Journal of Abnormal Child Psychology: Comorbidity of conduct and depressive problems at sixth grade: substance use outcomes across adolescence

FindArticles > Journal of Abnormal Child Psychology > June, 1998 > Article > Print friendly

Comorbidity of conduct and depressive problems at sixth grade: substance use outcomes across adolescence

Shari Miller-Johnson

The comorbidity of childhood and adolescent disorders has been largely ignored until recently (Caron & Rutter, 1991; McConaughy & Achenbach, 1994). One area that has received attention is for conduct disorder and depression (Capaldi, 1991, 1992; Loeber & Keenan, 1994). Comorbidity of these. problems is not uncommon (Angold & Costello, 1993; Loeber & Keenan, 1994). For example, Lewinsohn, Hops, Roberts, Seeley, and Andrews (1993) found a significant degree of comorbidity of depression and disruptive behavior disorders in a community-wide sample of adolescents. Although there are a number of complex methodological issues related to this area of study (see Angold & Costello, 1993; Caron & Rutter, 1991, for in-depth reviews), depression and conduct problems appear to occur at significantly higher rates than would be expected by chance and their comorbidity may have significant implications for nosology, treatment, and prognosis.

One outcome that is particularly relevant to comorbid depression and conduct problems is substance use (Rohde, Lewinsohn, & Seeley, 1996). Ties between antisocial behavior and subsequent adolescent substance use are a robust finding (Boyle, Offord, Racine, Szatmari, Fleming, & Links, 1992; Windle, 1990). Jessor and Jessor's problem behavior theory (1977, Jessor, 1992) has been used to explain associations between substance use and conduct problems. Within this framework, different problem behaviors are viewed as part of a broader deviance pattern that reflects a single, underlying syndrome and includes other norm-violating behaviors, such as delinquency, substance use, and risky sexual activity. In addition, risk and protective factors are shared across different problem behaviors. Thus, under the problem behavior theory, drug use is part of a deviance-prone life-style that emerges during adolescence.

Related to this, early substance use initiation has been shown to be a strong predictor of later use (Newcomb & Bentler, 1989). Windle (1990), using data from the National Longitudinal Survey of Youth, found that early adolescent substance use was the most consistent predictor of substance use 4 years later. Similarly, Farrell, Danish, and Howard (1992) found significant ties between previous and subsequent substance use in their sample of predominantly urban, African American teens. In part, this may explain associations between conduct problems and later substance use, in that these youth may be likely to initiate use earlier and therefore be more likely to use substances at a later point.

Ties between depression and substance use are less well understood, although associations between these variables have been found (Neighbors, Kempton, & Forehand, 1992; Simons, Whitbeck, Conger, & Melby, 1991). The self-medication hypothesis has been forwarded to explain this relationship. Under this model, substances are used to minimize distressing feelings associated with negative affective states (Khantzian, 1985). Related to this, stress and coping models of teen drug use (Wills & Filer, 1996) have theorized that substance use is an avoidant type of coping mechanism that helps individuals calm down when they ar depressed, resulting in a reduction of distressing feelings. A small number of studies have shown associations between using substances as a form of self-medication and depression (e.g., Weiss, Griffin, & Mirin, 1992). However, support for the self-medication hypothesis has been largely anecdotal and mixed at best.

Studies have generally focused on only one of these symptom domains, therefore, our understanding of comorbid depression and conduct problems as predictors of substance use is limited. Outcomes may vary across subgroups with conduct and/or depressive problems, and Caron and Rutter (1991) warn against possible false assumptions when comorbid conditions are not taken into account. First, relationships between a disorder and an outcome may really be due to a comorbid second condition. Second, by ignoring comorbidity, this implies that the meaning of condition is the same regardless of a second condition. The nature of influences between comorbid conditions remains unclear (Caron & Rutter, 1991), and developmental outcomes of comorbid conduct and depressive problems have received little attention.

One exception is a set of studies of early adolescent males comparing subgroups with elevated conduct and/or depressed symptoms (Capaldi, 1991, 1992). For most outcomes at sixth grade, an additive model was supported, with the comorbid boys experiencing combined adjustment problems of conduct and depressive conditions. However, for substance use, the comorbid group displayed significantly higher levels as compared with other groups. By Grade 8, substance use levels were equally high in both conduct problem groups (i.e., comorbid and conduct problems only) as compared with the depressed only and the nonproblem groups. Capaldi (1992) suggested that the combination of depression and conduct problems may be an important risk factor for substance use in early adolescence, although nondepressed, antisocial youth would "catch up" by Grade 8.

Different patterns were found by Henry et al. (1993) in their study of conduct problems and depressive symptoms at ages 11 and 15 and substance use at age 15. For males, depression at age 11 predicted substance use at age 15, whereas for females, no relationship was found between early symptoms and later substance use. For both males and females, relationships were strongest between concurrent conduct problems and drug use at age 15. The authors suggest that links between conduct problems and substance use emerge later in adolescence, while links between depression and substance use are more important earlier in development, at least for males.

This raises the important issue of gender in relation to conduct and depressive problems and subsequent substance use. During adolescence, rates of depression are higher, while rates of conduct problems are lower for females (McGee, Feehan, Williams, & Anderson, 1990). Substance use is somewhat higher among adolescent males, with the exception of cigarette use where levels have been shown to be equivalent by gender or even slightly higher for females (National Institutes on Drug Abuse, 1991, 1996). It has been speculated that associations between depression and substance use are higher for females (Pagliaro & Pagliaro, 1996; Windle, 1990). However, with the exception of the Henry et al. (1993) study, this question has not been examined with consideration of comorbid conduct and depressive problems and the impact this might potentially have on substance use.

There are other methodological reasons that warrant further examination of conduct and depressive symptoms and substance use. The bulk of studies have focused on Caucasian samples. However, youth from different racial groups may vary on characteristics related to substance use, such as family, peer, and contextual variables. In addition, rates of substance use are lower for African American youth in comparison with Caucasian youth (National Institute on Drug Abuse, 1991). Findings on racial differences between various risk factors and substance use have been inconsistent, with some studies showing variability (e.g., Newcomb, Maddahian, Skager, & Bentler, 1987), and other finding similar patterns (e.g., Gottfredson & Koper, 1996). Another important methodological concern is that short-term findings may vary from long-term patterns (Henry et al., 1993). Therefore, there is a need for additional longitudinal study of associations between conduct and depressive problems and substance use.

The purpose of this study is to examine the comorbidity of conduct and depressive symptoms and substance use outcomes in adolescence. This study addressed many of the methodological concerns of previous research. First, we were able to expand the range of samples in which these questions are examined through our investigation of an understudied population, namely, an urban, African American sample of males and females. Second, substance use was examined prospectively, thus allowing for examination of short- and long-term patterns. Third, the study included measures of conduct and depressive problems, thereby allowing for determination of the relative importance of both problems areas as predictors of substance use. based on findings from Capaldi (1991, 1992), we hypothesized that overall, substance use levels would be highest in youth with conduct problems, regardless of comorbid status. However, we also hypothesized that for females, levels would be highest for girls with comorbid conduct and depressive problems.

METHOD

Participants

Participants were part of a prospective, longitudinal study examining psychological functioning from third grade through young adulthood (Coie, Lochman, Terry, & Hyman, 1992; Coie, Terry, Lenox, Lochman, & Hyman, 1995). In the spring of 1984, 1985, and 1986, all third-grade children in 12 elementary schools were administered sociometric measures of peer status and social behavior. A total of 1,749 children were screened across the three cohorts at third grade (Cohort A: n = 588; Cohort B: n = 559; Cohort C: n = 602).

Within this larger sample (N = 1,749), longitudinal assessments were collected on a smaller subsample at 2-year intervals across adolescence (N = 622). Active consent was obtained from parents and youth. Interviews were completed by trained staff in the home and youth were interviewed without parents or other individuals present. Selection of the longitudinal sample was based on data from the third-grade sociometric survey. First, the liked most liked most and liked least scores were standardized within school. These standard scores were used to compute social preference (SP; liked most minus liked least) and social impact (SI; liked most plus liked least) scores. Peer status groups were then assigned as follows: rejected: SP score of less than -1.0; popular: SP score greater than 1.0; neglected: SI score of less than -1.0; controversial: SI score greater than 1.0; and average: both SP and SI scores between -1.0 and 1.0. A stratified, random sample from each of the original third-grade cohorts was chosen within the four nonrejected (i.e., popular, neglected, controversial, and average) social status groups. The sample was more heavily weighted for rejected status due to questions about peer rejection as a risk factor for negative outcomes. Of the 622 subjects in the longitudinal sample, 340 (54%) had complete data at Grades 6, 8, and 10 and were included in analyses. The proportion of the final study sample that fell into each of the peer status groups was average (50%), controversial (6%), neglected (6%), popular (9%), and rejected (29%). Attrition information is provided later in the paper.

The population served by the Durham city schools was predominantly African American (90%) and of low socioeconomic status (65%) of children in the school system were eligible for the free or reduced cost school lunch program). Given problems in interpreting sociometric data about children who are in an extreme minority of a school population (Kupersmidt & Coie, 1990), only African American subjects were included in the longitudinal follow-up, although sociometric data was obtained for all children in the school.s

Measures

Child Assessment Schedule. Subjects were interviewed using the Child Assessment Schedule (CAS; Hodges, 1987). The CAS, a semistructured interview for children, generates psychiatric symptom information. Psychometric properties of the CAS compare favorably with other standardized psychiatric interviews. Two summary indexes were derived from this measure: total score for conduct symptoms and total score for depression symptoms. The CAS was administered at Grades 6, 8, and 10.(4)

National Youth Survey. Subjects were interviewed using substance use items from the National Youth Survey (NYS; Elliot, Ageton, & Huizinga, 1985; Elliot, Huizinga, & Menard, 1989). The NYS assesses youth reports of delinquent behaviors. The NYS has displayed acceptable reliability and validity in a national probability sample, in terms of test-retest reliability and comparable prevalence rates with another national probability sample. A continuous score was generated that indicated how many times in the last year a subject reported using tobacco, alcohol, and marijuana. Given the skewed nature of the data, responses were recorded using a natural log+1 transformation. The NYS was administered at Grades 6, 8, and 10.

Group Classification

Subjects were classified into four groups related to increased levels of conduct and/or depressive symptoms based on sixth-grade data. It is noted that these categories do not denote formal diagnostic groups. Rather, they represent subjects who reported high levels of conduct and depressive symptoms relative to other subjects in the sample, Elevated scores may reflect the presence of the disorder and/or subclinical levels. It is noted that dichotomization is somewhat arbitrary and may increase the risk of false positives. However, elevated symptom levels can be relevant in terms of predicting functioning. The cutoff was chosen in a similar manner to previous research (Capaldi, 1991, 1992) to facilitate generalizability of findings.(5)

Raw scores for conduct and depressive symptoms from the CAS based on sixth-grade data were standardized within the sample, and then dichotomized into high/low groups using a z-score cutoff of greater than or equal to 0.5. The two scores were then crossed, and subjects were placed in one of four groups: (a) high on conduct and depressive symptoms (CDDE; n = 38, 11%); (b) high on conduct symptoms only (COND; n = 61, 18%); (c) high on depressive symptoms only (DEPR; n = 20, 6%); and, (d) low on conduct and depressive symptoms (NONE; n = 221, 65%). Group membership did not vary by gender.(6)

Attrition

Attrition was due primarily to the inability to locate subjects. Additionally, a smaller number of subjects refused to be interviewed. Youth participation rates were 84, 79, and 73% at Grades 6, 8, and 10. No differences were found in comparing subjects with complete data versus subjects with incomplete data at the three time points on sixth-grade conduct and depressive symptoms, tobacco, alcohol, or marijuana use or on third-grade peer status. A trend was found on third-grade peer ratings of aggression, with subjects having incomplete data displaying higher levels of aggression in comparison with subjects with complete data, t(584) = -1.92, p [less than] 10. This pattern was robust for males, t(298) = -2.77, p [less than] .01, while there were no differences in aggression between females with complete and incomplete data at Grades 6, 8, and 10.

RESULTS

First, descriptive information about substance use is provided. This is followed by results on early substance use in Grade 6 as a predictor of later use. Finally, findings on conduct and depressive problems as predictors of substance use at Grades 6, 8, and 10 are provided.

Descriptive Information: Substance Use at Grades 6, 8, and 10

Table I reports nontransformed substance use mean levels at Grades 6, 8, and 10. Levels were lowest at Grade 6 and rose substantially by Grade 10. Use was lowest for marijuana use. Substance use levels did not differ for males and females, with the exception of higher alcohol use for males at Grade 10, t(337) = 2.12, p [less than] .05, and a trend towards higher marijuana use for males at Grade 10, t(331) = -1.77, p [less than] .10. Reports of substance use in this sample were similar to national survey data for African Americans (National Institute on Drug Abuse, 1991, 1996).

Substance Use at Grade 6 as a Predictor of Later Use

Analyses first examined the question of whether the same subjects who reported use at Grade 6 continued to report use at Grades 8 and 10. Subjects were divided into dichotomous categories (use vs. nonuse) for tobacco, alcohol, and marijuana at the three assessment periods. For each of the three substances, chi-square values were highly significant from Grade 6 to 8, 6 to 10, and 8 to 10. Thus, substance use showed a high degree of continuity, with subjects reporting use at one period more likely to report continued use. The question of early use as predictor of later use was further examined using repeated measures analyses of variance (ANOVA), with continuous use as the outcome. Gender was entered first into the equation, followed by dichotomous use versus nonuse, and the interaction of gender by use, with separate analyses for each of the substances. The repeated measures ANOVAs showed robust effects for use for all three substances: tobacco, F(1,322) = 11.24, p [less than] .001; alcohol, F(1, 314) = 31.11, p [less than] .001; marijuana, F(1, 325) = 97.79, p [less than] .001. Effects for gender and group by gender were significant for alcohol: gender, F(1, 314) = 11.14, p [less than] .001; group by gender, F(1, 314) = 14.72, p [less than] .001 and marijuana: gender, F(1, 325) = 33.06, p [less than] .001; group by gender, F(1, 325) = 29.18, p [less than] .001. Levels were higher for males for these two substances. For alcohol use, males who reported use at sixth grade showed an increase across Grades 8 and 10, while levels for females who reported use at sixth grade varied only at eighth grade and converged at tenth grade. For marijuana, both males and females who reported use at sixth grade displayed higher levels in eighth and tenth grades. However, the difference between users and nonusers was relatively greater for males than for females.

Table I. Substance Use at Grades 6, 8, and 10

                     M         SD

Tobacco

Grade 6             0.47       3.23
Grade 8             7.64      61.82
Grade 10           43.92     177.36

Alcohol

Grade 6             0.67       3.53
Grade 8             3.78      24.60
Grade 10           11.00      55.79

Marijuana

Grade 6             0.10       0.55
Grade 8             4.32      53.48
Grade 10            4.04      26.50

Conduct and Depressive Problems as Predictors of Substance Use at Grades 6, 8, and 10

Repeated measures ANOVA were carried out for the following substance use outcomes at Grades 6, 8, and 10: (a) tobacco, (b) alcohol, and (c) marijuana. For ease of interpretation, nontransformed mean levels are reported in the text and figures that indicate the number of times the particular substance was used in the last year. Gender was entered first into the equation, followed by conduct and/or depressive problems group membership and the interaction of gender by group. Six planned contrasts (Table II) were carried out: (a) conduct problems only (COND) versus comorbid conduct and depressive problems (CDDE); (b) depressive problems only (DEPR) versus CDDE; (c) COND versus DEPR; (d) COND versus low on conduct and depressive problems (NONE); (e) DEPR versus NONE; (f) CDDE versus NONE. Contrasts are reported separately for males and females when the group by gender interaction was significant at a given time point. Given small samples for group by gender analyses, these results are considered exploratory and should be interpreted with caution. Trends at the 0.10 significance [TABULAR DATA FOR TABLE II OMITTED] level are noted for descriptive purposes. Time effects were examined using the multivariate Wilks's [Lambda] test.(7)

Tobacco

The repeated measures ANOVA showed a significant main effect for group, F(3, 318) = 5.55, p [less than] .01; effects for gender and group by gender were nonsignificant. As shown in Fig. 1, tobacco use was highest for the CDDE group at all three grades. At Grade 6, tobacco use was significantly higher in the CDDE group as compared with all other groups, suggesting a distinct pathway of tobacco use for youth with comorbid conduct and depressive symptoms. Higher tobacco levels in the CDDE group continued at Grades 8 and 10, with levels rising from 0.97 to 2.86. However, tobacco levels in the COND and NONE groups also rose at this time (Grades 8 and 10: COND: 0.45, 1.34; NONE: 0.23, 0.80), and the contrast comparing the CDDE and these groups was no longer significant at Grades 8 and 10. The CDDE group remained significantly higher than the DEPR group at all time points.

Time effects were highly significant, F(2, 317) = 10.72, p [less than] .001. The time by group by gender interaction was also significant, F(6, 634) = 2.66, p [less than] .05; univariate results indicated that the group by gender interaction was significant at Grade 10 only. At tenth grade, tobacco use was higher for males in the two conduct problem groups, as supported by either significant contrasts or trends in comparisons between both conduct problem groups and the NONE and DEPR groups, whereas for females, levels were even across groups and all contrasts were nonsignificant.

Alcohol

Significant main effects were shown for group, F(3, 310) = 14.47, p [less than] .001, and a trend was exhibited for gender, F(1, 310) = 3.32, p [less than] .10. The group by gender interaction was also significant, F(2, 310) = 3.98, p [less than] .01; univariate analyses indicated that this interaction was significant at Grade 10 only. Overall, alcohol levels were higher in the two conduct problem groups, and lower in the DEPR and the NONE groups. At Grade 6, alcohol levels for the CDDE and COND groups appeared similar (0.61 and 0.67, respectively), and were higher in comparison with other subjects, as supported by significant contrasts between these two groups and the DEPR and NONE groups. At Grade 8, alcohol use for comorbid youth increased considerably to 1.92, and was significantly greater than levels in all other groups (COND: 0.73, DEPR: 0.32; NONE: 0.35). By Grade 10, alcohol use increased in the COND group, and the two conduct problem groups again displayed higher levels (CDDE: 3.35, COND: 3.10) as compared with levels in the DEPR (0.61) and the NONE (1.05) groups [ILLUSTRATION FOR FIGURE 2 OMITTED].

Effects for time were highly significant, F(2, 309) = 27.28, p [less than] .001. Time also interacted significantly with gender, F(2, 309) = 4.64, p [less than] .05, group, F(6, 618) = 2.29, p [less than] .05, and group by gender, F(6, 618) = 2.18, p [less than] .05. Alcohol levels were equivalent by gender at Grades 6 and 8 (males: 0.34; females: -0.32). However, from Grade 8 to 10, females' use increased gradually (0.75 to 1.05), while males' use almost quadrupled (0.73 to 2.74). Alcohol use for males and females across the four groups appeared similar at Grades 6 and 8. However, at Grade 10, alcohol levels for males in the two conduct problem groups were significantly higher than in the NONE and the DEPR groups, while for females, alcohol levels appeared even across the four groups.

Marijuana

The repeated measures ANOVA displayed a significant main effect for group, F(3, 321) = 12.06, p [less than] .001. Effects for gender and group by gender were nonsignificant. Overall, marijuana levels were highest in the CDDE groups at Grades 6 and 8 (0.21 and 0.75), followed by levels in the COND group at these two grades (0.11 and 0.34). Levels in the DEPR and the NONE groups were near zero at Grades 6 and 8. Contrasts between the CDDE and the DEPR groups were significant at these two time points, and contrasts between the CDDE and the COND groups displayed trends towards significance. These patterns suggests a distinct trajectory for subjects with combined depression and conduct problems at these time points. At Grade 10, marijuana use in the COND group increased to 0.99 and was somewhat higher than levels in the CDDE group (0.68). Levels remained low in the DEPR and NONE groups over time, and contrasts comparing these two groups were nonsignificant at all time points [ILLUSTRATION FOR FIGURE 3 OMITTED].

Effects for time were significant, F(2, 320) = 13.92, p [less than] .001, as were interactions for time with gender, F(2, 320) = 4.64, p [less than] .05, group, F(6, 640) = 4.28, p [less than] .001, and group by gender, F(6, 640) = 3.52, p [less than] .01. The univariate group by gender interaction was significant at Grade 10 only. Levels were fairly equivalent by gender at Grades 6 and 8 (males: 0.05, 0.28; females: 0.12, 0.26). At Grade 10, use increased for males to 0.73, while levels for females remained even at 0.23. The increase for males was particularly evident in the two conduct problem groups. For females, use increased only in the COND group, although not to as great a degree as for males.

DISCUSSION

The purpose of the present study was to examine substance use outcomes of comorbid conduct and depressive problems across adolescence. Overall, many of the findings support earlier studies (Capaldi, 1991, 1992) and show that the two conduct problem groups exhibited the highest levels of substance use over time. However, at some time points, comorbid youth displayed elevated substance use levels that were over and above levels in either of the single problem groups. Substance use levels for depressed youth without comorbid conduct problems were low and generally even with levels for youth in the nonproblem group.

Youth who displayed high levels of conduct problems during early adolescence were at risk for increased levels of substance use from sixth through tenth grades, with similar trends for tobacco, alcohol, and marijuana use. This conclusion is supported by higher overall levels of substance use in the two conduct problem groups over time. In addition, the majority (though not all) of the contrasts comparing the two conduct problem groups to the remaining subjects are significant. Earlier conclusions about the role of antisocial behavior as a predictor of adolescent substance use (Boyle et al., 1992; Windle, 1990) are supported in this sample of urban, African American males and females. Substance use appears to be part of an interwoven set of antisocial behaviors, including aggression, rule-breaking, and violation of basic rights of others (Jessor & Jessor, 1977). Substance use may not be an outcome of early conduct problems, but rather a developmental manifestation of this broader syndrome that emerges during the teen years. In addition, as expected, early substance use is a robust predictor of later substance use. Therefore, a possible related mechanism of why conduct problem youth are at increased risk for later use is due to early initiation of substance use.

Some analyses revealed significantly higher levels of substance use for youth with comorbid conduct and depressive conditions. Specifically, tobacco use at Grade 6, alcohol use at Grade 8, and marijuana use at Grades 6 and 8 were higher for subjects with elevated conduct and depressive problems as compared with either of the single problem groups. Increased substance use in youth with comorbid conduct and depressive problems cannot be attributed to more extreme symptom levels, since symptoms in the comorbid group at Grade 6 were not higher as compared with the single problem groups. It may be, as Capaldi (1991) suggested, that comorbidity signals risk for early initiation, although conduct problem youth would "catch up" later on in their use of substances. Indeed, in the current sample, substance use levels appeared fairly even in the two conduct problem groups by Grade 10.

Another possibility is that the combination of these two symptom clusters comprises a meaningful pattern that results in a distinct developmental trajectory for substance use outcomes (Caron & Rutter, 1991). Patterson and Capaldi (1990) developed a "dual failure" model that provides a theoretical framework that may be useful in understanding comorbid conduct and depressive conditions. They hypothesized that antisocial behavior increases susceptibility to depression due to interference with social skill acquisition. Social deficits in antisocial boys may be further amplified by high levels of interpersonal conflict with peers, teachers, and parents. This conflict leads to rejection that is intensified over and above the sole effect of social skills deficits. Experiences of failure in school and difficulties in developing friendships may then lead to depressed mood. Thus, some youth with conduct problems may be at increased risk for developing depressive problems. Although purely speculative on our part, symptoms of depression that are prominent during the teen years, such as irritability, emotional lability, and highly reactive mood states, may interact with defiant, aggressive, and impulsive rule-breaking behaviors that characterize conduct difficulties. This may intensify problem behaviors, which can include increased substance use.

It is noteworthy that substance use levels were not elevated in youth with high levels of depressive symptoms without concurrent high levels of conduct symptoms. These findings are inconsistent with self-medication theories of substance use (Khantzian, 1985), and as previously stated, empirical support for this premise has been lacking. This may be especially true for African American youth, given the later age of onset and relatively lower substance use levels in adolescence (Gottfredson & Koper, 1996). Another possibility is that depression may lead to increased substance use only when it is accompanied by conduct problems. The importance of controlling for antisocial behavior when examining depression and associated outcomes is underscored by these findings. Additional studies across various racial and age groups during adolescence that examine motives and coping strategies related to substance use are needed to further examine theories related to mood regulation functions of substance use.

There is some evidence for differing developmental outcomes by gender. Group by gender interactions did not emerge until Grade 10, and long-term outcomes of conduct problems on subsequent substance use were stronger for males as compared with females. At tenth grade, substance use for males remained high in the two conduct groups, while for females, substance use'"levels were more even across groups. It may be that early conduct problems in females lead to different outcomes over time, such as teenage parenthood (Miller-Johnson et al., 1997). Given the small sample, results should be viewed with caution. Further exploration of possible gender variations in the prediction of substance use is warranted.

Several limitations in the study should be noted. First, subjects were exclusively African American, and therefore, findings may not generalize to other racial groups. Second, attrition analyses found that aggressive male subjects (based on peer ratings at Grade 3) were disproportionately missing in terms of having complete data at Grades 6, 8, and 10. It is possible that, if aggressive males were more highly represented in either the conduct only or the comorbid group, that differences between these two groups could become obscured. It is also possible that the influence of this differential attrition would be to lower levels of substance use in the sample, given that highly aggressive males would be more likely to be involved in drug behavior. Therefore, current findings may be biased to some extent and underestimate levels of substance use in the population under study. It is noteworthy that significant group differences in substance use outcomes did emerge, in spite of this attrition. Third, measures of conduct and depressive symptoms relied exclusively on youth reports, and findings may have differed if parent or teacher ratings were used (McConaughy & Achenbach, 1994). The number of subjects in the three problem groups was also small for examination of group patterns by gender, and these results should be considered exploratory in nature.

Also critical to note is that this study focused on substance use among adolescents, as opposed to substance abuse. Measures of intensity and quantity of use, adverse social consequences, and dependency symptoms were not available to make this discrimination (Newcomb & Bentler, 1989; Windle, 1996). Distinctions in adolescent drug behavior may have important implications in terms of varying causal pathways, risk for adult substance disorders, and intervention strategies. Therefore, it remains unclear the extent to which conduct problem youth in this sample were at risk for substance abuse. Future studies would benefit from inclusion of measures that take into account the multidimensional and heterogeneous nature of substance use.

The results of the current study have implications for interventions that address substance use. Conduct difficulties during early adolescence appear to be an important risk indicator for substance use. Increased levels of drug behavior emerged by Grade 6 in youth with conduct problems, and this pattern was stable across the teen years. Inclusion of a focus on skills designed to prevent substance use (e.g., resistance to peer pressure, assertion skills) appears warranted for youth with conduct problems. Such an intervention would most strategically take place during childhood before substance use is initiated and more difficult to change (Lochman & Wells, 1996). Some youth with conduct problems also appear to be at elevated risk for depression. Special attention to assessment of depression is critical when conduct problems are present. These internalizing symptoms are likely to be less overt, particularly given the often serious and overshadowing nature of conduct problems. Careful treatment and monitoring of clinical cases are important when comorbid conduct and depressive problems are exhibited.

In summary, the current study adds to the growing knowledge base on comorbid conduct and depressive conditions and substance use outcomes across adolescence. Findings underscore the importance of evaluating multiple symptom clusters when assessing psychological function. Results also have implications for preventive interventions designed to reduce substance use. Questions for future study include continued investigation of gender differences in exploring associations between comorbid conduct and depressive problems and careful examination of the heterogeneous nature of substance use among adolescents as an importance outcome.

ACKNOWLEDGMENTS

This research was supported by Grants R01 MY390140 and K05MH00797 from the Prevention Branch of the National Institute of Mental Health. We are grateful to the staff and children of the Durham, North Carolina schools for their cooperation in this longitudinal study.

4 Information on continuity of symptoms and diagnostic groupings across the three time points is available from the first author.

5 To illustrate the relevance of elevated symptom levels, Gotlib, Lewinsohn, and Seeley (1995) found that individuals with elevated depressive symptoms but who did not meet diagnostic criteria, did not differ from diagnosed individuals on most psychosocial measures. Similarly, Lahey et al. (1995) found that most youth with conduct disorder fluctuated above and below the diagnostic threshold over time, but remained high on symptom totals. Therefore, elevated symptom levels are not necessarily benign. The reader is referred to Waldman, Lilienfeld, and Lahey (1995) and Widiger (1992) for more detailed discussion of classification issues.

6 Although analyses were not statistically significant more females were present in the DEPR group (60% female) and more males were present in the COND group (57% male). In terms of depressive symptoms, group categorization was based on Grade 6 functioning, and gender differences in depression are generally not apparent until after pubertal onset (Nolen-Hoeksema & Girgus, 1994). In terms of conduct symptoms, the CAS included both aggressive and nonaggressive symptoms, which may have contributed to reduced gender variations in group composition. No significant differences were found in symptom levels between the comorbid group and each of the single problem groups.

7 Tables of means for group by gender analyses are available from the first author. Analyses were also carried examining continuous symptoms as predictors of substance use outcomes. Overall, results supported main effects for conduct symptoms as significant predictors of substance use. Significant interaction effects were shown between conduct and depressive symptoms in predicting alcohol use at Grade 8 and tobacco use at Grade 6. More detailed results of these analyses can be obtained from the first author.

REFERENCES

Angold, A., & Costello, E. (1993). Depressive comorbidity in children and adolescents: Empirical, theoretical, and methodological issues. American Journal of Psychiatry, 150, 1779-1791.

Boyle, M. H., Offord, D. R., Racine, Y. A., Szatmari, P., Fleming, J. E., & Links, P.S. (1992). Predicting substance use in late adolescence: Results from the Ontario child health study follow-up. American Journal of Psychiatry, 149, 761-767.

Capaldi, D. M. (1991). Comorbidity of conduct problems and depressive symptoms in early adolescent boys: I. Familial factors and general adjustment at grade 6. Development and Psychopathology, 3, 277-300.

Capaldi, D. M. (1992). Comorbidity of conduct problems and depressive symptoms in early adolescent boys: II. A 2-year follow-up at grade 8. Development and Psychopathology, 4, 125-144.

Caron, C., & Rutter, M. (1991). Comorbidity in child psychopathology: Concepts, issues, and research strategies. Journal of Child Psychiatry and Psychology, 32, 1063-1080.

Coie, J. D., Lochman, J., Terry, R., & Hyman, C. (1992). Predicting early adolescent disorder from childhood aggression and peer rejection. Journal of Consulting and Clinical Psychology, 60, 783-792.

Coie, J. D., Terry, R., Lenox, K., Lochman, J., & Hyman, C. (1995). Childhood peer rejection and aggression as predictors of stable patterns of adolescent disorder. Development and Psychopathology, 7, 697-713.

Elliott, D. S., Ageton, S.S., & Huizinga, D. (1985). Explaining delinquency and drug use. Beverly Hills: Siegel.

Elliott, D. S., Huizinga, D., & Menard, S. (1989). Multiple problem youth: Delinquency, substance use, and mental health. New York: Springer-Verlag.

Farrell, A.D., Danish, S. J., & Howard, C. W. (1992). Risk factors for drug use in urban adolescents: Identification and cross validation. American Journal of Community Psychology, 20, 263-286.

Gotlib, I. H., Lewinsohn, P.M., & Seeley, J. R. (1995). Symptoms versus a diagnosis of depression: Differences in psychosocial functioning. Journal of Consulting and Clinical Psychology, 63, 90-100.

Gottfredson, D.C., & Koper, C. S. (1996). Race and sex differences in the prediction of drug use. Journal of Consulting and Clinical Psychology, 64, 305-313.

Henry, B., Feehan, M, McGee, R., Stanton, W., Moffitt, T. E., & Silva, P. (1993). The importance of conduct problems and depressive symptoms in predicting adolescent substance use. Journal of Abnormal Child Psychology, 21, 469-480.

Hodges, K. (1987). Assessing children with a clinical interview: The Child Assessment Schedule. In R. J. Prinz (Ed.), Advances in behavioral assessment of children and families (pp. 203-233). Greenwich, CT: JAI.

Jessor, R., & Jessor, S. L. (1977). Problem behavior and psychosocial development: A longitudinal study of youth. San Diego, CA: Academic Press.

Jessor, R. (1992). Risk behavior in adolescence: A psychosocial framework for understanding and action. Developmental Review, 12, 374-390.

Khantzian, E. J. (1985). The self-medication hypothesis of addictive disorders: Focus on heroin and cocaine dependence. American Journal of Psychiatry, 142, 1259-1264.

Kupersmidt, J. B., & Coie, J. D. (1990). Preadolescent peer status, aggression, and school-adjustment as predictors of externalizing problems in adolescence. Child Development, 61, 1350-1362.

Lahey, B. B., Loeber, R., Hart, E. L., Frick, P. F., Applegate, B., Zhang, Q., et al. (1995). Four-year longitudinal study of conduct disorder in boys: Patterns and predictors of persistence, Journal of Abnormal Psychology, 104, 83-93.

Lewinsohn, P. M., Hops, H., Roberts, R. E., Seeley, J. R., & Andrews, J. A. (19993). Adolescent psychopathology: I. Prevalence and incidence of depression and other DSM-III-R disorders in high school students, Journal of Abnormal Psychology, 102, 133-144.

Lochman, J. E., & Wells, K. C. (1996). A social-cognitive intervention with aggressive children: Prevention effects and contextual implementation issues. In R. Dev. Peters & R. J. McMahon (Eds.), Preventing childhood disorders, substance abuse, and delinquency (pp. 90-110). Thousand Oaks, CA: Sage.

Loeber, R., & Keenan, K. (1994). Interaction between conduct disorder and its comorbid conditions: Effects of age and gender. Clinical Psychology Review, 14, 497-523.

McConaughy, S. H., & Achenbach, T. M. (1994). Comorbidity of empirically based syndromes in matched general populations and clinical samples. Journal of Child Psychiatry and Psychology, 35, 1141-1157.

McGee, R., Feehan, M., Williams, S., & Anderson, S. (1990). DSM-III disorders in a large sample of adolescents. Journal of the American Academy of Child and Adolescent Psychiatry, 29, 611-619.

Miller-Johnson, S., Winn, D. M., Coie, J. D., Hyman, C., Terry, R., Lochman, J., & Maumary-Gremaud, A. (in press). Parenthood during the teen years: A developmental perspective on risk factors for childbearing. Development and Psychopathology.

National Institute on Drug Abuse. (1991). National Household Survey on Drug Abuse: Main findings 1990. Rockville, MD: Author.

National Institute on Drug Abuse. (1996). National survey results on drug use from the Monitoring the Future Study, 1975-1995, Rockville, MD: Author.

Neighbors, B., Kempton, T., & Forehand, R. (1992). Comorbidity of substance abuse with conduct, anxiety, and depression disorder in juvenile delinquents. Addictive Behaviors, 17, 379-386.

Newcomb, M.D., Maddahian, E., Skager, R., & Bentler, P.M. (1987). Substance abuse and psychosocial risk factors among teenagers: Associations with sex, age, ethnicity, and type of school. American Journal of Drug and Alcohol Abuse, 13, 413-433.

Newcomb, M.D., & Bentler, P. M. (1989). Substance use and abuse among children and teenagers. American Psychologist, 44, 242-248.

Nolen-Hoeksema, S., & Girgus, J. S. (1994). The emergence of gender differences in depression during adolescence. Psychological Bulletin, 115, 424-443.

Pagliaro, A. M., & Pagliaro, L. A. (1996). Substance use among children and adolescents. New York: Wiley.

Patterson, G. R., & Capaldi, D. M. (1990). A mediational model for boys' depressed mood. In J. Rolf, A. S. Masten, D. Cicchetti, K. H. Neuchterlein, & S. Weintraub (Eds.), Risk and protective factors in the development of psychopathology (pp. 141-163). New York: Cambridge University Press.

Rohde, P., Lewinsohn, P.M., & Seeley, J. R. (1966). Psychiatric comorbidity with problematic alcohol use in high school students. Journal of the American Academy of Child and Adolescent Psychiatry, 35, 101-109.

Simons, R. L., Whitbeck, L. B., Conger, R. D., & Melby, J. N. (1991). The effect of social skills, values, peers, and depression on adolescent substance use. Journal of Early Adolescence, 11, 466-481.

Waldman, I.D., Lilienfeld, S. O., & Lahey, B. B. (1995). Toward construct validity in the childhood disruptive behaviors disorders. Classification and diagnosis in DSM-IV and beyond (pp. 323-363). In T. H. Ollendick & R. Prinz (Eds.), Advances in clinical child psychology (Vol. 17). New York: Plenum Press.

Weiss, R. D., Griffin, M. L., & Mirin, S. M. (1992). Drug abuse as self-medication for depression: An empirical study. American Journal of Drug and Alcohol Abuse, 18, 121-129.

Widiger, T. A. (1992). Categorical versus dimensional classification: Implications from and for research. Journal of Personality Disorders, 6, 287-300.

Wills, T. A., & Filer, M. (1996). Stress-coping model of adolescent substance use. In T. H. Ollendick & R. J. Prinz (Eds.), Advances in clinical child psychology (Vol. 18, pp. 91-132). New York: Plenum Press.

Windle, M. (1990). A longitudinal study of antisocial behaviors in early adolescence as predictors of late adolescent substance use: Gender and ethnic group differences. Journal of Abnormal Psychology, 99, 86-91.

Windle, M. (1996). An alcohol involvement typology for adolescents: Convergent validity and longitudinal stability. Journal of Studies on Alcohol, 57, 627-637.