Effectiveness of Naltrexone in a CommunityTreatment Program

Therese K. Killeen, Kathleen T. Brady, Paul B. Gold, Kit N. Simpson, Richard A. Faldowski,Clare Tyson, and Raymond F. Anton

Objective: In several large, well-designed, randomized, double-blind studies, the opiate antagonistnaltrexone demonstrated efficacy in the treatment of alcohol dependence. Specifically, when combinedwith certain psychosocial therapies, naltrexone reduces the number of drinking days, heavy drinking, andtime to relapse to alcohol use in alcohol-dependent individuals. Whether this efficacy can be generalized toindividuals who have alcohol use disorders and present for treatment at front-line community treatmentprograms has not been well established.

Methods: A total of 145 patients who presented for treatment at a rural community substance abusetreatment center were randomized to receive naltrexone 50 mg daily plus usual program treatment (n � 54),placebo plus usual treatment (n � 43), or usual treatment alone (n � 48) for 12 week. A total of 133 participantshad at least one follow-up visit. Primary outcome measures included percent days drinking, average drinks perdrinking day, average drinks per day, heavy drinking days (four or more for women and six or more for men),and time to first heavy drinking day. Secondary measures included changes in serum biological markers (alkalinephosphatase, alanine transaminase, aspartate transaminase, and �-glutamyltransferase), craving, and psy-chosocial functioning.

Results: In the intention-to-treat analysis, there were no between-group differences for any of theprimary drinking outcomes at 12 weeks. In post hoc exploratory analyses, the entire sample of participantswas divided into two new groups: (1) people who drank during the 2 weeks before the start of medication(entry drinkers) and (2) people who did not drink during this interval (entry abstainers). Entry abstainerswere at an advantage at study entry in that they were significantly more likely to have an inpatienthospitalization immediately before entry into outpatient treatment. Mixed-model analysis of variancerevealed a main effect for entry group at the 12-week treatment endpoint on the primary outcome measuresof percent days drinking, average drinks per drinking day, average drinks per day, heavy drinking days, andtime to first heavy drinking day. Participants in any of the randomized groups who were entry abstainers hadsignificantly better improvement on all of the primary outcome measures. The abstainer groups that wererandomized to placebo and usual treatment had significantly better outcomes than the entry drinkers inthose perspective groups. However, for the naltrexone-treated group, entry drinkers and entry abstainershad similar improvement in drinking-related outcomes.

Conclusions: These data suggest that naltrexone may offer particular benefit to patients who continueto drink during the early stages of the trial as compared with those who have achieved abstinence beforetreatment entry.

Key Words: Naltrexone, Community Treatment, Alcoholism, Treatment Outcome.

NALTREXONE, AN OPIATE antagonist, was ap-proved by the Food and Drug Administration for the

treatment of alcoholism in 1994. Specifically, in severallarge, well-controlled, double-blind studies, naltrexone wasshown to decrease percent of days drinking (PDD) anddrinks per drinking day (DDD), as well as delay relapse to

heavy drinking (Anton et al., 1999; O’Malley et al., 1992;Volpicelli et al., 1992). Furthermore, naltrexone may havea role in preventing progression to heavy drinking in indi-viduals who sample alcohol (Anton et al., 1999; Guardia etal., 2002; Heinala et al., 2001; Volpicelli et al., 1992). Otherstudies have demonstrated that the effects of naltrexoneare most robust in participants who adhere closely to themedication regimen (Chick et al., 2000; Oslin et al., 2002;Volpicelli et al., 1997).

The psychosocial therapy used in conjunction with nal-trexone has an impact on outcome. Cognitive behavioraltherapy was used effectively in conjunction with naltrexonein studies done by Anton et al. (1999), Balldin et al. (2003),Heinala et al. (2001), Kiefer et al. (2003), and O’Malley etal. (2003). O’Malley et al. (1992) found that the combina-

From the Department of Psychiatry and Behavioral Sciences, MedicalUniversity of South Carolina, Charleston South Carolina.

Received for publication March 24, 2004; accepted September 23, 2004.This research was supported by a NIAAA funded grant (R01 AA11747)

awarded to KTB.Reprint requests: Therese Killeen, PhD, APRN, Department of Psychiatry and

Behavioral Science, Medical University of South Carolina, 67 President Street,Charleston, SC 29425; Fax: 843-792-3982; e-mail: killeent@musc.edu.

Copyright © 2004 by the Research Society on Alcoholism.

DOI: 10.1097/01.ALC.0000145688.30448.2C

0145-6008/04/2811-1710$03.00/0ALCOHOLISM: CLINICAL AND EXPERIMENTAL RESEARCH

Vol. 28, No. 11November 2004

1710 Alcohol Clin Exp Res, Vol 28, No 11, 2004: pp 1710–1717

tion of supportive therapy and naltrexone was superior tonaltrexone and coping skills in producing abstinence over a12-week treatment period. However, in participants whosampled alcohol, naltrexone and coping skills were better inpreventing relapse to heavy drinking. In many of the posi-tive studies, manualized therapies were used by masters- ordoctoral-level therapists who received intensive therapytraining in an academic medical center (Anton et al., 1999;Balldin et al., 2003; Heinala et al., 2001; O’Malley et al.,1992, 2003). Studies that used less intensive or less specifiedtreatment had mixed results with regard to drinking out-comes (Chick et al., 2000; Gastpar et al., 2002; Guardia etal., 2002).

In a study by Krystal et al. (2001), 627 veterans wererandomized to naltrexone 50 mg daily for 12 months, nal-trexone 50 mg daily for 3 months followed by 9 months ofplacebo daily, or placebo daily for 12 months. All threegroups received 12-step facilitation therapy throughout thestudy. There were no differences in drinking outcomesbetween groups. In one commentary on the study, it wassuggested that the lack of efficacy of naltrexone in thisstudy might be related to the abstinence-based approach ofthe adjunctive 12-step therapy used (Sinclair et al., 2002). Itwas hypothesized that naltrexone may be effective only inprograms that focus on drink reduction and preventing slipsfrom becoming relapses. The notion of the moderatingimpact of adjunctive psychosocial treatment in determiningthe efficacy of naltrexone treatment is important becausemost treatment for alcoholism is delivered in communitytreatment programs that use a wide variety of psychother-apeutic treatment approaches, including strong emphasison abstinence and 12-step therapy.

More recently, a naltrexone depot was developed toaddress the problem of nonadherence to oral medications.In a large, double-blind, placebo-controlled study, 624 maleand female participants were randomized to receivemonthly injections of vivitrex 380 mg (naltrexone in inject-able suspension), vivitrex 190 mg, or placebo injections fora total of 6 months. More than 60% of participants receivedall six injections. Heavy drinking was significantly dimin-ished in the vivitrex 380 mg group versus the placebo group.This effect was significant in men but not in women. Men inthe vivitrex 190 mg group also had a significant reduction inheavy drinking compared with the placebo group (Garbuttet al., 2004).

A major limitation of previous naltrexone studies is thegeneralizability of findings to the typical alcohol-abusingpatient commonly seen in community treatment programs.Interpretation of these studies is limited by the stringentinclusion and exclusion criteria used for enrollment. Thislimitation is compounded by the suggestion that naltrexonemay be effective only in combination with specific psycho-social therapies. Effectiveness studies have the advantageof exploring medication usage in a more naturalistic setting.In the real-world treatment setting, patients with alcoholuse disorders present into treatment with many problems,

including psychiatric and medical comorbidity. Most stud-ies required at least 5 days of abstinence and psychosocialstability, as well as exclusion of other substances of abuse,psychotropic medications, and legal involvement (Anton etal., 1999; Chick et al., 2000; Guardia et al., 2002; Heinala etal., 2001; Krystal et al., 2001; O’Malley et al., 1992; Volpi-celli et al., 1992). In addition, some studies discontinuedmedication in patients who relapsed (Guardia et al., 2002;Kiefer et al., 2003). Exploring subgroups of patients whorespond to naltrexone may help to explain the conflictingresults seen in previous studies.

In this study, the safety and effectiveness of naltrexonetreatment combined with usual treatment was compared tothe usual treatment with placebo and usual treatment alonein a front-line community treatment program. To enhancethe generalizability of the findings, we applied minimalexclusion criteria.

MATERIALS AND METHODS

The study was submitted and approved by a convened full InstitutionalReview Board. Individuals who entered treatment at an outpatient com-munity treatment program for an alcohol use disorder were recruited forthis study. Participants were either approached by the research assistant atinitial intake or referred by their counselor within the first 2 weeks oftreatment. The research assistant explained the study to patients whoentered treatment for alcohol use disorders. When an individual wasinterested in participation, informed consent was obtained. A brief inclu-sion/exclusion questionnaire was used to determine initial study eligibility.Because this was an effectiveness study in a naturalistic clinic setting, onlyexclusion criteria that were essential for patient safety and study validitywere applied. Inclusion criterion was a current alcohol use disorder withreported drinking in the 30 days before study entry. Exclusion criteria were(1) current addiction to opiates; (2) women who were pregnant, breast-feeding, or of childbearing age and not practicing effective birth control;(3) serious medical conditions or liver enzymes more than three times thenormal range; (4) cognitive dysfunction to an extent that impaired under-standing of informed consent or assessments; and (5) having had �10 daysof outpatient treatment in the past 3 months.

Participants were evaluated by a psychiatrist or a psychiatric clinicalnurse specialist for physical examination and psychiatric diagnoses. Sub-stance use disorders and co-occurring psychiatric diagnoses were madeusing DSM-IV criteria (American Psychiatric Association, 1994).

Participants were randomized into one of three conditions: naltrexone50 mg daily for 12 weeks plus treatment as usual, placebo daily for 12weeks plus treatment as usual, and treatment as usual alone. The averagetime that elapsed between enrollment and start date was 6.61 � 6.6 days.For participants in the usual treatment alone group, study start dates weredetermined by yoking them to the number of days between consent dateand medication start for the previous participant in the medication treat-ment group.

Urn randomization was used to balance certain variables across groups(Stout et al., 1994; Wei and Lachin, 1988). Urn variables were gender,having one or more other Axis I psychiatric diagnoses, having one or moreother substance use disorders, alcohol dependence severity as determinedby a score of 22 or higher on the Alcohol Dependence Scale (ADS),court-ordered treatment, treatment intensity (�6 hr/week versus �6 hr/week), and being in a controlled environment for �3 days in the past 30days before study entry.

This study was conducted between February 1998 and June 2003 in apublic treatment program in a rural community ~30 miles from Charles-ton, South Carolina. The program serves ~1000 patients per year in avariety of services, including intensive outpatient, court related, preven-

NALTREXONE IN A COMMUNITY TREATMENT PROGRAM 1711

tion, aftercare, and counseling programs. Study participants were enrolledin several different treatment programs offered at the clinic. Treatmentintensity ranged from 1 to 2 hr per week of group and/or individualtherapy to intensive group programs for 3 to 4 hr, 4 days per week for ~6weeks. Therapy was eclectic and included 12-step and relapse preventionapproaches. Patients were strongly encouraged to attend AlcoholicsAnonymous self-help groups in the community. As per clinic policy,breathalyzer and random urine drug screens were done routinely, andpatients who continued to drink or use other substances while in treatmentwere often referred by their counselors to inpatient treatment.

Assessments

The Time Line Follow-Back (TLFB) self-report drinking was the pri-mary outcome measure (Sobell and Sobell, 1995). Using a calendar,individuals recall estimates of their daily drinking over a specified timeperiod. Relevant calendar events are used as probes to enhance recall.This was collected at baseline for the 90 days before study entry and at 12,24, and 48 weeks of follow-up.

The following measures were collected at baseline and 12, 24, and 48weeks. Psychosocial functioning was assessed with the Addiction SeverityIndex (ASI), a semistructured interview that collects data in seven differ-ent psychosocial areas that are adversely affected by alcohol and drugs(McLellan et al., 1992). The Obsessive Compulsive Drinking Scale(OCDS), a 14-item self-report questionnaire, was used to assess alcoholcraving. Specifically, subscale scores address the obsessive thoughts asso-ciated with craving for alcohol and compulsive drinking behavior (Antonet al., 1996). Alcohol severity was assessed using the ADS, a 25-itemself-report questionnaire (Skinner and Allen, 1982).

A symptom checklist to evaluate medication adverse effects and liverenzymes, including �-glutamyl transferase (GGT), alanine aminotransfer-ase (ALT), and aspartate aminotransferase (AST), were collected duringthe treatment study at weeks 4, 8, and 12. Carbohydrate-deficient trans-ferrin, a biological drinking marker, was collected at baseline and week 12.

Participants in the two medication groups were seen at baseline and 4,8, and 12 weeks by the physician or the clinical nurse specialist formedication management. At each study visit, 40 tablets of the medicationwere dispensed. Adherence to medication was measured using theAPREX Medication Electronic Monitoring System (MEMS), an elec-tronic system designed to monitor pill bottle openings. Service utilization,TLFB, symptom checklist, liver enzymes, vital signs, breathalyzer reading,pill count, and MEMS cap readings for the two medication groups wereobtained at the monthly visits. Participants were not given any feedbackabout adherence information obtained from the MEMS cap reading butwere encouraged to take their medications. Participants in the usualtreatment alone were interviewed for service utilization and self-reportdrinking (TLFB) at 4 and 8 weeks either by telephone or in person if theywere attending their treatment program. At 12 weeks, all three groupswere seen in person for a comprehensive follow-up assessment that in-cluded TLFB, ASI, OCDS, ADS, and biological markers. Follow-up visitswere also conducted at 24 and 48 weeks after study entry for all studyparticipants. These data will be reported in a later publication.

Analysis

Baseline group differences were examined with the use of analysis ofvariance for continuous data and �2 tests for dichotomous data for theintention-to-treat analysis. Primary and secondary outcome variables wereanalyzed using analysis of covariance with baseline measures as covariates.Primary dependent measures included ADD, PDD, DDD, and heavydrinking days (HDD) as measured by the TLFB. Heavy drinking daysconsisted of four or more drinks per day for women and six or more drinksper day for men. Secondary outcome measures included biological mark-ers as measured by GGT, ALT, and AST; craving as measured by theOCDS; and psychosocial functioning as measured by the ASI. Kaplan-Meier survival analysis was performed on time to first drink and time tofirst HDD.

In conducting the initial data analysis, it was noted that a cluster ofparticipants were drinking at the time of entry into the study, whereas theother participants reported no alcohol consumption after signing informedconsent. Because this phenomenon of entry drinking differences mighthave an impact on outcome, in a post hoc analysis, participants weredivided into two groups called entry drinkers and entry abstainers. Entrydrinkers were participants who reported any drinking in the 2 weeksbefore medication initiation. Entry abstainers were participants who wereabstinent during this same interval. Treatment outcome data analysis wasperformed on the three randomized groups categorized into entry drink-ers and entry abstainers. These six groups were naltrexone entry drinkers,naltrexone entry abstainers, placebo entry drinkers, placebo entry abstain-ers, usual treatment entry drinkers, and usual treatment entry abstainers.

RESULTS

Of the 191 patients screened, 145 were enrolled andrandomized into one of the three groups. A total of 133participants had at least one follow-up visit. Baselinedemographic characteristics of the randomized groupsare displayed in Table 1. No between-group baselinedifferences were detected. In general, participants mostlywere male (63%), were white (76%), were an average of 37.3(�8.75) years of age, and had an average of 11.7 (�2.2) yearsof education. There were no demographic differences be-tween participants who were randomized and participantswho were screening dropouts. The gender and ethnicity of thissample are representative of the county clinic population(62% male, 72% white). Sixty-six percent of the state clinicpopulation is male and 62% is white (Department of Alcoholand Other Drug Abuse Services, 2003). Forty-nine percent ofthe state population is male and 70% is white. Blacks consti-tute the majority of the minority population in the state. Thenumber of blacks who reside in rural counties across the statevaries from as little as 7 to as much as 70% (US CensusBureau; quickfacts.census.gov/qfd/ststes/45000.html).

Sixty-eight (51%) participants had a comorbid psychiat-ric disorder, and 47 (35%) had a substance use disorder inaddition to alcohol abuse/dependence. The majority (88%)of participants were enrolled in an intensive outpatient (4days/week) program. The only difference between groupswas employment status. There was a trend (p � 0.06) formore participants in the naltrexone group to be unem-ployed than in the other two groups.

Medication adherence rates are displayed in Table 2.Adherence to medication was assessed by the percentageof MEMS cap openings, which represents the percentageof days that participants took their medication. Therewere no significant adherence differences between thenaltrexone and placebo groups. The mean percentage ofdays that participants on naltrexone were adherent was0.51 (�0.35) versus 0.61 (�0.31) for placebo. Sixty-sevenpercent of participants who stopped medicationsdropped out of clinic treatment. There were no differ-ences in reasons for termination of medication betweenthe groups.

Follow-up for the entire group was 79% at 4 weeks, 74%at 8 weeks, and 72% at 12 weeks. Participants in the

1712 KILLEEN ET AL.

placebo group had a slightly lower follow-up rate (67%)than the treatment as usual (75%) and the naltrexone(76%) groups at 12 weeks. Those with 12-week follow-upwere not significantly different from dropouts with regardto demographics or baseline drinking variables. Dropoutswere distributed equally across the three randomizedgroups. However, there was a trend for dropouts to beactively drinking at the time of study entry (56% for drop-outs vs. 38% for those who completed 12-week follow-up;�2 � 3.98; df � 1; p � 0.06).

Table 3 displays adverse events experienced by partic-ipants in both medication groups during the 12-weektreatment period. In general, naltrexone was well toler-ated despite that many participants were actively drink-ing and/or using other substances of abuse, experiencingother comorbid disorders, and/or taking other psycho-tropic medications. Significantly more participants in thenaltrexone group experienced daytime sleepiness,

fatigue, and dizziness as compared with the placebogroup (p � 0.05). There was also a trend for moreparticipants in the naltrexone group to experience irri-tability and nausea as compared with the placebo group.There were no significant differences in liver enzymevalues (GGT, AST, and ALT) at 12 weeks between thetwo medication groups. All three groups had lower meanGGT (67.9 � 74.8 at baseline vs. 48.4 � 70.1 at 12weeks), ALT (41.4 � 35.7 at baseline vs. 32.2 � 18 at 12weeks), and AST (33.4 � 20.4 at baseline vs. 30.8 � 15.2at 12 weeks) at 12 weeks compared with baseline (p �0.001).

Table 1. Baseline Characteristics by Treatment Group

Total(n � 133)

Naltrexone(n � 51)

Placebo(n � 36)

Usualtreatment(n � 46)

Statisticp value

Age 37.3 � 8.75 37.9 � 7.8 35.5 � 7.8 38.0 � 10.3 F � 1.41p � 0.25

GenderMale 84 (63%) 30 (58%) 20 (56%) 34 (74%) �2 � 3.59Female 49 (37%) 21 (41%) 16 (44%) 12 (26%) p � 0.17

EthnicityWhite 101 (76%) 39 (76%) 29 (81%) 33 (72%) �2 � 2.12Black 29 (22%) 11 (22%) 7 (19%) 11 (23%) p � 0.16Other 3 (2%) 1 (2%) 2 (4%)

Marital statusMarried 37 (28%) 20 (39%) 13 (29%) 11 (23%) �2 � 7.21Separated/divorced 55 (41%) 21 (41%) 14 (31%) 22 (47%) p � 0.51Single/widow 41 (31%) 9 (18%) 16 (36%) 13 (28%)

Education 11.7 � 2.2 11.66 � 1.9 12.0 � 2.9 11.9 � 2.0 F � 0.218p � 0.8

Employed 73 (55%) 22 (43%) 20 (56%) 31 (67%) �2 � 5.75p � 0.06

Other Axis I 68 (51%) 27 (53%) 18 (50%) 23 (50%) �2 � 0.11p � 0.95

Other substance use disorder 47 (35%) 18 (35%) 14 (39%) 15 (33%) �2 � 0.38p � 0.84

ADS score �22 33 (25%) 12 (24%) 6 (17%) 15 (33%) �2 � 2.82p � 0.24

Court ordered 77 (58%) 29 (57%) 24 (67%) 24 (52%) �2 � 1.77p � 0.41

Controlled environment �3 days in past month 102 (77%) 11 (22%) 10 (22%) 10 (28%) �2 � 0.55p � 0.76

Previous treatments 70 (53%) 24 (47%) 24 (67%) 22 (48%) �2 � 5.53p � 0.24

Treatment intensity �6 hr per week 117 (88%) 45 (88%) 32 (89%) 40 (87%) �2 � 0.08p � 0.96

Table 2. Medication Adherence Between Groups

Naltrexone(n � 51)

Placebo(n � 36)

Statisticp value

Stopped medication 24 (47%) 12 (33%) �2 � 1.639p � 0.27

Stopped because of sideeffects

9 (38%) 4 (25%) �2 � 0.563p � 0.71

Mean % days taking 51%�0.35 61%�0.31 F � 2.639medication p � 0.17

Table 3. Adverse Events

SymptomNaltrexone

(n)Placebo

(n) Statistic

Trouble falling asleep 5 3 NSTrouble staying asleep 4 4 NSDaytime sleepiness 10 2 0.039Nervousness 3 1 NSIrritability 9 2 NS (0.064)Depressed 2 0 NSInability to relax 2 2 NSAbdominal pain 4 2 NSNausea and/or vomiting 10 3 NS (0.098)Decreased appetite 4 1 NSFatigue 5 0 0.041Headache 8 7 NSDizziness 5 0 0.041Itching 1 2 NS

NS � p � 0.05.

NALTREXONE IN A COMMUNITY TREATMENT PROGRAM 1713

Primary Outcome

TLFB data for baseline and 12-week treatment periodfor all three groups are displayed in Fig. 1. For all threegroups, there was a significant decrease in alcohol con-sumption during the 12-week period as compared withalcohol consumption in the 90 days before informed con-sent (baseline). There were no significant between-groupdifferences in ADD (F � 2.02; df � 2; p � 0.14), DDD(F � 1.59; df � 2; p � 0.21), PDD (F � 1.15; df � 2; p� 0.32), and HDD (F � 1.48; df � 2; p � 0.23) duringthe 12-week study period. There were no between-groupdifferences in time to first heavy drinking day (�2 � 2.36;p � 0.3) for those who had 12-week data available or upuntil the time of dropout. Forty-eight percent of participantsin the usual treatment group, 41% of participants in thenaltrexone group, and 33% of participants in the placebogroup had at least one episode of heavy drinking during the12-week treatment period (�2 � 1.17; p � 0.28).

Craving, as measured by the OCDS, was significantlyreduced at 12 weeks for all groups, but there were nobetween-group differences. Similarly, there was significantimprovement for all groups in psychosocial function asmeasured by the ASI but no between-group differences.

One of the difficulties in detecting between-group differ-ences lies in the dramatic decrease in drinking in all groupsduring the 12-week study period. A large subgroup ofindividuals stopped drinking after study entry during theassessment period before medications were initiated, andwe were interested in further exploring and characterizingthis group. For a post hoc exploratory analyses, the entiresample of participants was divided into two new groups: (1)

people who drank during the 2-week period after signingconsent before the start of medication (entry drinkers) and(2) people who did not drink during this interval (entryabstainers). Thus, each treatment group was subdividedinto two groups on the basis of entry drinking, resulting insix groups. Baseline characteristics of participants in thesegroups are displayed in Table 4. There were no between-group differences in age, gender, ethnicity, marital status,employment, education, comorbid psychiatric disorder,previous treatment, and treatment intensity. There was,however, a significantly larger number of individuals whohad been in a controlled environment for �3 days in themonth before study entry in the entry abstainer group,indicating that a higher number of these individuals hadbeen in inpatient alcohol treatment. Entry drinkers werealso more likely to be enrolled in less intense treatment asusual.

An analysis of the TLFB data then was performed takinginto account entry drinking status as well as randomizedgroups. There were no significant differences between ran-domized groups in the proportion of participants who wereentry drinkers and entry abstainers (�2 � 1.54; df � 2; p �0.5). Analysis of covariance revealed a main effect for entrygroup and for interaction between entry group and treat-ment groups as randomized for the 12-week treatmentendpoint on PDD (F � 5.66; df � 5; p � 0.001), ADD (F� 3.69; df � 5; p � 0.004), HDD (F � 3.76; df � 5; p �0.003), and DDD (F � 2.62; df � 5; p � 0.027). As mightbe expected, the entry drinker group had higher PDD,DDD, HDD, and DDD during the 12-week treatmentperiod as compared with entry abstainers in all three treat-

Fig. 1. Drinking outcomes for randomizedgroups. TAU, treatment as usual; PLB, pla-cebo; NTX, naltrexone. *p � 0.001 withintreatment differences from baseline to 12weeks.

1714 KILLEEN ET AL.

ment groups. These differences were significant for thetreatment as usual and placebo groups. For the naltrexonegroup, however, there were not significant differences be-tween the entry drinkers and entry abstainers in drinkingoutcomes during the 12-week study. Entry drinkers in thenaltrexone group had significantly lower ADD (p � 0.01),PDD (p � 0.01), and HDD (p � 0.006) than entry drinkersin the treatment as usual group. However, drinking out-comes in the naltrexone entry drinkers were not signifi-cantly different from the placebo entry drinkers. As can beseen in Fig. 2, for the naltrexone-treated group, the im-

provement in drinking-related outcomes in the entrydrinker group was similar to that seen in the abstainergroup. In the placebo and usual treatment groups, entryabstainers had significantly better drinking outcomes thanthe entry drinkers.

Survival analysis curves for time to first HDD for therandomized treatment groups divided into entry abstainerand entry drinkers are presented in Fig. 3. For the placeboand naltrexone groups, there were no differences betweentime to first HDD for entry drinkers and entry abstainers(p � 0.12, p � 0.2; log-rank �2, respectively). For the

Table 4. Baseline Characteristics by Entry Abstainer and Entry Drinker Groups

Entryabstainers(n � 81)

Entrydrinkers(n � 52) Statistic p value

Age 38.1 � 8.25 35.9 � 9.38 t � 1.44 0.15Gender

Male 48 (59%) 36 (69%)Female 33 (41%) 16 (31%) �2 � 1.353 0.27

EthnicityWhite 65 (80%) 36 (69%)Black 15 (19%) 14 (27%)Other 1 (1%) 2 (4%) �2 � 2.490 0.29

Marital statusMarried 23 (28%) 14 (27%)Separated/divorced 36 (44%) 19 (37%)Single 19 (23%) 19 (37%) �2 � 3.88 0.42

Education 11.6 � 2.4 12.2 � 2.0 F � 2.30 0.13Employed 45 (55%) 34 (56%) �2 � 0.01 1.0Other axis I disorder 45 (56%) 23 (44%) �2 � 1.62 0.22Other substance use 30 (37%) 17 (33%) �2 � 0.262 0.71ADS score �22 22 (27%) 11 (21%) �2 � 0.612 0.5Court ordered 43 (53%) 34 (65%) �2 � 1.970 0.21Controlled environment �3 days in past month 25 (31%) 6 (12%) �2 � 6.617 0.01*Previous treatment 43 (53%) 27 (52%) �2 � 0.110 0.94Treatment intensity �6 hr per week 6 (7%) 10 (19%) �2 � 4.183 0.06

Data are n � SD or %.

Fig. 2. Drinking outcomes at 12 weeksadjusted for baseline in randomized treatmentgroups and entry status. ED, entry drinkers;EA, entry abstainers.

NALTREXONE IN A COMMUNITY TREATMENT PROGRAM 1715

usual treatment group, there was a significantly longertime to first HDD in the entry abstainers compared withthe entry drinkers ( p � 0.0002; log-rank �2). There wereno differences between the drinking entry groups on anyof the craving, psychosocial functioning, or biologicalmeasures.

DISCUSSION

This study was designed to explore the use of naltrexonein a group of alcohol-dependent individuals who presentedfor treatment at front-line community treatment programsusing minimal exclusion/inclusion criteria and not control-ling the adjunctive psychosocial treatment. As such, thisstudy is an effectiveness trial designed to build on thedemonstrated efficacy of naltrexone in well-controlled clin-ical trials. Typically, individuals who present for treatmentat community treatment programs have numerous psycho-social, medical, and psychiatric comorbidities that oftenmake them ineligible for efficacy studies. The philosophy,intensity, and content of therapy offered in the treatmentprogram can affect outcomes. It has been shown that the

efficacy of naltrexone varies depending on the adjunctivepsychosocial therapy being delivered. Previous studies haveshown lower relapse rates when cognitive behavioral ther-apy or relapse prevention is used in conjunction with nal-trexone (Anton et al., 1999; Heinala et al., 2001; Krystal etal., 2001). One study found higher rates of continuousabstinence with naltrexone plus supportive therapy(O’Malley et al., 1996). O’Malley et al. (2003) found com-parable improved drinking outcomes using primary caremanagement plus naltrexone as with cognitive behavioraltherapy plus naltrexone. In a large study that used 12-stepfacilitation therapy aimed at abstinence with naltrexone orplacebo, there were no differences in drinking outcomesbetween groups at 13 and 52 weeks (Krystal et al., 2001).

In the present study, 85% of the participants were re-ferred to intensive treatment services (3–4 hr, 4 days/week).Treatment was eclectic and abstinence based. Patients werestrongly encouraged to attend community 12-step meet-ings. As in other studies dealing with heterogeneous and“real life” treatment settings (Chick et al., 2000; Krystal etal., 2001), we found no differences on drinking outcomemeasures among the naltrexone, placebo, and usual treat-ment groups. As with these studies, we had a substantialattrition rate, marginal medication adherence, and a treat-ment program that was abstinence based. As in other ef-fectiveness trials, patients with psychosocial instability,polysubstance use, and psychiatric comorbidity are lesslikely to adhere to medications. The use of specific adher-ence enhancement procedures or the use of depot (inject-able) naltrexone might circumvent those barriers.

In the treatment setting for this study, participants whocame to treatment and reported drinking/drug use or testedalcohol/drug positive on biological indices were often re-ferred by their counselors to inpatient treatment. This clinicpolicy is likely to have contributed to the high study attri-tion. Of interest in the present study, participants who weretaking naltrexone and reported drinking just before theinitiation of study medications did as well on drinkingoutcomes as participants who were abstinent immediatelybefore initiation of study medications. In the treatment asusual and placebo groups, the entry drinkers had substan-tially poorer outcomes as compared with entry abstainers.This suggests that naltrexone may have preferential efficacyon the subgroup of alcohol-dependent individuals who areactively drinking while in treatment. This finding is consis-tent with the findings by Heinala et al. (2001), who foundthat naltrexone plus coping skills therapy was beneficial inpatients who were actively drinking. They proposed thatnaltrexone might react by extinguishing reinforced drinkingbehavior, making it essential that an individual continuewith some drinking after naltrexone initiation for the med-ication to exert its therapeutic effect. Anton et al. (1999)found that participants in the naltrexone group had pro-longed time to relapse after sampling alcohol than partic-ipants in the placebo group. This outcome has clinicalsignificance in that it prevents a slip from becoming a

Fig. 3. Meyer-Kaplan survival curves for days to first heavy drinking in ran-domized groups and entry status.

1716 KILLEEN ET AL.

full-blown relapse. Taken together, these studies offersome explanation for the conflicting results reported in theliterature concerning the effectiveness of naltrexone in al-cohol use disorders. Importantly, these studies also suggestthat there may be a subgroup of alcoholics who are morelikely to respond to naltrexone. Subtyping of alcohol-dependent individuals such that specifically tailored treat-ment can be delivered is an area that has received muchattention and could improve the cost-effectiveness of treat-ment enormously.

One of the limitations of this study is related to overallstudy and subgroup sample size. The study was powered todetect a small to moderate effect size (0.4 on Cohen’srange) for the psychosocial variables based on previousstudies at our institution. However, the heterogeneity inour population resulted in large SDs in most of our mea-sures. Furthermore, the large impacts of both usual treat-ment alone and usual treatment with placebo therapyeroded the number of subjects in the active drug group whocould benefit from the effect of naltrexone. For example, atan � level of 0.05 with a two-sided test, we had only 51%power to detect a decrease of 1.4 on ADD at 12 weeks asa result of naltrexone once the effect of usual treatment wasaccounted for. The only reason that we found a statisticallysignificant difference for the subgroup of clients who drankat study entry and were randomized to naltrexone was thatthe incremental effect size as a result of naltrexone for thisgroup was very large (1.1 on Cohen’s range).

The present study has important implications for the useof naltrexone in community treatment programs. Naltrex-one that is given to patients who are having difficultyachieving abstinence at the onset of outpatient treatmentmay prevent such patients from being prematurely referredto expensive inpatient services.

REFERENCES

American Psychiatric Association (1994) Diagnostic and Statistical Manualof Mental Disorders. 4th ed. American Psychiatric Association, Wash-ington, DC.

Anton RF, Moak DH, Latham PK (1996) The obsessive-compulsive as-pects of craving: A new method of assessing outcome in alcoholismtreatment studies. Arch Gen Psychiatry 53:225–231.

Anton RF, Moak DH, Waid LR, Latham PK, Malcolm RJ, Dias JK (1999)Naltrexone and cognitive behavioral therapy for the treatment of out-patient alcoholics: Results of a placebo-controlled trial. Am J Psychiatry156:1758–1764.

Balldin J, Berglund M, Borg S, Mansson M, Bendtsen P, Franck J,Gustafsson L, Hallidin J, Nilsson L-H, Stolt G, Willander A (2003) A6-month controlled naltrexone study: Combined effect with cognitivebehavioral therapy in outpatient treatment of alcohol dependence.Alcohol Clin Exp Res 27:1142–1149.

Chick J, Anton R, Checinski K, Croop R, Drummond DC, Farmer R,Labriola D, Marshall J, Moncrieff J, Morgan MY, Peters T, Ritson B(2000) A multicentre, randomized, double-blind, placebo-controlled

trial of naltrexone in the treatment of alcohol dependence or abuse.Alcohol 35:587–593.

Department of Alcohol and Other Drugs Abuse Services (2003) Totalclient intakes for 2003: country � race � gender. Substance AbuseAgency Management System. Report, Columbia, SC.

Garbutt, JC, Kranzler H, Gastfriend D, O’Malley S, Pettinati H, Silver-man B, Loewy J Ehrich E (2004) A phase III study of vivitrex (naltrex-one in injectable suspension) in alcohol-dependent adults: Efficacy andtolerability (abstract). Alcohol Clin Exp Res 28:148A.

Gastpar M, Bonnet U, Boning J, Mann K, Schmidt LG, Soyka M, Weter-ling T, Keilstein V, Labriola D, Croop R (2002) Lack of efficacy ofnaltrexone in the prevention of alcohol relapse: Results from a Germanmulticenter study. J Clin Psychopharmacol 22:592–598.

Guardia J, Caso C, Arias F, Gual A, Sanahuja J, Ramirez M, Mengual I,Gonzalvo B, Seguro L, Trujols J, Casas M (2002) A double-blind,placebo-controlled study of naltrexone in the treatment of alcohol-dependence disorder: results from a multicenter clinical trial. AlcoholClin Exp Res 26:1381–1387.

Heinala P, Alho H, Kiianmaa K, Lonnqvist J, Kuoppasalmi K, Sinclair JD(2001) Targeted use of naltrexone without prior detoxification in thetreatment of alcohol dependence: A factorial double-blind, placebo-controlled trial. J Clin Psychopharmacol 21:287–292.

Kiefer F, Jahn H, Tarnaske T, Helwig H, Briken P, Holzbach R, Kampf P,Stracke R, Baehr M, Naber D, Widemann K (2003) Comparing andcombining naltrexone and acamprosate in relapse prevention of alco-holism: A double-blind, placebo-controlled study. Arch Gen Psychiatry60:92–99.

Krystal JH, Cramer JA, Krol WF, Kirk GF, Rosenheck RA (2001) Nal-trexone in the treatment of alcohol dependence. N Engl J Med 345:1734–1739.

McLellan AT, Kushner H, Metzger D, Peters R, Grissom G, Pettinatti H,Argeriou M (1992) The Fifth Edition of the Addiction Severity Index.J Subst Abuse Treat 9:199–213.

O’Malley SS, Jaffe AJ, Chang G, Rode S, Schottenfeld RS, Meyer RE,Rounsaville B (1996) Six month follow-up of naltrexone and psycho-therapy for alcohol dependence. Arch Gen Psychiatry 53:217–224.

O’Malley SS, Jaffe AJ, Chang G, Schottenfeld RS, Meyer RE, RounsavilleB (1992) Naltrexone and coping skills therapy for alcohol dependence.A controlled study. Arch Gen Psychiatry 49:881–887.

O’Malley SS, Rounsaville BJ, Farren C, Namkoong K, Wu R, Robinson J,O’Conner PG (2003) Initial and maintenance naltrexone treatment foralcohol dependence using primary care vs specialty care. Arch InternMed 163:1695–1704.

Oslin DW, Pettinati H, Volpicelli JR (2002) Alcoholism treatment adher-ence: Older age predicts better adherence and drinking outcomes. Am JGeriatr Psychiatry 10:740–747.

Sinclair J, Alho H, Shinderman M (2002) Naltrexone in the treatment ofalcohol dependence. N Engl J Med 346(17):1130.

Skinner HA, Allen BA (1982) Alcohol dependence syndrome: Measure-ment and validation. J Abnorm Psychol 91:199–209.

Sobell LC, Sobell MB (1995) Alcohol Timeline Follow Back Users’ Manual.Addiction Research Foundation, Toronto, Ontario, Canada.

Stout RL, Wirtz PW, Carbonari JP, Del Boca FK (1994) Ensuring bal-anced distribution of prognostic factors in treatment outcome research.J Stud Alcohol Suppl 12:70–75.

Volpicelli JR, Alterman AI, Hayashida M, O’Brien CP (1992) Naltrexonein the treatment of alcohol dependence. Arch Gen Psychiatry 49:876–880.

Volpicelli JR, Rhines KC, Rhines JS, Volpicelli LA, Alterman AI,O’Brien CP (1997) Naltrexone and alcohol dependence. Role of subjectcompliance. Arch Gen Psychiatry 54:737–742.

Wei L, Lachin J (1988) Properties of the urn randomization in clinicaltrials. Control Clin Trials 9:345–364.

NALTREXONE IN A COMMUNITY TREATMENT PROGRAM 1717