Preadolescent conduct disorder behaviors predict relapse and progression of addiction for adolescent alcohol and drug abusers

0145-6008/95/1906-1528$03.00/0 ALCOHOLISM: CLINICAL AND EXPERIMENTAL RESEARCH

Vol. 19, No. 6 December 1995

Preadolescent Conduct Disorder Behaviors Predict Relapse and Progression of Addiction for Adolescent

Alcohol and Drug Abusers Mark G. Myers, Sandra A. Brown, and Mariam A. Mott

Increased awareness of the heterogeneity among alcohol and drug abusers has led to examination of the influence of concomitant psy- chopathology on the course of addiction and addictive relapse. Re- search among adult alcohol and drug abusers has found a detrimen- tal influence of comorbid psychopathology on treatment outcome. Evidence exists for prevalent comorbidity, in particular conduct dis- order (CD), among clinical samples of teen alcohol and drug abusers. However, little information is currently available on the relation and influence of psychopathology on outcome after treatment for ado- lescent alcohol and drug abuse. The current study examines the relation between CD, as assessed by the extent of preadolescent CD behaviors occurring before the initiation of drug use, and outcome for 131 adolescents over 2 years after inpatient treatment for alcohol and drug abuse. Drawing on recent conceptualizations of the pro- cess of relapse and progression of addictive behavior, CD is hypoth- esized to influence the process of relapse by altering coping efforts and intentions and increasing exposure to potential relapse situa- tions. Results reveal that the extent of preadolescent CD behaviors is positively related to alcohol involvement in the 2 years after treat- ment, and that much of this relation is accounted for by posttreat- ment cognitive coping, motivation for alcohol abstinence, interper- sonal problems, and exposure to alcohol use. Contraty to expectations, preadolescent CD behaviors did not predict posttreat- ment drug use. These findings provide evidence for the influence of psychopathology on the process of relapse and progression of ad- diction after treatment for adolescent alcohol and drug abuse, and suggest that teens with early life CD behaviors may be at greater risk for continued alcohol abuse.

Key Words: Adolescent Alcohol and Drug Abuse, Comorbidity, Conduct Disorder, Relapse.

ECOGNITION OF the heterogeneity among alcohol R and drug abusers has directed increased attention to the presence and influence of concomitant psychopathol- ogy on the course of addiction. Research consistently re- veals a high co-occurrence between psychiatric disorders

From the Department of Psychiatry (M.G.M., S.A.B.), University of Cal- ifornia San Diego School of Medicine; and the Department of Psychology (S.A.B.), Veterans Affairs Medical Center, San Diego, California; and the Treatment Foster Family Services Division (M.A.M.), Father Flanagan’s Boys Town, Omaha, Nebraska.

Received for publication March 20, 1995; accepted June 7, 1995 This study was supported by the National Institute on Alcohol Abuse and

Alcoholism (Grant AA07033-06), the National Institute on Drug Abuse (Grant DA 09181-01), and the Research Service of the Department of Veterans Affairs.

Reprint requests: Mark G. Myers, Ph.D., Department of Psychology (116B), Veterans Affairs Medical Center, 3350 La Jolla Village Drive, San Diego, CA 92161.

Copyright 0 1995 by The Research Society on Alcoholism.

and substance abuse,’ high levels of psychopathology among clinical samples of adult alcoholics,2 as well as a significant relation between such symptomatology and treatment outcome for alcohol and other drug a b ~ s e . ~ ” For example, adult Antisocial Personality Disorder (ASPD) is found related to poorer outcome after treatment for alcoholism.6 However, little information currently exists regarding comorbid psychopathology in relation to treat- ment outcome for adolescent alcohol and drug abusers.

Investigations examining psychopathology among ado- lescent alcohol and drug abusers find that conduct disorder (CD) is the most common comorbid psychiatric diagnosis. For example, a recent study of 226 adolescents on an inpatient alcohol and drug abuse unit found that 54% of the teens met criteria for CD, a greater prevalence than for any other d i~orde r .~ In an examination of affective comor- bidity among 156 teens hospitalized on a dual diagnosis unit for teen alcohol and drug abusers, Bukstein et a1.8 reported that 70% of these teens met criteria for CD, 51% for affective disorders, and 14% were diagnosed with both CD and major depression. An investigation of 57 adoles- cents consecutively admitted for treatment of alcohol and drug abuse found that 42% of the patients met diagnostic criteria for CD.9 A chart review of adolescents admitted to an outpatient alcohol and drug abuse treatment program found that 60% had a nonalcohol and drug use psychiatric diagnosis, with 40% meeting criteria for CD.1° In examin- ing 43 inpatient adolescent alcohol abusers, Tarter et a1.l’ categorized the majority as demonstrating externalizing symptoms and behavioral dyscontrol. In the same sample, those teens with externalizing problems were found to demonstrate greater severity for both substance involve- ment and psychiatric disturbance than those with internal- izing symptomatology.12 Thus, investigations to date con- sistently find that CD is a prevalent comorbid diagnosis in clinical samples of alcohol and drug-abusing teens, and also suggest that this subgroup may evidence the greatest sever- ity of substance involvement and psychopathology. Yet, little is currently known regarding the influence of CD on outcome after teen alcohol and drug abuse treatment.

Little information is available exploring the relationship between CD and adolescent alcohol and drug abuse treat- ment outcome. We have previously reported findings that both a diagnosis of CD based on behaviors independent of

1528 Alcohol CIin Exp Res, Vol19, No 6, 1995: pp 1528-1536

CD BEHAVIORS AND RELAPSE 1529

drug and alcohol use and a greater number of such behav- iors assessed during treatment were significantly related to greater use of alcohol and drugs 12 and 24 months after treatment.13 We have also found that alcohol and drug- abusing teens displaying CD behaviors before drug involve- ment are at greater risk for an adult diagnosis of ASPD.14 These findings of poorer outcome for CD teens mirrors the relationship between adult ASPD, for which CD is a pre- cursor,15 and poorer treatment outcome among adult alco- h o l i c ~ . ~ , ~ Thus, examination of the relation between CD and treatment outcome is of particular relevance, because CD teens are at risk for becoming treatment-resistant ASPD adult alcohol and drug abusers.

Despite increased attention to comorbid psychopathol- ogy and alcohol and drug abuse treatment outcome, the relation between psychopathology and adolescent addic- tion progression is poorly understood. The body of research examining the process of relapse provides one starting point for exploring the influence of psychopathology on treatment outcome. Recently formulated biobehavioral models of addiction conceive of relapse as part of a process of addiction.16 Current models of addiction relapse high- light the importance of the process of coping and suggest that individuals who lack adequate coping resources to manage situational demands (i.e., stressors or temptations to use) are at greater risk for resuming alcohol and drug use.I7-l9 Available evidence from studies of adolescent re- lapse support the significant role of coping in addiction relapse. For example, our studies examining cognitive and behavioral coping strategies of adolescent alcohol and drug abusers find that utilization of strategies varies between teens with different posttreatment and that coping predicts subsequent drug and alcohol use s t a t ~ s . ~ ~ , ~ ~ In addition, appraisal of potential relapse situations, or cognitive vigilance, has been found to play a role in teen relapse such that teens who return to alcohol and/or drug use seem to underestimate the personal risk involved in tempting situations.20221 Thus, the coping process has been implicated in alcohol and drug abuse treatment outcome for adults and teens. One possible pathway whereby comor- bid psychopathology affects alcohol and drug abuse treat- ment outcome is through influence on various aspects of relapse risk.

Within the context of a psychosocial model of coping and relapse, the known correlates of CD may elevate the vul- nerability to addictive relapse and influence posttreatment clinical course by increasing risk factors and decreasing protective factors for addiction progre~sion.~~ Although clear evidence of causality in the etiology of CD is lacking, certain factors consistently emerge as correlates for delin- quent and antisocial y o ~ t h . ~ ~ , ~ ~ Among these are disturbed interpersonal relationships, including disrupted family en- v i ronmen t~~~ (e.g., familial conflict, parental addiction, and criminality) and symptoms of hyperactivity, such as impul- sivity and limited attention span.2g For example, family conflict, familial alcohol, and drug abuse and other prob-

lematic interpersonal relations may provoke stress that taxes coping abilities and increases the attraction of alcohol and drug use as a method of coping. Difficult interpersonal relationships, delinquent peers, and familial substance abuse may also influence posttreatment alcohol and drug use by providing poorer social resources and increased posttreatment exposure to substances and substance-abus- ing models.29 Furthermore, the impulsivity associated with CD may influence outcome through poor decision making (i.e., entering risky situations with little forethought), or a lack of cognitive vigilance (i.e., a failure to assess accurately the potential risk for relapse in certain situations). Finally, motivation for abstinence may be lower among CD teens for a variety of reasons, such as the salience of alcohol and/or drug reinforcement, attractiveness of alcohol and/or drug use as a means for managing personal difficulties, or peer pressure and modeling of alcohol and drug use. Lower motivation for abstinence may also result in minimization of perceived risk of less potent substances (e.g., alcohol), or lead to greater alcohol and drug use among CD teens by diminishing efforts to avoid potential substance use or use situations. In conceptualizing relapse, it is important to distinguish between isolated relapse episodes versus the progression of addiction, because the previously discussed factors will exert differential influence on each. For exam- ple, temptation-coping skills and impulsivity may be more linked with a return to alcohol and drug use after a period of abstinence (i.e., discrete episodes of use), whereas cog- nitive factors (e.g., motivation) and environmental factors, such as ongoing interpersonal conflict and substance avail- ability, might be expected to have more influence on con- tinuing alcohol and drug use (i.e., progression of addic- tion).

The diagnosis of CD can be difficult to establish in the context of adolescent developmental changes and alcohol and drug abuse. Delinquent behaviors may emerge in ad- olescence independent of substance involvement, precede and lead to adolescent alcohol and other drug use and

or follow from teen drug i n ~ o l v e m e n t . ~ ~ ’ ~ ~ Fur- thermore, several efforts to distinguish types of CD be- havior have recently been developed.3495 For example, M ~ f f i t ~ ~ has proposed a taxonomy that aims to distinguish between teens likely to persist in deviant behavior versus those who will not. According to this taxonomy, a majority of teens display deviant behaviors during adolescence but desist as they enter adulthood. A subgroup of youth evi- dence life-course-persistent antisocial behaviors character- ized by deviant behaviors that begin in early childhood and continue throughout life. Consistent with existing data,26,36 Moffit’s taxonomy suggests that teens with an earlier onset of CD behavior and who display a greater range and sever- ity of deviant behaviors are more likely to persist in these behaviors as development progresses. Because CD-type behaviors are particularly common among alcohol and drug-abusing teens (e.g., the problem behavior syn- d r ~ m e ~ ~ ) and to avoid confounding CD behaviors pro-

1530 MYERS ET AL.

voked by substance involvement, the present study will utilize the frequency of CD behaviors occurring before the onset of drug use (i.e., preadolescent CD behaviors) as an indicator of potential antisocial psychopathology.

In our conceptualization of the process of a return to addictive behavior, factors that decrease ability to cope successfully with or manage relapse risk situations and increase the attractiveness of alcohol and drug use are postulated to increase the likelihood of relapse and pro- gression of addiction. To this end, the current study exam- ines whether the relation between preadolescent CD be- haviors and alcohol and drug use outcome is in part accounted for by aspects of CD that may mediate the coping process. The process of relapse to an addictive life-style is herein exemplified by cumulative alcohol and drug use during the 2 years after treatment for adolescent alcohol and drug abuse. Cognitive, behavioral, and envi- ronmental features of CD are hypothesized to provoke a resurgence of addictive behaviors. Specifically, the vari- abies hypothesized to influence the outcome process in- clude coping strategies and appraisal in relapse risk situa- tions, motivation for abstinence, impulsivity, conflict in significant relationships, family relations and parental alco- hol and drug use, addiction within the social resource net- work, and exposure to alcohol and drug using models. We predict that a greater number of CD behaviors preceding the onset of drug use will correspond with poorer treatment outcome over the 2 years following adolescent alcohol and drug abuse treatment. It is also anticipated that more pre- adolescent CD behaviors will correspond with poorer cop- ing skills and lower relapse risk appraisal, less motivation for abstinence, greater impulsivity, poorer interpersonal and familial relationships, greater parental substance use, lower quality of social support, and greater exposure to models for alcohol and drug use. It is anticipated that the aforementioned posttreatment variables will account for a portion of the relation between preadolescent CD behav- iors and 2-year treatment outcome.

METHODS

Subjects

Subjects for the present study were drawn from 163 teens interviewed during an inpatient stay at two adolescent alcohol and drug treatment programs in metropolitan San Diego, CA. Both programs were based on the Alcoholics Anonymous 12-step treatment approach and had compa- rable lengths of stay (mean = 4-6 weeks). These facilities provided programs that include family meetings, individual and group counseling, therapeutic recreation, and school. Teen subjects were selected from consecutive admissions to the rehabilitation programs, and screened to exclude adolescents with DSM-III-R38 axis I diagnoses that predated the onset of their alcohol or drug abuse (e.g., affective disorder and schizo- phrenia), teens for whom no resource person (typically a parent) was available to provide corroborative information for preadolescent and posttreatment functioning, and those residing >50 miles from the re- search offices.

The present sample includes 144 teens who completed 2-year posttreat- ment interviews (87% of the initial sample). Of these 144, 13 subjects institutionalized for more than half of either the first year or second year

Table 1. Characteristics of Adolescent Alcohol and Drug Involvement during the Previous 3 Months Assessed at the Time of Treatment, 1 Year and 2 Years

Posttreatment [Mean (SD)] _____

Treatment 1 year 2 years Average daydmonth 11.2 (14.1) 4.8 (6.9) 6.9 (9.9)

Quantity/frequency of 58.7 (92.6) 21.4 (36.7) 38.8 (73.7) using alcohol

alcohol use (drinks/ occasion x days/ month)

using drugs

different drua Woes

Average times/month 35.4 (17.4) 9.4 (14.7) 7.3 (12.1)

Average no. of 2.3 (1 2) 1.3 (1.1) 0.6 (1.2)

posttreatment were excluded from analyses, because living in a controlled environment reduces opportunity for use of alcohol and other drugs and may influence the other variables considered herein (e.g., exposure to alcohol and drug use, extent of interpersonal problems, etc.). The exclu- sion of institutionalized teens resulted in a final sample of 131 for the present analyses. Thus, the sample examined consists of 58 females and 73 males, averaging 16.1 years of age (SD = 1.2, range = 13-18) at the time they entered treatment. These adolescents came from varied socioeco- nomic backgrounds, ranging from manual laborers to college-educated professionals (Hollingshead Socioeconomic Status index; mean = 29.2, SD = 12.4), and of whom 80% were Caucasian, 7% Hispanic, 5% African American, and 8% from other ethnic backgrounds.

Of the adolescents assessed during treatment, 48% met DSM-111-R criteria for CD only when delinquent behaviors related to their alcohol or drug involvement were included (secondary CD), whereas another 47% met the CD criteria independent of their alcohol and other drug involve- ment (primary CD). Initial alcohol and drug use characteristics for this sample are based on self-reported lifetime and current (consumption over the 3 months before admission) alcohol and drug use, as assessed with the Customary Drinking and Drug Use Record39 (CDDR). Alcohol use at the time of treatment was summed for beer, wine, and liquor, resulting in an average of 11.2 (SD = 14.1) days/month drinking alcohol and a quantity/ frequency (average drinks per occasion X average drinking days per month) of 58.7 (SD = 92.6). Drug use in the previous 3 months was summed across seven substances (marijuana, amphetamines, cocaine, opi- ates, barbiturates, hallucinogens, and inhalants) resulting in an average of 35.4 (SD = 17.4) times/rnonth using drugs and 2.3 (SD = 1.2) different types of drugs used. Note that because frequency of use was summed across multiple substances, the average days per month can exceed 30 days. See Table 1 for alcohol and drug use information at intake, 1 year, and 2 years posttreatment.

Comparison of the excluded subjects with those included in the present analyses yielded a significant difference on age [mean = 15.3 (SD = 1.4) vs. mean = 16.1 (SD = 1.2), respectively], but no differences emerged on other demographics, pretreatment alcohol and drug involvement mea- sures, nor current CD diagnosis.

Procedure

During treatment, each teen and resource person (i.e., parent) com- pleted a 90-min independent structured clinical interview39 during the second or third week of hospitalization. Teen and parent pairs were each assigned different interviewers to ensure confidentiality and promote teen self-disclosure of socially undesirable activities. Follow-up interviews were conducted at 6 months, 1 year, and 2 years postdischarge, with parents/ resource persons and adolescents again interviewed separately.

Measures

Demographics and Psychosocial Functioning at the Time of Treatment. The intake structured clinical inte~vie$~"~ assessed sociodemographic, medical, and psychiatric history, as well as drinking and drug use infor- mation for the adolescent and hisher family.

CD BEHAVIORS AND RELAPSE 1531

Pretreatment Alcohol and Drug Use. Details of alcohol and drug use were assessed using the CDDR.39 The CDDR elicits quantity and fre- quency of alcohol and drug use, substance withdrawal and DSM-111-R dependence symptoms, and life problems related to alcohol and drug use?’

Pretreatment History of CD Behaviors. Each teen and parent indepen- dently completed a second structured interview (the CD Questionnaire) focusing on the adolescent’s behavior as related to DSM-III-R3’ diagnos- tic criteria for CD and ASPD. For each reported CD and ASPD behavior, participants were asked: age of initiation, extent to which the behavior was

ment Scale45 (FES). Scores of the conflict scale of the FES were summed over the follow-up time points to provide a measure of posttreatment family conflict, with higher scores reflecting greater familial conflict. Also included was a structured follow-up interview composite measure of in- terpersonal problems (at home, at school, with peers, etc.) summed across follow-up time points.

Posttreatment Exposure and Access to Alcohol and Drugs. Social support was assessed using Sarsason’s social support followed with a Lazarus and Folkman social network evaluation modified to con- sider the alcohol and drug use pattern of each identified support.29

directly related to alcohol or drug use (e.g., during use episodes), indi- rectly related to alcohol or drug use (e.g., in an effort to obtain money to buy drugs), or occurred independent of any alcohol or drug involvement. Subsequent to both interviews, teen and parent data were compared to determine the most reliable, conservative estimate of the incidence of CD and ASPD behaviors and their level of occurrence independent of alcohol or other drug involvement.

Posttreatment Alcohol and Drug Use. Teens and resource persons were interviewed separately using a follow-up version of the CDDR to obtain

Specifically, based on the alcohol and drug use information, supports were classified into three substance involvement categories: high-quality sup- ports (no drinking or drug use), moderate quality (occasional or social drinking or drug use with no evidence of substance use related problems), and low-quality (probable or definite alcohol or drug abusers). The present analyses utilized scores reflecting the ratio of low quality to total number of supports, each summed over the follow-up interviews. Addi- tional measures of exposure included items from the structured follow-up interview assessing parental alcohol and drug use (each rated separately

information regarding teen posttreatment alcohol and other drug use. In addition, information regarding details of the initial relapse situation was collected as part of the follow-up structured clinical inter vie^.^' Standard procedures to increase reliability of alcohol and drug use measures were

for mother and father, on a 7-point scale from nondrinkerhser to problem drinkerher) and adolescent exposure to alcohol and drug use [each rated separately from 1 (minimal exposure) to 10 (maximal exposure), each summed over the follow-up interviews].

Teen alcohol and drug use information was corroborated by resource person interviews, and in the few cases of discrepant reports, other resources were utilized (e.g., medical records, additional resource persons, etc.). The majority of the sample (>85%) reported posttreatment RESULTS

alcohol and drug use during the follow-up period, with teens consistently disclosing more use than did resource persons. Previous studies4’ have established that alcohol and drug abusers can provide reliable drinking and drug use data when similar procedures are used (e.g., assurance of confidentiality, multiple sources of data, and corroborative interviews).

Analytic Strategy

hypothesized relations between CD behaviors, CD charac- teristics hvDothesized to influence relaDse. and Dosttreat-

The two-stage analysis was designed to investigate the

, I 1 ’ Teen motivation for abstinence was assessed using the structured fol-

low-up interview. Motivation for abstinence was rated separately for al- coho1 and drugs, scores ranging from 1 (definitely won’t drinwuse) to 100 (definitelv will drinkhe). and a summaw score was created from each of

merit alcohol drug use outcome. In the first stage, correlational analyses were performed to assess the rela- tions between preadolescent CD behaviors and alcohol and

the follow-up interviews. Posttreatment Coping with Relapse Situations. Coping was assessed using

the Adolescent Relapse Coping Questionnaire,21.22 an instrument de- signed to examine coping responses to a commonly reported adolescent alcohol and drug abuser relapse situation. This measure assesses coping responses in the face of a social offer to use alcohol and drugs and includes three cognitive appraisal items, each rated on a 1-10 Likert scale: per- ceived importance of the situation (not important to extremely important), difficulty of coping with the situation (not difficult to very difficult), and likelihood of successfully abstaining in the situation (definitely would use

drug use, followed by correlations between early CD be- haviors and the characteristics of CD during adolescence.

a hierarchical multiple regression was used to exam- ine the extent to which the characteristics of teen CD found

the rela- tion between early CD behaviors and adolescent addiction treatment outcome.

to preadolescent CD behaviors

to definitely would not use). The 33 coping response items comprise three factors: cognitive and behavior problem solving (coping factor l), self- critical thinking (coping factor 2), and abstinence-focused coping (coping factor 3).23 To avoid transient fluctuations in coping responses that might be produced by treatment, responses to the coping measure at 1-year posttreatment were used in the analyses, because these are roughly rep- resentative of posttreatment coping, and these scores cannot be meaning- fully summed over multiple time points.

Impulsivi@ Impulsivity was assessed using the impulsivity scale of the Jackson Personality Research Form (JPRF-E).44 This variable was only assessed at the intake interview, but was used herein, because impulsivity is considered a relatively stable trait. In addition, forethought before relapse was assessed in relation to the teen’s initial posttreatment episode of alcohol or drug use assessed on the structured clinical interview. Teens were asked to describe how long they had been thinking about drinking/ using before the relapse, with a 7-point scale ranging from “did not make a conscious decision to drink/use” to “months before the relapse.” This variable was examined only for teens who had used during the posttreat- ment period.

Posttreatment Interpersonal Conflict. Family relations were assessed using parental responses to the conflict subscale of the Family Environ-

Preadolescent CD Behaviors Preadolescent CD behaviors were calculated by summing

the number of DSM-111-R criteria reported for each subject before their initial drug use experience. This variable did not display significant skewness or kurtosis. Overall, the teens averaged 1.1 (SD = 1.2; range = 0-6) different CD criteria before drug use, with 60% reporting at least one type of CD behavior preceding drug use. Among teens reporting any preadolescent CD behaviors, the most com- monly reported behaviors were: frequent lies (79%), steal- ing without confrontation (68%), repeated truancy (35%), fights (24%), property destruction (24%), cruelty to others (14%), and cruelty to animals (13%). For these most fre- quent CD behaviors preceding drug use, the average age at which the behaviors were first exhibited range from 7.5 (SD = 2.8) years for multiple fights to 11.9 (SD = 1.9) years old for truancy.

1532 MYERS ET AL.

Posttreatment Alcohol and Drug Use Alcohol and other drug use outcome measures were

reduced by utilizing principal components analysis to pro- duce composite outcome variables. Included in the analysis were 13 alcohol and drug use variables representing inten- sity of use [cumulative 2 years quantity/frequency of alcohol use; cumulative 2 years drinking days/month; number of drugs used, 2 years cumulative and currently (past 3 months); cumulative 2 years timedmonth using drugs; use- related problems (2 years cumulative and current life prob- lems related to alcohol; cumulative problems related to drugs in the past 2 years); and withdrawal and dependency symptoms (alcohol withdrawal symptoms, in the past 2 years cumulative and currently; cumulative 2 years alcohol dependence symptoms; current drug withdrawal symptoms; past 2 years drug dependence symptoms]. Three of the original items were omitted, one because of skewness (cur- rent alcohol withdrawal symptoms) and two for complex loading on the derived factors (cumulative dayshonth us- ing alcohol past 2 years, and cumulative number of drugs used past 2 years). The principal components analysis was followed by a varimax rotation and yielded a two-factor solution. The first factor, alcohol involvement (eigenvalue = 4.91, variance accounted for = 49%), consisted of five items: current (i.e., 3 months before interview) alcohol- related problems; cumulative alcohol-related problems past 2 years; cumulative 2 years quantity/frequency of alcohol use; cumulative 2 years alcohol withdrawal symptoms; and past 2 years cumulative alcohol dependence symptoms. The second factor, drug involvement (eigenvalue = 1.37; vari- ance accounted for = 14%), included five items: past 2 years cumulative drug-related problems; cumulative times/ month using drugs in the past 2 years; current number of different drugs used; and past 2 years cumulative drug withdrawal symptoms and drug dependence symptoms.

Correlational Analyses Pearson correlations were conducted to examine the

relations between preadolescent CD behaviors and adoles- cent alcohol and drug use outcome, and adolescent char- acteristics of CD (Table 2).

First assessed were the relationships between preadoles- cent CD behaviors and posttreatment alcohol and drug involvement. The correlation between preadolescent CD behaviors and 2-year composite alcohol use (r = 0.23, p 5 0.01) was significant, but the composite posttreatment drug use variable was not correlated with the preadolescent CD measure (r = -0.05). Thus, CD behaviors preceding sub- stance involvement were related to posttreatment alcohol but not drug use.

Next examined were the relations between CD behaviors and relapse risk measures of coping, appraisal, motivation for abstinence, and impulsivity. The three coping factor scores and appraisal questions yielded only one significant correlation with CD, that for coping factor 2, self-critical

Table 2. Correlations (f j of CD Behaviors Occurring Before Initial Drug Involvement with Alcohol and Drug Use Outcome and Teen CD Characteristics

No. of different CD behaviors before initial

drug involvement

Alcohol and drug use outcome Composite P-year alcohol use Composite P-year drug use

Coping Teen CD characteristics

Coping factor 1 Coping factor 2 Coping factor 3

Importance Difficulty Self-efficacy

Motivation for abstinence from alcohol use Motivation for abstinence from drug use Relapse forethought JPRF impulsivity

Interpersonal problems Family conflict (FES)

Ratio of low quality to total supports

Mother’s alcohol use Mother’s drug use Father’s alcohol use Father’s drug use

Alcohol exposure Drug exposure

Relapse risk appraisal

Motivation and impulsivity

Interpersonal and family conflict

Social support

Parent substance use measures

Exposure to substance-abusing models

0.23’ -0.05

0.00 0.34’

-0.03

-0.1 1 0.15 0.04

0.18t 0.05

-0.01 -0.05

0.20t 0.17

0.08

0.12 -0.03

0.07 0.13

027* 0.06

* Two-tailed correlation, p 5 0.01. t Two-tailed correlation, p < 0.05.

statements (r = 0.34, p < 0.01). Teen motivation for absti- nence from alcohol use, but not drug use, was significantly related to preadolescent CD behaviors (r = 0 . 1 8 , ~ < 0.05). Neither the extent of forethought before initial relapse in the first year posttreatment (i.e., how long before this situation had you thought of using) nor the JPRF impul- sivity scale scores were significantly correlated with extent of early CD behaviors.

Interpersonal difficulties assessed utilizing the interper- sonal problems composite score were significantly related to early life CD behaviors (r = 0.20, p < 0.05). However, family conflict as measured by scores on the conflict sub- scale of the FES did not relate significantly with early CD behaviors. Average FES conflict scale scores across the follow-up interval were similar for the current sample in comparison with those reported for a normative sample of nondistressed families42 [mean = 3.17 (SD = 1.75) vs. mean = 3.18 (SD = 1.91), respectively].

In examining quality of social resources, the ratio of low-quality supports to total number of supports was not significantly related to CD. Cumulative ratings of parental alcohol and drug use were also not related to preadolescent CD behaviors. Examination of parental substance use sta- tus, averaged over the follow-up interviews shows that 32% of the fathers could be classified as heavy drinkers (i.e., reported as drinking heavily at least sporadically) versus

CD BEHAVIORS AND RELAPSE 1533

Table 3. Hierarchical Regression Analysis Predicting 2-Year Alcohol Use Composite Score from Correlates of CD and Early CD Behaviors

variance explaining alcohol use outcome. These findings are consistent with a model whereby posttreatment CD

P ssc characteristics mediate the relation between preadolescent CD behaviors and the alcohol use composite variable. Be-

Self-critical statements coping (coping factor 2) 0.09 0.08 Motivation for abstinence (from alcohol) o.15 o.12 cause preadolescent CD criteria were not correlated with Interpersonal problems 0.21 0.19' posttreatment drug use, a mediationai model was not as- Exposure to alcohol use

CD behaviors preceding drug use 0.07 0.06

Variable

Step 1

o.26 0.19' sessed for this outcome. Step 3

Note: R2 = 0.31 for step 1 (p < 0.01); AR2 = 0.004 for step 2 (NS). * p < 0.05.

DISCUSSION

The current study examined the hypothesis that one form

Hierarchical Multiple Regression The mediational model analytic procedures proposed by

Barron and K e n n e ~ ~ ~ were utilized to assess whether the relation between preadolescent CD behaviors and post- treatment alcohol and drug use was mediated by the pro- posed cognitive, behavioral, and environmental features of adolescent CD. The relation found above between pread- olescent CD behaviors and alcohol use outcome (r = 0.23), demonstrated that CD behaviors accounted for -5% of the variance in the posttreatment alcohol composite variable. A hierarchical multiple regression was then conducted to examine whether the relation between preadolescent CD behaviors and alcohol use outcome was explained by ado- lescent CD characteristics (Table 3). Along with preado- lescent CD behaviors, significantly related adolescent CD characteristics [coping factor 2 (self-critical statements), motivation for alcohol abstinence, interpersonal problems summary score, and exposure to alcohol use] were included in the regression. The first step consisted of the four CD characteristic variables entered as a block that collectively accounted for a significant proportion of variance (s?) in predicting alcohol use outcome [s? = 0.31,1;(4,95) = 10.5, p < 0.011. Number of early CD behaviors was entered as the second block and did not add significantly to the pre- diction of alcohol use outcome (increment in s? = 0.004). To assess the independent contribution of each variable to the prediction of alcohol use, we examined for each pre- dictor variable the standardized regression coefficients (p), which represent the weighted contribution of each variable independent of other predictors, and squared semipartial correlations (ssc), which represent the common variance between the predictor and independent variable after re- moving the influence of all other predictor variables in the analysis. The standardized regression coefficients (p) and ssc for the four correlates of CD variables revealed that exposure to alcohol ( p = 0.26; ssc = 0 . 1 9 ; ~ < 0.05) and interpersonal problems ( p = .21; ssc = 0.19; p < 0.05) independently accounted for significant portions of the

mise the coping process, increase risk situations, or in- crease access and exposure to alcohol and drug use. Our hypotheses were derived from adult-based addiction re- lapse mode1s,l6-l8 and were generally supported by the present data as applied to a sample of teen alcohol and drug abusers. The results reveal that a greater diversity of CD behaviors before the initiation of drug involvement corresponds with greater alcohol, but not drug, involve- ment during the 2 years after treatment. As predicted, posttreatment variables characteristic of adolescent CD account for most of the relation between early life CD behaviors and 2-year alcohol involvement outcome. These data suggest that: (1) teens displaying more early-life CD behaviors may have poorer alcohol use outcome after treat- ment for alcohol and drug abuse than other substance- abusing teens; and (2) the influence of CD on the post- treatment course of addiction occurs through domains, including both lifestyle and cognitive processes. These find- ings provide initial evidence for the influence of one form of psychopathology, CD, on the process of relapse and progression of addiction after treatment for adolescent alcohol and drug abuse.

The present data refine and extend previous findings of a relation between a diagnosis of CD based on behaviors independent of alcohol and drug use and categorical treat- ment outcome (i.e., abstainers, minor relapsers, and major relap~ers'~). There are a number of reasons for the ob- served relationship between early CD behaviors with post- treatment alcohol but not drug use. Examination of indi- cators of posttreatment alcohol use suggests a greater range and variability in alcohol use as a result of increasing alcohol involvement for the entire sample over the 2-year time period. Conversely, both frequency and diversity of drug use seem to decrease during the posttreatment inter- val. The differential patterns of alcohol versus drug use are reflected in the relative proportions of variance accounted for by the posttreatment composite factors (49% vs. 14%, respectively), which may explain the lack of observed rela- tions between early-life CD behaviors and drug use out-

1534 MYERS ET AL

come. The observed increase in alcohol use by teens in this sample may reflect the substantial normative increases in the use of alcohol during late adole~cence.~~ Furthermore, the differences in alcohol versus drug consumption might also reflect the considerable societal pressure generated against teen drug use (e.g., “just say no”) that peaked during the period of data collection for this study (late 1980s). Such efforts may further solidify cognitive distinc- tions teens make between alcohol and other drugs. Thus, the overall increase in alcohol use and reduction in drug use may parallel normative peer patterns of substance in- volvement. Finally, it may be that treatment was differen- tially effective, such that alcohol use was influenced less than drug use. Of note, however, the increase in alcohol involvement seems more pronounced for those who display preadolescent CD behaviors. If, as suggested by existing data,26,35 more extensive deviant behaviors during early life are an indicator for the temporal persistence of antisocial behaviors, the present relation between preadolescent CD behaviors and greater alcohol involvement may continue into young adulthood. Of note, whereas childhood antiso- cial behaviors represent a well-documented risk factor for teen alcohol and drug abuse, the current results are among the first to demonstrate that these teens may be at greater risk for continued problems with alcohol after treatment than are other substance-abusing teens.

Preadolescent CD behaviors were associated with nega- tive self-statements reflecting self-blame and self-criticism (e.g., realize you brought the situation on yourself; criticize or lecture yourself) in coping responses to a relapse risk situation. As suggested by Marlatt,I8 an increased propen- sity for self-criticism in high-risk situations may influence alcohol use by leading to diminished self-efficacy for suc- cessful coping in relapse situations and by generating neg- ative affect, which in itself may motivate further alcohol use. Thus, these cognitive factors seem particularly impor- tant in understanding the addiction relapse process for adolescents. The present findings diverge from our predic- tions for CD behaviors and relapse risk appraisal variables. In previous studies, we have found that of adolescent re- lapse risk appraisal variables, greater perceived importance and higher self-efficacy correspond with less posttreatment alcohol and drug involvement in the first 6 months after treatment.20 In addition, general self-efficacy has also been related to 1-year outcome for teem2’ The present lack of relation between appraisal and early-life CD behaviors sug- gests that teens with more extensive CD behaviors before the onset of substance use do not differ from other alcohol and drug-abusing teens with respect to appraisal character- istics associated with relapse.

Motivation for alcohol and drug use, forethought before relapse, and impulsivity were assessed as cognitive-style variables hypothesized to influence the process of relapse for adolescent alcohol and drug abusers through decisions regarding both exposure to and attempts to cope with relapse risk situations. Of these, only teen motivation for

posttreatment alcohol abstinence was related to preadoles- cent CD behaviors. Differences in motivation for alcohol versus drug abstinence may be linked to peer influences and the normative patterns suggested herein, in that alco- hol use may have been more acceptable than drug use. Our measure of motivation for alcohol abstinence is interpreted as an indicator of cognitive state (behavioral intention) and might influence posttreatment alcohol use through teen decision making such that those with less motivation for abstinence may fail to avoid potential drinking situations or to use coping efforts in such situations. The lack of an observed relation between early CD behaviors with relapse forethought and impulsivity may reflect hyperactivity, which is common among teen substance abusers,’ rather than CD in this clinical sample of alcohol and drug-abusing adolescents.

Of the interpersonal functioning variables assessed, the composite measure of interpersonal difficulties was related to CD behaviors. This finding is consistent with data on correlates of CD, indicating that CD teens exhibit conflic- tual interpersonal relationship^.^^ As hypothesized, it may be that greater interpersonal conflict leads to more relapse risk situations and/or taxes coping resources, thereby con- tributing to alcohol use or increasing the attraction of alcohol use as a means for managing the affective conse- quences of conflict. The direction of relationship cannot, however, be ascertained from the present data; it may be that greater alcohol involvement exacerbates interpersonal problems. That CD behaviors were not significantly related to either the measure of family conflict or parental alcohol and drug use may reflect that families in this clinical sample are more disrupted than in the general population, thus limiting the ability to detect difference^.^^ This hypothesis receives partial support in the present data in that approx- imately one-third of the fathers in this sample are rated as heavy drinkers, which may lead to greater familial disrup- tion. However, the family conflict measure scores in our sample do not differ from those for nondistressed families.

Examination of the relation between preadolescent CD behaviors with exposure to substance use and social re- sources found a significant relation only with exposure to alcohol use. Greater exposure to alcohol-using models for teens with more CD behaviors likely reflects multiple in- fluences. For example, increased exposure may result from elevated availability of alcohol in the environment (family and community), peer alcohol use, or more frequent expo- sure to situations wherein alcohol is present (as suggested by the relation of CD behaviors with less motivation for alcohol abstinence). Thus, increased exposure to alcohol use may result from environmental factors, as well as se- lective decision making (i.e., choosing to attend situations presenting opportunities for alcohol use).

Of the aspects of teen CD found related to early-life CD behaviors, interpersonal problems and exposure to alcohol use are found to contribute significantly to the prediction of alcohol use outcome. Thus, of the CD characteristics hy-

CD BEHAVIORS AND RELAPSE 1535

pothesized to influence posttreatment outcome, environ- mental factors seem more influential than cognitive-style variables in predicting poorer course with respect to alco- hol involvement for teens with more preadolescent CD behaviors. As previously mentioned, we anticipated that variables such as coping skills and other cognitive variables were more likely to influence discrete incidents of relapse, whereas progression of addiction would relate more strongly with environmental variables, such as interper- sonal conflict and exposure to alcohol use. This distinction between discrete relapse episodes and the progression of addictive relapse is important in the present context, be- cause our composite outcome variable provides a global measure of alcohol involvement over time and as such better represents the course and progression of alcohol use. Thus, the limited relations demonstrated herein with cop- ing efforts and motivation may reflect the methods used rather than the relative importance of these variables in influ- encing initial or discrete episodes of a return to alcohol use.

The predictions for this study were based on CD theory and highlight the process of addictive relapse for teens transitioning from middle to late adolescence (during the ages of 16 to 18), which is an important period for the formation of personal identity, social role adjustment, and separation from family of origin. For adolescents in this age range, the peer group is the primary source of identification and a powerful influence on beha~ior.~' Clearly, alcohol use becomes increasingly common and continues to be more socially acceptable than drug use. It may be that the increased prevalence of alcohol use during late adolescence acts to diminish the perceived danger or negative conse- quences of alcohol use and fosters a discrimination be- tween substances legal for adults (alcohol) versus illegal psychoactive substances. For teens with greater preadoles- cent CD behaviors, peers may influence the increased choice of alcohol, particularly in response to circumstances such as increased interpersonal conflict. The influence of exposure to alcohol as a significant factor in the relation between early life CD and adolescent posttreatment alco- hol use highlights the importance for teens of alcohol avail- ability and peer alcohol involvement.

For several reasons, the current interpretation of the process of relapse and progression of addiction in relation to CD must be viewed with caution. First, early-life CD behaviors occurring before the onset of drug involvement were utilized to provide an indicator of CD and progression to adult ASPD?l This approach has the advantage of pro- viding a measure of deviant behavior independent of the subsequent influence of alcohol and drug involvement. However, the use of a continuous rather than categorical approach limits the ability to draw inferences as to the meanings of the present results with regard to the diagnosis of CD. A second issue involves the use of composite vari- ables designed to characterize behaviors and experiences over a 2-year period, thereby precluding conclusions as to the direction of causality. Because it is likely that the

variables examined herein relate in a reciprocal fashion, the statements as to the influence of correlates of CD on the process of posttreatment alcohol use must, at this time, be viewed as hypotheses. Finally, the sample excluded teens institutionalized for extended periods and consists predom- inantly of Caucasian lower-middle to middle class youth and their families. The process of relapse, progression of addiction, and related factors may differ in dissimilar clin- ical samples of alcohol and drug-abusing teens.

The present data suggest the value of assessing CD behaviors in temporal relation to alcohol and drug use history for alcohol and drug-abusing teens as a means of identifying those at greater risk for continued alcohol in- volvement. Support is also provided for the utility of the measures used herein, the CDDR and the CD Question- naire. Furthermore, targeting cognitive factors, such as motivation for alcohol abstinence, focusing on interper- sonal functioning and emphasizing the risk involved in exposure to alcohol-using models, may be particularly im- portant for alcohol and drug-abusing teens with a history of extensive CD behaviors preceding the onset of drug in- volvement.

Further research is needed to replicate and clarify the present findings. The finding of differential prediction of CD behaviors to alcohol versus drug involvement needs to be replicated in an independent sample. In addition, more comprehensive information is needed to allow identifica- tion of teens at risk for poorer treatment outcome because of early-life CD behaviors. For example, more detailed information regarding the extent and type of CD behaviors preceding drug use that are most related to elevated post- treatment alcohol involvement is needed to assist in iden- tifying teens at risk for poorer treatment outcome. Also, it is necessary to assess other important factors that may influence the process of relapse and addiction progression, such as coping with different types of relapse risk situations, coping with life stress, and interpersonal skills. Finally, further prospective studies of the process of relapse in relation to CD behaviors are needed to address causal relations between CD behaviors, correlates of CD, and treatment outcome.

REFERENCES 1. Regier DA, Farmer ME, Rae DS, Locke BZ, Deight SJ, Judd LL,

Goodwin F K Comorbidity of mental disorders with alcohol and other drug abuse. JAMA 264:2511-2518, 1990

2. Hesselbrock MN, Meyer RE, Keener JJ: Psychopathology in hos- pitalized alcoholics. Arch Gen Psychiatry 42:1050-1055, 1985

3. Rounsaville BJ, Dolinsky ZS, Babor TF, Meyer RE: Psychopathol- ogy as a predictor of treatment outcome in alcoholics. Arch Gen Psychi- atry 44:505-513, 1987

4. Booth BM, Yates WR, Petty F, Brown K Patient factors predicting early alcohol-related readmissions for alcoholics: Role of alcoholism se- verity and psychiatric co-morbidity. J Stud Alcohol 52:37-43, 1991

5. Carroll KM, Power MD, Bryant K, Rounsaville BJ: One-year fol- low-up status of treatment-seeking cocaine abusers-Psychopathology and dependence severity as predictors of outcome. J Nerv Mental Dis 181~71-79, 1993

1536 MYERS ET AL.

6. Schuckit MA, Schwei MG, Gold E: Prediction of outcome in inpatient alcoholics. J Stud Alcohol 47:219-224, 1986

7. Stowell RJA, Estroff TW: Psychiatric disorders in substance-abus- ing adolescent inpatients: A pilot study. J Am Acad Child Adolesc Psy- chiatry 31:1036-1040, 1992

8. Bukstein OG, Brent DA, Kaminer Y: Comorbidity of substance abuse and other psychiatric disorders in adolescents. Am J Psychiatry

9. Dimileo L. Psychiatric symptoms in adolescent substance abusers. Am J Psychiatry 146:1212-1214, 1989

10. Monopolis SJ, Brooner RK, Jadwisiak RM, Marsh E, Schmidt C W Preliminary report on psychiatric comorbidity in adolescent substance abusers, in Harris L (ed): Problems of Drug Dependence. NIDA Research Monograph 105. Washington, D.C., DHHS, 1990

11. Tarter R, Kirisci L, Hegedus A, Mezzich, A: Heterogeneity of adolescent alcoholism, in Babor TF, Hesselbrock VM, Meyer RE, Shoe- maker W (eds): Types of Alcoholics: Evidence from Clinical, Experimen- tal, and Genetic Research, vol 708. New York, New York Academy of Sciences, 1994, pp 172-180

12. Mezzich AC, Tarter R, Kirisci L, Clark D, et al: Subtypes of early age onset alcoholism. Alcoholism 17:767-770, 1993

13. Mott MA, Myers MG, Tammariello CF, Brown SA: Conduct Dis- order Diagnosis Is Related to Long-Term Adolescent Drug Treatment Outcome. Poster presented at the American Psychological Society Annual Meeting, San Diego, June 1992

14. Stewart DG: Antisocial behavior and long-term outcome of sub- stance abuse treatment, in Brown SA (Chair): Long Term Outcome Among Adolescents Following Alcohol and Drug Treatment. Symposium paper presented at the annual convention of the American Psychological Association, Los Angeles, CA, 1994

15. Robins LN, Price R K Adult disorders predicted by childhood conduct problems: Results from the NIMH epidemiological catchment area project. Psychiatry 54:116-132, 1991

16. Brownell KD, Marlatt GA, Lichtenstein E, Wilson GT: Under- standing and preventing relapse. Am Psychol 41:765-782, 1986

17. Brown SA, Mott MA, Richter SS, Copeland AL: Stress, vulnera- bility and adult alcohol relapse. Alcoholism 16:362, 1992

18. Marlatt GA: Relapse prevention: Theoretical rationale and over- view of model, in Marlatt GA, Gordon JR (eds): Relapse Prevention. New York, Guilford Press, 1985, pp 3-70

19. Wills TA, Shiffman S: Coping and substance use: A conceptual framework, in Shiffman S, Wills TA (eds): Coping and Substance Use. Orlando, Academic Press, 1985, pp 3-24

20. Myers MG, Brown SA: Coping and appraisal in relapse risk situa- tions among adolescent substance abusers following treatment. J Adolesc Chem Depend 1:95-190, 1990

21. Myers MG, Brown SA: Coping responses and relapse among ado- lescent substance abusers. J Subst Abuse 2:177-190, 1990

22. Myers MG, Brown SA, Mott MA: Coping as a predictor of ado- lescent substance abuse treatment outcome. J Subst Abuse 5:15-29, 1993

23. Myers MG, Brown S A The Adolescent Relapse Coping Question- naire: Psychometric validation. J Stud Alcohol (in press)

24. Brown SA, Vik PW, Patterson TL, Grant I, Schuckit MA: Stress, vulnerability and adult alcoholic relapse. J Stud Alcohol (in press)

25. Loeber R: Development and risk factors of juvenile antisocial behavior and delinquency. Clin Psychol Rev 1O:l-41, 1990

26. Robins LN: Conduct disorder. J Child Psychol Psychiatry 32:193- 212, 1991

27. Kazdin AE: Treatment of antisocial behavior in children: Current status and future directions. Psychol Bull 102:187-203, 1987

28. Prinz R, Connor P, Wilson C Hyperactive and aggressive behaviors in childhood: Intertwined dimensions. J Abnorm Child Psychol 9:191-202, 1981

29. Richter SS, Brown SA, Mott M A The impact of social support and

14611131-1141, 1989

self-esteem on adolescent substance abuse treatment outcome. J Subst Abuse 3:371-386, 1991

30. Johnston L, O’Malley P, Eveland L Drugs and delinquency: A search for causal connections, in Kandel DB (ed): Longitudinal Research on Drug Use: Empirical Findings and Methodological Issues. Washington, D.C., Hemisphere, 1978, pp 137-156

31. Robins LN: The consequences of conduct disorder in girls, in Olweus D, Block J, Radke-Yarrow M (eds): Development of Antisocial and Prosocial Behavior. Orlando, Academic Press, 1993, pp 385-414

32. van Kammen WB, Loeber R: Are fluctuations in delinquent activ- ities related to the onset and offset in juvenile illegal drug use and drug dealing? J Drug Issues 24:9-24, 1994

33. Brown SA, Mott MA, Stewart M Adolescent alcohol and drug abuse, in Walker CE, Roberts MC (eds.): Handbook of Clinical Child Psychology, ed 2. New York, John Wiley & Sons, 1992, pp 677-693

34. Loeber R, Wung P, Keenan K, Giroux B, Stouthamer-Loeber M, van Kammen WB, Maughan B: Developmental pathways in disruptive child behavior. Dev Psychopath01 5:103-133, 1993

35. Moffit TE: Adolescence-limited and life-course-persistent antiso- cial behavior: A developmental taxonomy. Psychol Rev 100:674-701,1993

36. Loeber R: The stability of antisocial and delinquent childhood behavior: A review. Child Dev 53:1431-1446, 1982

37. Donovan J, Jessor R: Structure of problem behavior in adolescence and young adulthood. J Consult Clin Psychol 53:890-904, 1985

38. American Psychiatric Association: Diagnostic and Statistical Man- ual of Mental Disorders, ed 3-rev. Washington, D.C., American Psychiat- ric Association, 1987

39. Brown SA, Creamer VA, Stetson BA. Adolescent alcohol expect- ancies in relation to personal and parental drinking patterns. J Abnorm

40. Brown SA, Vik PW, Creamer V A Characteristics of relapse fol- lowing adolescent substance abuse treatment. Addict Behav 14:291-300, 1989

41. Stewart DG, Brown SA: Withdrawal and dependency symptoms among adolescent substance abusers. Addiction 90:627-635, 1995

42. Babor TF, Brown J, del Boca F K Validity of self-reports in applied research on addictive behaviors: Fact or fiction? Behav Assess 125-31, 1990

43. Maisto SA, Connors GJ: Using subject and collateral reports to measure alcohol consumption, in Litten RZ, Allen JP (eds): Measuring Alcohol Consumption: Psychosocial and Biochemical Methods. Totowa, NJ, Humana Press, 1992, pp 73-96

44. Jackson DN: Personality Research Form, ed 3. Port Huron, MI, Research Psychologists Press, 1984

45. Moos RH, Moos BS: Family Environment Scale Manual, ed 3. Palo Alto, Consulting Psychologists Press, 1994

46. Sarason IG, Levine HM, Bashon RB, Sarason BR: Assessing social support: The social support questionnaire. J Pers SOC Psychol 44:127-139, 1983

47. Barron RM, Kenney D A The moderator-mediator variable dis- tinction in social psychological research: Conceptual, strategic, and statis- tical considerations. J Pers SOC Psychol 51:1173-1182

48. National Institute on Drug Abuse: National Survey Results on Drug Use from the Monitoring the Future Study, 1975-1992 (National Institute of Drug Abuse, DHHS Publication No. ADM 93-3597). Wash- ington, D.C., US. Government Printing Office, 1993

49. Stewart MA, Brown SA: Family functioning following adolescent substance abuse treatment. J Subst Abuse 5:327-337, 1993

50. Kelly JA, Hansen DJ: Social interactions and adjustment, in van Hasselt VB, Hersen M (eds): Handbook of Adolescent Psychology. New York, Pergamon Press, 1987, pp 131-146

51. Brown SA, Gleghorn AA, Schuckit MA, Myers MG, Mott MA: Conduct disorder among adolescent alcohol and drug abusers. J Stud Alcohol (in press)

Psycho1 96:117-121, 1987