change their risk behaviours in some cities before formal interventions were established, fol-

targeting out-of-treatment

IDUs, includ-

12

R.E. Booth et al. / Drug and Alcohol Dependence 42 (1996) I I-20

ing interventions stressing responsible drug use, have reported reductions in the frequency of drug injecting (Neaigus et al., 1990, Booth and Wiebel, 1992,Anderson et al., 1993),needle sharing (Watters et al., 1990, Siegal et al., 1991, Stephenset al., 1991), and injecting in shooting galleries (Sufian et al., 1991). Unfortunately, however, at least one-third of IDUs continue high-risk injecting practices (Battjes and Pickens, 1993, Booth, 1994) and relapse to risk behavior is common among those who initiate risk reduction (Des Jarlais et al., 1991). These findings indicate the importance of emphasizing interventions that go beyond encouraging responsible drug use. Substance abuse treatment is a potentially powerful method to reduce the threat of HIV infection among drug users. Treatment has been associated with reduced drug injecting (Hubbard et al., 1988, Yancovitz et al., 1991), fewer HIV risk behaviours (Ball et al., 1988, Des Jarlais et al., 1990), and lower HIV seroconversion (Woody et al., 1992, Moss et al., 1994), as well as less criminal activity (DeLeon et al., 1979, Coombs, 1981, Simpson, 1981). Central to the effectiveness of treatment is its duration (Stimmel et al., 1978, Simpson et al., 1979, Des Jarlais et al., 1981). Outcomes for patients receiving less than 90 days of treatment are not significantly different from those for individuals receiving no treatment, an indication that 90 days is the minimum time for treatment to be effective (Simpson, 1979, 1981, Simpson and Sells, 1981). While these findings support the effectiveness of substance abuse treatment duration, the absence of randomized comparison groups as controls prevents establishing casual linkages. Clients with greater motivation, hence prognosis, may stay in treatment longer due to a self-selection bias. It is also important to note that patients in these studies were either already in the process of entering treatment or in treatment when the investigations began. It is unclear what factors were associated with treatment entry. It is also not clear what effect treatment might have had in reducing drug use, and consequently HIV risk, if compared to non-treatment interventions. The present investigation was designed to address these two issues, specifically: what factors, including client characteristics and various program and community interventions, could account for treatment entry; and what is the relative importance of treatment in reducing drug use and crime when compared to other non-treatment interventions.

sites were part of a National Institute on Drug Abuse, (NIDA) lCooperative Agreement designed to evaluate the efficacy of various community-based intervention strategies in reducing risk behaviours for HIV infection among IDUs (NIDA, 1989). Eligibility criteria included: (a) age over 17 years; (b) self-reported injection of opiates in the 30 days before the interview; (c) no treatment for substance abuse in the past 30 days; (d) not too intoxicated or otherwise dysfunctional to respond to questions; and (e) verification of drug use through visual inspection for signs of recent venipuncture and urinalysis for morphine, as well as for cocaine, metabolites. In most cites, Abuscreen ONTRAK (Roche Diagnostic Systems) was used for urinalysis. Prior to data collection, each site developed a plan for sampling subjects. Because it is impossible to randomly sample from a ‘hidden’ population of IDUs, recruitment locations within cities were selected using available local indicators (e.g. arrests for drug-related crimes, treatment admissions, observational studies). Selection was based on high densities of IDUs in targeted areas relative to other geographical areas in each city. In all sites, subjects were recruited by community outreach workers who explained the study and scheduled interviews. The study was again described by the interviewer, and informed consent obtained. Baseline data were collected between January, 1992 and December, 1993. Subjects were interviewed initially (tl) using a NIDA-developed structured questionnaire, the Risk Behavior Assessment(RBA). They were paid a modest stipend after the interview, as compensation for their time as research participants, and offered free HIV testing and counseling. Approximately 6 months later (t2), respondents were re-interviewed with the Risk Behavior Follow-up Assessment (RBFA). The RBA and RBFA questionnaires assesseddemographics, drug use, sexual behaviours, medical histories, and HIV/AIDS risk behaviours in the 30 day period prior to the interview. Reliability and validity assessmentsof the RBA support its adequacy as a research tool for this population (Weatherby et al., 1994, Dowling et al., 1994). All data underwent strict quality control before they were available for analyses. Procedures were reviewed and approved by institutional review boards at each site. 2.1. Interventions

2. Methods This prospective study included out-of-treatment opiate injectors recruited from 15 U.S. cities*. These ’ Anchorage, Columbus/Dayton, Denver, Detroit, Flagstaff, Houston, New Orleans, New York City, Long Beach, Miami, Philadelphia, Portland, Tucson, San Juan, and Washington, DC.

Sites randomly assigned subjects to one of two interventions, either a NIDA-developed Standard Intervention (SI) (NIDA, 1992), or the SI plus a site-specific Enhanced. Intervention (EI). The SI, a manual-driven intervention, consisted of HIV pre-test counseling within 7 days of the RBA, optional HIV testing, and provision of test results, with post-test counseling, approximately 14 days after the HIV test. In the pre-test

R.E. Booth et al. / Drug and Alcohol Dependence 42 (1996) 11-20

counseling session, nine cue cards describing basic information about AIDS and how to reduce the risk of HIV infection were discussed with the subject. Subjects also rehearsed how to clean injection equipment and use a condom, using an anatomical model. At the post-test counseling session, test results were provided and six of the original nine cue cards were shown along with one of two sets of three new cards, based on whether or not the subject was HIV infected. Subjects declining to have their blood drawn could remain in the study if they attended both counseling sessions. If either session was missed, however, they were excluded from the national assessmentand, thus, from this study as well. At both sessions, counselors provided written materials, including a description of the study, a list of drug treatment agencies in the area (i.e. ‘passive’ treatment referral), a list of HIV prevention and testing agencies, and if requested by the subject, a list of non-HIV related social or economic services. EIs were developed individually at each site. Components included education, motivation, strategizing, and involvement in community organization designed to create acceptance and adoption of HIV-related risk reduction behaviours (NIDA, 1995). All EIs addressed drug and sexual risk factors for HIV infection, barriers to behavioral change, supports for risk reduction, and ways to avoid situations where risk activities were likely to occur. At each site, EI sessions were conducted after the two SI sessions and typically consisted of one to three sessions and the use of audio visual materials. Two-thirds of the sites employed health educators to present the intervention, while in the remaining sites, nurses, health psychologists, or social workers served as interventionists. The vast majority of the interventions were conducted in project offices. In seven cities, the EI included ‘active’ referral to substance abuse treatment, in which program staff assisted clients in scheduling and arranging for treatment admission. In the other eight cities, treatment referral for the EI was the same as that for the SI (i.e. passive referral). Due to site differences in the EI, analyses (see below) controlled for the number of intervention sessions and active versus passive treatment referral, in addition to the site itself. All sites employed community outreach workers to recruit subjects for study participation. Typically, these staff were recovering drug users, or individuals familiar with the drug subculture, and from the communities targeted for intervention. In this way they had relatively easy accessto the IDU population. The responsibility of the outreach worker was to penetrate social circles of IDUs, develop a relationship of trust with them, explain the project’s purposes, and recruit study participants. In this process, outreach workers distributed free prevention materials, including bleach

13

and condoms, and provided basic information about HIV infection and AIDS. Since the majority of their time was spent in the community, they were available for personalized, in-depth, interventions when approached by members of the target population. Unlike office-based sessions, these interventions were not planned and focused on issues raised by the subject. To assesssubstance abuse treatment, we defined a treatment group made up of respondents who had been in methadone maintenance, drug-free outpatient, or residential therapy for at least 90 days between the two interviews. This information was obtained at t, when subjects were asked if they had been in treatment since t, and, if so, for how many days. Measures of program and community interventions, other than treatment, included: intervention by community outreach workers; the number of planned and unplanned program intervention sessions provided to each subject; SI versus EI; and, in separate analyses, active treatment referral within the EI versus SI and passive treatment referral within the EI versus SI. Information regarding interventions by community outreach workers was obtained from the subject at t,. The number of planned and unplanned intervention sessions attended was recorded by each program during the course of conducting interventions and submitted on a standardized form. Site directors supplied information on whether their EI included active or passive treatment referral. 2.2. Statistical procedures

SAS for Windows was used to perform all statistical analyses (SAS, 1992). For univariate comparisons, chi-square tests were used to examine the relationship between categorical variables; two sample t-tests were used to examine the relationship of continuous variables. In cases of highly skewed distributions, twotailed Fisher’s Exact Test and Wilcoxon rank sum test were used, respectively. To assess the independent contribution of demographics, type of drugs used, total injection frequency, treatment, and program and community interventions on behavior change for continuous variables (e.g. total times injected), analysis of covariance was used, adjusting for initial scores in the outcome variable at t,. For categorical variables (e.g. injected opiates), we used multiple logistic regression, again controlling for the effect of the outcome variable at t,. Multiple logistic regression was also used to predict factors associated with treatment entry, with the same independent variables used to assess behavior change (except treatment). Because of site differences, notably in the EI, 14 dummy variables were created to indicate the 15 sites and included in all linear and logistic regression models.

14

R.E. Booth et al. /Drug

and Alcohol Dependence 42 (1996) 11-20

3. Results

The average time between t, and t2 interviews was 186 days (range 150 to 284). Of the 4443 subjects assessedat t,, 2973 (67%) were recontacted and interviewed at t,. Analyses of demographics and drug use at t, showed that those lost-to-follow-up were more likely to be male, not African-American, and opiate injectors who also injected cocaine. Frequency of drug injecting and crack smoking was greater among subjects not re-interviewed than those re-interviewed. There were no differences found in arrest history, prior treatment, employment status, or t, morphine or cocaine urine results. To determine if there were demographic, prior arrest, prior treatment, or drug use differences, including types of drugs used and their frequency of use, EI subjects lost-to-follow-up were compared to SI subjects lost-to-follow-up. No differences were found. The 2973 IDUs interviewed at t2 are the subjects of this report. They averaged 38.8 years of age, 73.0% were male, 46.0% African-American, 26.0% white, and 22.9% Latino. Only 5.9% were employed full time. Most (86.0%) had a prior arrest and 52.2% had previously been in substance abuse treatment. All subjects had injected opiates in the 30 days prior to t, and most had injected other drugs and/or smoked crack cocaine: 67.0% injected heroin, 38.0% speedball (a combination of heroin and cocaine), 3.8% other opiates; 61.3% cocaine and 51.2% smoked crack. In this same time period, the average number of injections of heroin was 42.0, speedball 35.1, other opiates 13.2, and cocaine 38.5. In addition, subjects reporting crack use smoked an average of 53.1 times in the 30 days prior to the interview. Among the 2678 subjects tested for HIV at t,, 7.2% were infected. No significant t, differences were found between subjects assigned to the SI and EI on demographics, drug use, or, at t,, on receipt of interventions by community outreach workers (approximately 40% in each group reported receiving community outreach intervention). However, EI participants were seen during the study an average of 1.6 times in planned and unplanned intervention sessions, in addition to the two SI sessions,while SI participants were seen an average of 0.4 times (t = 14.35, P < 0.0001). 3.1. Treatment entry

Overall, 250 subjects (8.4%) entered and remained in treatment for at least 90 days between t, and t,. Of those, 74.8% were still in treatment during the 30 days prior to the tz interview. Clients in treatment included 60.2% who received methadone maintenance, 28.4% drug-free outpatient, and 28.8% residential treatment. These figures exceed 100% since 42 subjects (16.8%) entered more than one treatment modality. Selected

demographic and baseline characteristics for the intreatment and out-of-treatment cohorts are presented in Table 1. Subjects in treatment for 90 days or more were more likely to have been opiate only injectors than opiate injectors who also injected cocaine or smoked crack, as indicated by both self-reports and urinalyses. Prior treatment and being white or Latin0 were also associated with receiving treatment. The average age of treatment recipients was 39.6 compared to 38.7 for those not receiving treatment (ns). Treatment recipients averaged 76.5 injections in the 30 days prior to the t, interview, compared to 65.8 for non-treatment recipients (ns). Results from the logistic regression analysis of predictors of treatment entry and retention at 90 days are shown in Table 2. None of the demographic nor site dummy variables were significant, however six variables were significant, including all three intervention variables: assignment to the EI; intervention by community outreach workers; and participation in fewer planned and unplanned intervention sessions. Among subjects assigned to the EI, 9.8% entered treatment compared to 6.8% in the SI (P < 0.01 in univariate analysis, table not shown). Of those who reported they were intervened with during the study by community outreach workers, 10.1% entered treatment compared to 7.0% for those not seen by community outreach workers who also entered treatment (P < 0.01). Subjects entering treatment averaged.58 intervention sessionscompared to.83 for those not entering treatment (P < 0.01). As Table 2 illustrates, treatment entry was also more likely among opiate injectors who did not inject cocaine than among injectors of both opiates and cocaine, as well as among those with prior treatment experiences compared to those who had never entered treatment prior to this study. Sites differed in whether they actively or passively referred clients into treatment as part of their EI. Of the 15 study sites, seven used active treatment referral and eight passive treatment referral. Two dummy variables were created to determine if the association observed between the EI and treatment entry was a result of an overall effect of the EI, whether or not it emphasized treatment, or a result of active treatment referral within the EI. The first dummy variable compared the effects of EI/active-referral to SI, while, in the second, EI/passive-referral was compared to SI. Independent measures used in the prior logistic regression were also included in the model. In separate analyses, results showed that the EI/active-referral was independently, and significantly, associated with entering treatment when contrasted with the SI (OR= 1.58, CI= 1.18-2.13, P < O.Ol), while assignment to the EI/passive-referral was not. A comparison between SI, EI/passive referral, and EI/active referral (table not shown), found treatment rates of 6.8%, 8.6%, and 10.6%, respectively (P <

15

R.E. Booth et al. / Drug and Alcohol Dependence 42 (19%) 11-20

Table 1 Baseline demographics and drug use (past 30 days) of 2973 injection drug users who did or did not receive 90 days of treatment before follow-up Variable

% of Treated subjects (N)

% of Untreated subjects (N)

x2

P

Gender Male Female

69.2 (173) 30.8 (77)

73.3 (1997) 26.7 (726)

1.99

ns

Ethnicity African American White Latin0

40.2 (94) 33.8 (79) 26.1 (61)

49.2 (1273) 26.9 (695) 24.0 (620)

7.73

0.05

Prior arrest Yes No

83.8 (207) 16.2 (40)

84.0 (2255) 16.0 (429)

0.01

ns

Prior treatment YeS No

70.4 (176) 29.6 (74)

50.5 (1375) 49.5 (1347)

36.29

0.000

Injected heroin Yes No

79.1 (197) 20.9 (52)

65.9 (1789) 34.1 (927)

18.10

0.000

Injected cocaine Yes No

48.8 (122) 51.2 (128)

62.4 (1695) 37.6 (1020)

17.93

0.000

Injected speedball Yes No

41.6 (104) 58.4 (146)

37.7 (1023) 62.3 (1694)

1.52

ns

Smoked crack Yes No

41.4 (103) 58.6 (146)

52.1 (1417) 47.9 (1303)

10.51

0.001

UA positive: heroina Yes No

62.7 (133) 37.3 (79)

53.6 (1215) 46.4 (1052)

6.53

0.02

UA positive: cocaine” Yes No

57.1 (129) 42.9 (97)

69.0 (1767) 31.0 (795)

13.49

0.000

YJA indicates urinalysis.

0.01). As observed earlier, prior treatment (OR = 2.04, CI = 1.52-2.71, P c O.OOOl),intervention by community outreach workers (OR = 1.75, CI = 1.33-2.27, P < O.OOOl),injecting opiates (OR = 1.59, CI = 1.142.22, P < O.Ol), not injecting cocaine (OR = 0.68, CI = 0.51-0.88, P < O.Ol), and fewer number of planned and unplanned interventions received from the program (OR = 0.86, CI = 0.77-0.95, P < 0.01) were also positively associated with treatment entry and retention. 3.2. Treatment effects

Differences in total self-reported frequency of drug injecting between t, and t2 for subjects who did or did not enter treatment appear in Fig. 1. Mean levels of the

outcome variables and corresponding two standard errors of the means are plotted. In an analysis of covariante, three variables were significantly associated with total drug injection at t, after controlling for site: drug injection frequency at t, (regression coefficient = 0.36, CI = 0.33-0.39, P c 0.0001); treatment entry (regression coefficient = - 27.25, CI = - 36.64- 17.86, P < O.OOOl),and Latin0 versus African-American ethnicity (regression coefficient = 15.43, CI = 7.1l-23.76, P < 0.001). In Fig. 2, self-reported use of each drug is plotted, along with urinalysis findings for cocaine and morphine. Use of each drug declined, as did the average number of injections, for respondents in both groups. Self-reported change was greater, and supported by urinalysis, for those who entered treatment

16

R.E. Booth et al. / Drug and Alcohol Dependence 42 (1996) 11-20

Table 2 Summary of logistic regression analysis: predictors of treatment entry and retention among 2973 injection drug usersa 95% CI

P

2.04

(1.52-2.70)

1.59 1.72 1.59

(1.20-2.08) (1.32-2.22) (0.51-0.88) (1.15-2.22)

0.85

(0.77-0.95)

0.0001 0.01 0.0001 0.01 0.01 0.01

Independent variable

Adjusted odds ratios

Prior treatment EI (vs. SI) Seen by outreach worker Injected cocaine Injected heroin Number of interventions received

0.67

“Since the dependent variable in this analysis is a postive rather than a negative outcome, variables with odds ratios greater than 1.00 are favorable; the independent variables shown above, plus demographics (age, gender, and ethnicity), site, injected speedballs and smoked crack, were adjusted for one another to determine their unique contribution to the dependent variable.

than for those who did not. Subjects who entered treatment were also less likely to have been arrested by t, than those who did not enter treatment (14.1% versus 24.2%, P < 0.0001). Associations with drug use and crime at t,, using client characteristics and interventions as independent variables, are summarized in the logistic regression shown in Table 3. Most of the site dummy variables were significant in the analyses. For clarity they are not shown in the table. Higher frequencies of drug use and life-time arrests at t, were significantly, and independently, associated with higher scores at t, in each analysis. Whites and Latinos were more likely to report continued drug injection at t, than African-Americans, however, they were less likely to be smoking crack. Substance abuse treatment was consistently associated with reductions in drug use, from two-and-a-half to three-fold, in both self-reports and urinalyses, as well as with fewer arrests. One significant association, crack smoking, was found with group assignment, favouring the EI over the SI. The number of program intervention sessions received was protective in the analyses of injecting cocaine, injecting speedball, and cocaine urinalysis. Intervention by outreach workers was associated with subjects’ positive urine for cocaine. To determine if the associations found between substance abuse treatment and drug use and arrests were due to differences in treatment modality, in particular,

Fig. 1. Comparisons between treated and not treated opiate injectors: self-reported number of drug injections in preceding 30 days at I, and t2.

the possible ‘protection’ offered by residential treatment, we conducted separate logistic regression analyseswith each of the three modalities (tables not shown). On all substance use variables, the effects of treatment were similar to those shown on Table 3 and, with the exception. of self-reported speedball injection for clients in methadone maintenance and morphine urinalysis for drug-free outpatients, results were also significant. Findings showed that the highest significance levels and odds ratios were in residential treatment, followed by drug-free outpatient care and methadone maintenance. Arrest rates at follow-up indicated that all modalities were protective, however, significance was reached only with subjects in residential treatment. Finally, we assessedpredictors of overall drug-free status at t,, using both aggregated self-report data for the four drug types and, in a separate analysis, aggregated data from urinalyses. At t,, 22% of respondents reported not using any drugs in the past 30 days. In logistic regression, with the independent variables from Table 3, subjects who self-reported abstinence at follow-up were more likely to have received treatment versus no treatment (OR = 3.33, CI = 2.56-4.55, P < -

100 ReportedUrine

eported peedball ejection

dcaine ljection

Reported Urine Crack Cocaine Smoking Positive

Fig. 2. Comparisons between treated and not treated opiate injectors: self-reported injected drug use, crack smoking, and urinalyses in preceding 30 days at t, and t,. Note: error bars represent 2 standard errors derived from the binomial distribution.

17

R.E. Booth et al. 1 Drug and Alcohol Dependence 42 (1996) 11-20

Table 3 Summary of logistic regression analyses: predictors of drug use and arrests among 2973 injection drug users Independent variable Adjusted odds ratios (95% CI) Injected heroin Client characteristics Baseline drug use 13.87**** or arrests (11.06-17.41) Age (older) ;;:Ko4) 1.24* Female (vs. male) (1.01-1.51) White (vs. African ;iB:;f:T40) American) Latin0 (vs. African 2.46**** American (1.95-3.10)

Injected cocaine

;K;r73) 1.01’ (1.00-1.02) 1.01 (0.83-1.22) 1.45*** (1.18-1.78) 2 33**** (;.89-2.88)

Injected speedball

Smoked crack

UA positive heroin

UA positive cocaine

Arrested

9.24**** (7.51-11.37)

11.59** (9.45-14.21) 1.oo (0.99-1.01) 1.14 (0.92- 1.41) 0.41**** (0.32-0.52) 0.50**** (0.39-0.64)

7.72+*** (6.29-9.48) 1.01 (1.00-1.02) 1.16 (0.93-1.46) 1.03 (0.81-1.31) 1.42** (1.11-1.83)

4.25*** (3.50-5.16) 1.oo (0.98-1.01) 1.32** (1.07-1.63)

3.11**** (2.25-4.30) 0.96**** (0.95-0.97) 0.88 (0.71-1.08)

;;:;I::,,) 0.59**** (0.47-0.74)

:$-1.60) 1.07 (0.84-1.35)

0.3s**** (0.27-0.54) 1.09 (0.89-1.33) 0.93 (0.86-1.00)

0.37**** (0.26-0.51) 0.99 (0.82-1.20) 0.94* (0.88-1.00)

0.53** (0.36-0.78) 1.10 (0.92-1.33) 0.99 (0.94-1.06)

1.05 (0.86-1.28)

1.29** (1.07-1.56)

1.12 (0.93-1.34)

1.01

(1.00-1.02) 0.98 (0.78- 1.23) 1.06 (0.82- 1.36) 1.79**** (1.40-2.28)

Interventions Treatment EI (vs. SI) Number of program interventions Seen by out reachworker

;li1E48) 1.03 (0.86- 1.23) 1.04 (0.98-1.11)

;GK62, 1.12 (0.94- 1.33) 0.90*** (0.85-0.95)

;lZZ61, 1.20 (0.98-1.47) ;&3:Y90)

;K;T45) 0.79* (0.65-0.95) 1.09** (1.02-1.16)

1.14 (0.96-1.37)

0.98 (0.82-1.16)

1.21 (0.99- 1.48)

1.12 (0.93-1.36)

*p<0.05; **p
O.OOOl),and to have been white rather than AfricanAmerican (OR = 1.43, CI = 1.12-1.79, P < 0.01). Older aged (OR = 0.97, CI = 0.96-0.98, P < 0.0001) and female (OR = 0.69, CI = 0.56-0.86, P < 0.001) subjects were more likely to have reported continued drug use. A similar logistic regression, using combined urinalyses results, showed that subjects who received treatment were nearly four times more likely to have drug-negative urines at t2 than those not receiving treatment (OR = 3.70, CI = 2.70-5.00, P < 0.0001). Other factors associated with drug-negative urines were being white (OR = 2.13, CI = 1.64-2.78, P < 0.0001) or Latin0 (OR = 1.49, CI = 1.14-1.96, P < 0.01). Similar to selfreported drug use, females had higher rates of drugpositive urines than did males (OR = 0.67, CI = 0.76-0.85, P < 0.01). Aside from the associations observed on the number of intervention sessions,interventions other than treatment were of relatively little significance. Additional univariate analysis of EI compared to SI found no significant differences. 4. Discussion Nearly 3000 out-of-treatment opiate injectors were followed approximately six months after receiving an initial risk assessmentinterview and HIV testing and

counseling. Both IDUs who went to treatment and those who did not reported reduced drug use by t,. However, subjects who entered and stayed in treatment for at least 90 days reported significantly greater reductions in drug injection and crack smoking, corroborated by urinalyses, and they were less likely to have been arrested, than subjects who did not enter treatment. At the follow-up interview, three-fourths of this group were still in treatment. Despite the significance of these findings, the study’s design does not permit inferences about casual linkages between substance abuse treatment and outcomes. Treatment entry was associated with two prominent factors. First, subjects randomly assigned to the EI intervention, particularly in those locations where the EI included active treatment referral, were more likely to have entered and stayed in treatment for at least 90 days than subjects assigned to the SI. This finding underscores the need for interventions that assist outof-treatment IDUs in the treatment entry process, including scheduling intakes, providing transportation, and waiving admission fees. Second, subjects intervened with by community outreach workers were more likely to have entered treatment than subjects lacking interventions by outreach workers. This finding offers strong support for the efforts of community outreach workers advocating for IDUs. Increased treatment en-

18

R.E. Booth et al. / Drug and Alcohol Dependence42 (1996) 11-20

try was also predicted by fewer intervention sessions, a probable artifact due to lack of availability, or desire, of those who entered treatment to receive interventions. Despite the effect of EI intervention with active referral and the role of community outreach intervention in facilitating treatment entry, fewer than 10% of the IDUs in this study entered treatment and remained for 90 days. In view of the pivotal role IDUs play in HIV transmission, to both other IDUs and non-IDUs (Friedland and Klein, 1987, Des Jarlais and Friedman, 1987) and the protective influence of treatment in reducing HIV risk behaviours (Ball et al., 1988, Des Jarlais et al., 1988, 1990), this finding is troubling. It underscores the need for new intervention approaches that increase the appeal of treatment as an option to high-risk behaviours. At a minimum, this low treatment rate calls for increased community outreach to IDUs, incorporating active treatment referral and the removal of barriers to rapid intake (Woody et al., 1975). Interventions that could augment community outreach and potentially improve the attractiveness of treatment, such as motivational interviewing (Miller and Rollnick, 1991, Rollnick et al., 1992) and role induction (Stark and Kane, 1985, Ravndal and Vaglum, 1992), as well as approaches that could increase retention rates following entry, such as structured contingency programs (McCall et al., 1984, Stitzer et al., 1986), should be considered. Since only 15% of IDUs are estimated to be in treatment at any time (Schuster, 1988) we also need to further our understanding of the determinants of treatment entry and retention. Opiate injectors who also injected cocaine were less likely to have entered and stayed in treatment than those who only injected opiates. Treatment for opiate/ cocaine injectors was successful, however, since both self-reported and urine-verified cocaine use at follow-up was significantly less among those in- than out-of-treatment. Numerous investigators have reported relationships between injecting cocaine, needle risk, and HIV (Chaisson et al., 1989, Guydish et al., 1990, Wiebel et al., 1990, Friedman et al., 1993). High treatment attrition rates among cocaine abusers have frequently been noted (Agosti et al., 1991, Kleinman et al., 1992, Gainey et al., 1993, Gawin and Kleber, 1994) prompting suggestions for more intensive and intrusive treatment (Kleinman et al., 1992), as well as strategies that build upon social networks of non-users and provide incentives for life-style change (Gainey et al., 1993). Our findings from IDUs in 15 cities further highlight the need for developing more effective interventions for cocaine abusers. This study also found that Latinos were less likely to have stopped injecting drugs than were African-Americans, although they were more likely to have stopped smoking crack. These findings may, in part, be due to the persistence of initial drug use patterns. At both

interviews, Latinos were the most likely to report injecting opiates and African-Americans the least likely, while African-Americans had the highest rate of crack smoking and Latinos the lowest. The finding that females were less likely to have ceased using drugs, according to both aggregated self-reports and urinanalysis, was unexpected, since there were no differences found in rates of treatment entry according to gender. We could find no studies directly relating to this topic with out-of-treatment IDUs, although an assessmentof early attrition from treatment for cocaine abuse found that females were significantly more likely than males to have completed treatment (Agosti et al., 1991). Similarly, an investigation of factors associated with entry into methadone maintenance found that females were more likely than males to have entered this modality (Schutz et al., 1994). Findings in the present study are inconsistent with these results and suggest the need for further investigation into the role gender plays in reducing high risk drug behaviours, including treatment entry and retention. There are several limitations to this study. First, many of the results we reported were based on self-reports; these data may have been affected both by recall and social desirability. However, the short time period (30 days) respondents were asked to recall may have minimized recall error. Most importantly, while social desirability cannot be eliminated, urinalyses consistently supported self-report data. Second, approximately 30% of respondents interviewed initially were not located for follow-up. These subjects tended to use drugs more frequently at t, than those returning for the second interview, although no differences were found in urinalyses at t1 between those interviewed versus not interviewed at tZ. Becausedifferences were not found in those lost-to-follow-up according to the assigned intervention, subjects lost in each group may have been similar to one another in their prognosis and eventual outcomes. Third, the 6-month period between interviews may not be an adequate time frame to study the persistent effect of interventions on behavior change. Finally, due to variations across sites, notably in the EI, which was not standardized, pooled data are not directly comparable. Although we feel these data are appropriate to examine for general trends across sites, after adjusting for site in the linear and logistic regression models, it is recommended that future national studies, such as the one described here, consider standardized manual-driven interventions for all conditions. To our knowledge, this is the only large-scale prospective assessmentof the effects of treatment entry on substance use by opiate injectors not in treatment when the study began. It is also the only large-scale study we are aware of in which self-reports of drug use were corroborated by urinalyses. Results offer evidence that substance abuse treatment can reduce drug ‘use

R.E. Booth et al. / Drug and Alcohol Dependence 42 (1996) 11-20

and, consequently, risk for HIV infection, as well as arrests, among out-of-treatment IDUs. By comparing the effects of substance abuse treatment in reducing substance use, as well as criminal activity, with other program and community interventions designed to modify drug-related behaviours, we extend previous findings supporting the role of drug treatment (Bale et al., 1980, Hubbard et al., 1988, Des Jarlais et al., 1990, Yancovitz et al., 1991). The 6-month interval between interviews did not allow sufficient time to assessthe protective effect of treatment on HIV seroconversion. However, a recent longitudinal study found that attendance in methadone maintenance of 12 months or more was highly protective, and that declines in seroconversion coincided with changes in risk behavior, including reduced cocaine injection (Moss et al., 1994). Our findings support the efficiency of treatment in reducing drug use compared to interventions emphasizing safer drug use. We extend others’ observations that patients continue in treatment longer when they receive rapid initial response and individual attention (Stark, 1992). We show that they are more likely to enter treatment if actively referred and that intervention by community outreach workers increasestreatment rates. Educational interventions combined with active treatment referral influence treatment-seeking, which in turn influence the risk of HIV transmission through drug injection. Acknowledgements Support for this study was provided by the National Institute on Drug Abuse, grant DA-06912. Parts of this paper were presented at the 1995 Conference on AIDS and Drug Abuse, Scottsdale, AZ, June 9, 1995. We acknowledge the assistance provided by Richard Needle, Ph.D., M.P.H. and Francisco Pinto, M.P.H. We would also like to thank the anonymous reviewers for their thoughtful and helpful suggestions concerning the original manuscript. References Agosti, V., Nunes, E., Stewart, J.W. and Quitkin, F.M. (1991) Patient factors related to early attrition from an outpatient cocaine research clinic: a preliminary report. Int. J. Addict. 26, 327-334. Anderson, M.D., Smereck, G.A. and Braunstein, MS. (1993) LIGHT model: an effective intervention model to change high-risk AIDS behavior among hard-to-reach urban drug users. Am. J. Drug. Alcohol Abuse 19, 309-325. Bale, R.N., Van Stone, W.W., Kuldau, J.M., Engelsing, T.M., Elashoff, R.M. and Zarcone, V.P. (1980) Therapeutic communities versus methadone maintenance. A prospective controlled study of narcotic addiction treatment: design and one-year followup. Arch. Gen. Psychiatry 37, 179-193. Ball, C.C., Lange, W.R., Myers, C.P. and Friedman, S.R. (1988) Reducing the risk of AIDS through methadone maintenance treatment. J. Health Sot. Beh. 29, 214-226.

19

Battjes, R.J. and Pickens, R.W. (1993) Trends in HIV infection and AIDS risk behaviours among injecting drug users in selected US cities. IX International Conference on AIDS, Berlin, Abstract PO-Cl S-2950. Booth, R.E. (1994) Predictors of unsafe needle practices: injection drug users in Denver. J. of Acquir. Immun. Defic. Syndr. 7, 504-508. Booth, R.E. and Wiebel, W.W. (1992) The effectiveness of reducing needle-related risks for HIV through indigenous outreach to injection drug users. Am. J. Addict. 1, 277-288. Campbell, C.A. (1990) Women and AIDS. Sot. Sci. Med. 30, 407415. Centers for Disease Control and Prevention (1990) HIV Monthly Surveillance Report, MMWR. Centers for Disease Control and Prevention (1993) HIV Monthly Surveillance Report, MMWR. Chaisson, R.E., Bacchetti, P., Osmond, D., Brodie, B., Sande, M.A. and Moss, A.R. (1989) Cocaine use and HIV infection in intravenous drug users in San Francisco. J. Am. Med. Assoc. 261, 5615. Coombs, R.H. (1981) Back on the streets: therapeutic communities impact upon drug users. Am. J. Drug Alcohol Abuse 8, 185-201. DeLeon, G., Andrews, M., Wexler, H.K. et al (1979) Therapeutic community dropouts: criminal behavior five years after treatment. Am. J. Drug Alcohol Abuse 6, 2533271. Des Jarlais, D.C. and Friedman, S.R. (1987) HIV infection among intravenous drug users: epidemiology and risk reduction. AIDS 1, 67-76. Des Jarlais, D.C., Joseph, H. and Dole, V.P. (1981) Longterm outcomes after termination from methadone maintenance treatment. Ann. NY Acad. Sci. 362, 231-238. Des Jarlais, D.C., Friedman, S.R. and Stoneburner, R.L. (1988) HIV infection and intravenous drug use: critical issues in transmission dynamics, infection outcome, and prevention. Rev. Infect. Dis. 10, 151-158. Des Jarlais, D.C., Friedman, S.R., Novick, D. et al (1989) HIV-l infection among intravenous drug users in Manhattan, New York City, from 1977 through 1987. J. Am. Med. Assoc. 261, 10081012. Des Jarlais, DC., Friedman, S.R. and Casriel, C. (1990) Target groups for preventing AIDS among intravenous drug users. Il. The ‘hard’ data studies. J. Consult. Clin. Psychol. 58, 50-56. Des Jarlais, D.C., Abdul-Quader, A. and Tross, S. (1991) The next problem: maintenance of AIDS risk reduction among intravenous drug users. Int. J. Addict. 26, 1279-1292. Dowling, S., Johnson, M.E., Fisher, D.G. et al (1994) Reliability of drug users self-reported recent drug use. Assessment 1, 383-392. Friedland, G.H. and Klein, R.S. (1987) Transmission of the human immunodeficiency virus. N. Engl. J. Med. 317, 1125-1135. Friedman, S.R., Rodriguez, A., Zunzunegui, V. et al. (1993) Similarities in syringe sharing among drug injectors in Madrid and New York City. American Public Health Association Annual Meeting, San Francisco. Gainey, R.R., Wells, E.A., Hawkins, J.D. and Catalano, R.F. (1993) Predicting treatment retention among cocaine users. Int. J. Addict 28, 487-505. Gawin, F.H. and Kleber, H.D. (1994) Cocaine abuse treatment: open trial with desipramine and lithium carbonate. Arch. Gen. Psychiatry 41, 905. Guinan, M.E. and Hardy, A. (1987) Epidemiology of AIDS in women in the United States, 1981 through 1986. J. Am. Med. Assoc. 257, 19-22. Guydish, J.R., Abramowitz, A., Woods, W., Black, D.M. and Sorenson, J.L. (1990) Changes in needle sharing among intravenous drug users: San Francisco, 198661988. Am. J. Public Health 80, 995-997.

20

R.E. Booth et al. / Drug and Alcohol Dependence 42 (1996) 11-20

Hubbard, R.L., Marsden, M.E., Cavanaugh, E., Rachal, J.V., Ginzberg, H.M. (1988) Role of drug abuse treatment in limiting the spread of AIDS. Rev. Infect. Dis. 10, 377-384. Kleinman, P.H., Friedman, ST., Mauge, C.E., Goldsmith, D.S., Des Jarlais, D.C. and Hopkins, W. (1987) Beliefs and behaviours regarding AIDS: a survey of street intravenous drug users about AIDS prevention. III International Conference on AIDS, Washington, DC. Kleinman, P.H., Kang, S.-Y., Lipton, D.S., Woody, G.E., Kemp, J. and Millman, R.B. (1992) Retention of cocaine abusers in outpatient psychotherapy. Am. J. Drug Alcohol Abuse 18, 29943. McCall, M.E., Stitzer, M.L., Bigelow, G.E. and Liebson, J.A. (1984) Contingency management interventions: effects on treatment outcome during methadone detoxification. J. Appl. Behav. Anal. 17, 35-43. Miller, W.R. and Rollnick, S. (1991) Motivational Interviewing: Preparing People to Change Addictive Behaviours. Guilford Press, New York. Moss, A.R., Vranizan, K., Gorter, R., Bacchetti, P., Watters, J. and Osmond, D. (1994) HIV seroconversion in intravenous drug users in San Francisco, 1985-1990. AIDS 8, 223-231. National Institute on Drug Abuse (1989) A Cooperative Agreement for AIDS Community-Based Outreach/Intervention Research. Program Announcement DA 90-02. Department of Health and Health and Human Services, Public Health Service. National Institute on Drug Abuse National Institute of Health (1992) The Standard Intervention of the Cooperative Agreement Program for Community-Based Outreach/Intervention Research. Department of Health and Health and Human Services, Public Health Service. National Institute on Drug Abuse National Institute of Health. (1995) Cooperative Agreement for AIDS Community-Based Outreach/Intervention Research Program, 1990-Present.Community Research Branch, Division of Epidemiology and Prevention Research, Department of Health and Health and Human Services, Public Health Service. Neaigus, A., Sufian, M., Friedman, S.R. et al (1990) Effects of outreach intervention on risk reduction among intravenous drug users. AIDS 2, 253-271. Ravndal, E. and Vaglum, P. (1992) Different intake procedures: the influence of treatment start and treatment response - a quasi-experimental study. J. Subst. Abuse. Treat. 9, 53-58. Rollnick, S., Heather, N. and Bell, A. (1992) Negotiating behaviours change in medical settings: the development of brief motivational interviewing. J. Mental Health 1, 25-37. SAS Institute. Inc. (1992) The SAS System for Windows 3.10, Release 6.08. Cary, NC. Schuster, C.R. (1988) Intravenous drug use and AIDS prevention. Public Health Rep. 103, 261-266. Schutz, C.G., Rapiti, E., Vlahov, D. and Anthony, J.C. (1994) Suspected determinants of enrolment into detoxification and methadone maintenance treatment among injecting drug users. Drug Alcohol Depend. 36, 129-138. Selwin, P.A., Feiner, C., Cox, C.P., Lipshutz, C. and Cohen, R.L. (1987) Knowledge about AIDS and high-risk behavior among intravenous drug users in New York City. AIDS 1, 247-254. Siegal, H.A., Falck, R.S., Carlson, R.G. and Ling, L. (1991) Changes in needle risk behaviours: a preliminary comparative analysis of the Dayton-Columbus NADR program’s standard and enhanced

intervention tracks at six months. Research in Progress, NOVA Research Co., Washington, DC, October, pp. 12-16. Simpson, D.D. (1979) The relation of time spent in drug abuse treatment to post-treatment outcome. Am. J. Psychiatry 136, 1449-1453. Simpson, D.D. (1981) Treatment from drug abuse: follow-up outcomes and length of time spent. Arch. Gen. Psychiatry 38, 875880. Simpson, D.D. and Sells, S.B. (1981) Highlights of the DARP Follow-up Research on the Evaluation of Drug Abuse Treatment Effectiveness. Institute of Behavioral Research, Texas Christian University, Fort Worth, TX. Simpson, D.D., Savage, L.V. and Lloyd, M.R. (1979) Follow-up evaluation of drug abuse during 1969 to 1972. Arch. Gen. Psychiatry 36, 772-780. Stark, M.J. (1992) Dropping out if substance abuse treatment: a clinically oriented review. Clin. Psychol. Rev. 12, 93-l 16. Stark, M.J. and Kane, B.J. (1985) General and specific psychotherapy role induction for substance-abusing clients. Int. J. Addict. 20, 1135-1141. Stephens, R.C., Feucht, T.E. and Gibbs, B.H. (1991) The impact of level of program intervention on the risk behavior of IVDUs in Cleveland. Research in Progress, NOVA Research Co., Washington, DC, October, pp. 22-26. Stimmel, B., Goldberg, J., Cohen, M. and Rotkopf, E. (1978) Detoxification from methadone maintenance: risk factors associated with relapse to narcotic use. Ann. NY Acad. Sci. 311, 183-180. Stitzer, M.L., Bickel, W.K., Bigelow, G.E. and Liebson, J.A. (1986) Effect of methadone dose contingencies on urinalysis test results of polydrug-abusing methadone-maintenance patients. Drug Alcohol Depend. 18, 341-348. Sufian, M., Friedman, S.R., Curtis, R., Neaigus, A. and Stepherson, B. (1991) Organizing as a new approach to AIDS risk reduction for intravenous drug users. In: Cocaine, AIDS, and Intravenous Drug Use (Friedman, S.R., Lipton, D.L. and Stimmel, B., eds.), Hawthorn Press, New York, pp. 89-98. Watters, J.K., Downing, M., Case, P., Lorvick, J., Cheng, Y. and Fergusson, B. (1990) AIDS prevention for intravenous drug users in the community: street-based education and risk reduction. Am. J. Community Psychol. 18, 587-596. Weatherby, N.L., Needle, R., Cesari, C. et al. (1994) Validity of self-reported drug use among injection drug users and crack cocaine users recruited through street outreach. J. Ed. Prog. Plan. 17, 347.-355. Wiebel, W., Ouellet, L., Guydan, C. and Samairat, N. (1990) Cocaine injection as a predictor of HIV risk behaviours. VI International Conference on AIDS, San Francisco. Woody, G., O’Hare, K., Mintz, J. and O’Brien, C. (1975) Rapid treatment intake: a method for increasing retention rate of heroin addicts seeking methadone treatment. Comprehensive Psychiatry 16, 165-169. Woody, G.E., Metzger, D.S., McClelland, A.T. et al. (1992) Two year incidence rates for HIV and HTLV-I/II among IDUs in- and out-of-methadone treatment. 54th Annual Scientific Meeting College on Problems of Drug Dependence, Keystone, Colorado. Yancovitz, S., Des Jarlais, D., Peyser, N. et al. (1991) A randomized trial of an interim methadone clinic. Am. J. Public Health 81, 118551191.