The Relationship Between Mental Health and Substance Abuse Among Adolescents

4.5 Relationship of Syndromes to Substance Use

Logistic regression was used to assess the relationship between syndromes and substance use for the different age and gender groups. For these analyses, the eight YSR syndrome scales were used to predict each of the individual measures of substance use (e.g., past month cigarette smoking). The analyses were conducted for the entire sample and for each specific age and gender group (e.g., 12 to 13 year old males). Results are shown in Tables 4.8 to 4.14. Delinquent Behavior was the syndrome most associated with substance use, followed by scales measuring Social Problems and Attention Problems. Delinquent Behavior was consistently associated with substance use even when the alcohol and drug use item was omitted from the scale. This result indicates that mild to serious substance use is associated with various other forms of delinquent behavior in a national sample of adolescents. In addition, this finding confirms the placement of the alcohol and drug use item in the Delinquent Behavior scale (Achenbach, 1991).

Because logistic regression models are nonlinear, for continuous independent variables the contribution of the regression coefficent in estimating the probability of the outcome variable dependson the value of the variable of interest and the values of other variables in the model.8 To illustrate key findings, different values of the syndrome of interest are selected to illustrate the difference in the estimated probability of substance use. The value for all other syndromes is set at the respective mean. For example, the estimated probability of past month marijuana use for adolescents aged 12 to 17 with the lowest possible raw score for Delinquent Behavior (0) was approximately .02. The probability of past month marijuana use increased with higher raw scores on the Delinquent Behavior syndrome. The estimated probability of past month marijuana use was .40 for adolescents with Delinquent Behavior raw scores of 10 (out of 20). The estimated probability for adolescents with raw scores of 15 was .81. For adolescents with the highest Delinquent Behavior raw score, the estimated probability of past month marijuana use was .96. A similar pattern was indicated for other substances, for substance dependence, and for need for treatment. For adolescents with low raw scores for Delinquent Behavior, the estimated probability of need for illicit drug abuse treatment was less than .01. In contrast, for those with the highest raw score (20), the probability of needing treatment for substance abuse was .94.

The relationship between measures of Social Problems and substance use is more complex. The association is negative in each analysis. Adolescents who use substances are less likely to report the behaviors or experiences comprising the Social Problems scale. This relationship does not depend on substance type or severity, or on adolescent age or gender. Substance use is influenced by social factors such as vulnerability to peer pressure and association with older adolescents. Association with other adolescents may provide more opportunities for substance use. The relationship between specific items comprising the Social Problems scale and substance use indicators is considered further in the item-level analysis.

The relation of Attention Problems to substance use is more straightforward. The positive coefficients for this scale indicate that substance use is associated with the presence of Attention Problems. This relationship does not differ by age or gender. For adolescents with low raw scores for Attention Problems, the estimated probability of past month alcohol use was lower. For those with the lowest possible raw score (0), the estimated probability of past month alcohol use was .15. For those with a raw score of 6 for Attention Problems, the probability of past month alcohol use increased to .19 and for those with a raw score of 12 the probability increased to .24. For those with the highest possible raw score for Attention Problems, the probability of past month alcohol use was estimated to be .30. A similar pattern was indicated for past month cigarette smoking. For adolescents with the highestpossible raw score for Attention Problems (18), the estimated probability of smoking was approximately .39 vs. an estimated probability of .12 for those with the lowest raw score for Attention Problems (0).

Associations of other syndromes to substance use were more specific to age and gender, and dependent on the substance use indicator. A negative association with substance use was indicated for the scale measuring withdrawal (Withdrawn). Those adolescents who reported substance use were less likely to report the items comprising this scale. This pattern was most evident for cigarette use. The Somatic Complaints syndrome was associated with alcohol use, alcohol dependence, and substance dependence for females, and with cigarette use for 14- and 15-year-old males. Males aged 12 to 13 who were determined to be in need of substance abuse treatment reported fewer Somatic Complaints.

The Anxiety/Depression scale was significantly related to substance abuse in six of the analyses.

For example, the estimated probability of past month cigarette smoking was estimated to be .15 for those with the lowest possible raw scores for the Anxiety/Depression syndrome. With higher Anxiety/Depression raw scores the estimated probability of past month cigarette use increased. For those with Anxiety/Depression raw scores of 16 (out of 32) the probability of past month cigarette smoking was estimated to be .20. For those with the highest possible raw scores for Anxiety/Depression, the probability of past month smoking was estimated to be .27. The coefficient was positive in five of the analyses-higher Anxiety/Depression scores were associated with elevated scores on substance use indicators. There was no discernible pattern for age, gender, or type of substance. In one instance, however, this relationship was negative-for 14- to 15-year-old males reporting alcohol use.

Similarly, the Thought Problems syndrome was negatively associated with cigarette and alcohol use for 16- to 17-year-old males, although the significance levels were marginal (p<.10). The Aggressive Behavior syndrome scale was positively associated with alcohol dependency (p<.05) and substance dependency (p<.10) for 12- and 13-year-old females. Aggressive Behavior was negatively associated with drug dependency for 16- to 17-year-old males.

Item-level analysis. To study the relationship of syndromes and substance use further, the individual items from the three scales most consistently associated with substance use were examined. Items from the Delinquent Behavior, Social Problems, and Attention Problems scales were used as independent variables in logistic regression analyses to predict use of specific substances, dependency, and need for treatment for the entire sample. Table 4.15 summarizes the results and presents the number of measures of substance use associated with specific items from each of the three syndromes. The table also presents the general direction of the association. For example, the item, "I cut classes or skip school" from the Delinquent Behavior scale was significantly associated with seven substance use indicators.The positive association indicates that adolescents who report truancy are more likely to also report different kinds of substance use, dependence, and need for treatment.

The majority of substance use indicators (cigarette use, alcohol use, marijuana use, alcohol dependence, drug dependence, substance dependence, and need for treatment) were significantly predicted by 10 of the 25 individual items examined. Among these, eight predicted every indicator of substance use.

Of the 10 items most predictive of substance use, five concerned forms of deviant behavior from the Delinquency scale: stealing, swearing, cutting classes, hanging around with troublemakers, and running away from home. Adolescents who reported these behaviors were more likely to use substances. The item, "feeling confused or in a fog," from the Attention Problems scale, was also positively associated with all seven indicators of substance abuse. Another item from the Attention Problems scale, "I have trouble sitting still," was positively associated with three measures depicting a serious substance problem. These were Alcohol Dependent, Substance Dependent, and Need Treatment. Four items included in the Social Problems scales were negatively related to substance use. Adolescents who used substances were less likely to acknowledge getting teased, a preference for younger friends, acting too young, or being overdependent on adults. Nine items were associated with three or fewer substance use indicators, and six items showed no association with any of the substance use indicators.

It should be noted that items appearing in other subscales may significantly predict substance use, especially for specific age-by-gender groups, but further research is required to determine this. The analysis presented here permitted the discovery of items that were most associated with substance abuse in the general sample, from those subscales or constructs that were most predictive of substance use.

---

8 The equation for predicting the probability of membership in a substance use group from a logistic regression model is as follows: P(User)= (e(B₀ +X₁*B₁ +X₂*B₂...+X_k*B_k))/(1+e(B₀ +X₁*B₁ +X₂*B₂...+X_k*B_k)), where B₀ is the intercept term, X₁, X₂ ...X_k are values of the independent variables in the model and B₁, B₂, ... B_k are regression coefficients for each of the independent variables (Hosmer & Lemeshow, 1989; pp. 25-26).

This page was last updated on June 16, 2008.

SAMHSA, an agency in the Department of Health and Human Services, is the Federal Government's lead agency for improving the quality and availability of substance abuse prevention, addiction treatment, and mental health services in the United States.