Abstract
The efficacy of Multisystemic therapy (MST) in treating adolescent aggression has been established, however, not all youth and their families benefit from MST. One reason for this treatment variability could be the failure to distinguish between different aggressive subtypes with different risk factors, developmental prognoses and treatment needs. We investigated whether changes in maternal depression over MST would lead to different outcomes for two aggressive subtypes: pure externalizers (EXT) and mixed externalizers/internalizers (MIXED). Forty-two EXT and 27 MIXED youth and their families underwent MST for six months. Maternal depression, youth externalizing and internalizing behaviour were assessed before and after MST. Results showed a marginally greater change in externalizing for EXT youth. In addition, reductions in maternal depression were related to successful treatment outcomes for MIXED youth only. Our findings have implications for MST clinicians, namely the importance of reducing depressive symptoms in mothers of MIXED youth to improve their outcomes.
Results
Preliminary analyses
Concerning the entire sample, repeated measures ANOVA revealed a significant reduction in externalizing behaviour, F(1, 66) = 207.75, p < .001, partial η2 = .76 from pre- to post-treatment. Results also indicated a significant reduction in maternal depression over the course of treatment, F(1, 68) = 7.92, p < .01, partial η2 = .10.
When we compared the two subtypes on pre-treatment levels of impairment, ANOVAs revealed that MIXED and EXT youth did not differ in externalizing symptoms at pre-treatment, and that, by definition, MIXED youth had significantly more internalizing symptoms than EXT youth at pre-treatment, F(1, 67) = 121.67, p < .001, partial η2 = .65.
While each subtype showed a significant reduction in externalizing behaviour from pre- to post-treatment, there was a trend towards an interaction between Aggressive subtype and pre- to post-externalizing change, F(1, 65) = 2.85, p < .10, partial η2 = .04 indicating a tendency towards greater change for EXT youth (see Fig. 1).
Externalizing behaviour from pre- to post-treatment in MIXED and EXT youth.
Fig. 1.
Externalizing behaviour from pre- to post-treatment in MIXED and EXT youth.
Figure options
Hypothesis 1: testing pre-treatment differences between subtypes on maternal depression
An ANOVA revealed no significant differences between EXT and MIXED on maternal depression at pre-treatment. However, when we classified mothers as depressed or non-depressed, chi-square analysis revealed a significant difference in the proportion of depressed mothers in the MIXED and EXT groups at pre-treatment, χ2(1) = 5.02, p < .05. As expected, mothers of MIXED youth were more likely to be depressed than mothers of EXT youth (see Fig. 2).
Frequency of maternal depression in MIXED and EXT youth.
Fig. 2.
Frequency of maternal depression in MIXED and EXT youth.
Figure options
Hypothesis 2: testing the relationship between changes in maternal depression and changes in externalizing behaviour for the two subtypes
To test our second hypothesis that a reduction in maternal depression would be related to a reduction in externalizing problems for MIXED, but not EXT youth, a moderator analysis was conducted using EXT/MIXED classification as the moderator variable (Aggressive Subtype). Using guidelines outlined in Baron and Kenny (1986), a two-way ANOVA was conducted to investigate whether there was a statistical interaction between Aggressive Subtype (dichotomous moderator) and externalizing change2 on the dependent variable (post-maternal depression, controlling for pre- maternal depression). A significant interaction would indicate that aggressive subtype moderates the relationship between maternal depression change and externalizing change, warranting separate linear regression analyses for each subtype.
The moderator analysis demonstrated a significant interaction effect between Aggressive Subtype and externalizing change on maternal depression change, F(6, 47) = 4.20, p < .01, partial η2 = .35 indicating that the association between a change in maternal depression and externalizing change was different for EXT and MIXED youth.
To examine this association more closely by subtype, separate linear regression analyses were performed. Results showed that, for MIXED youth, externalizing change accounted for an additional 40% of the variance in post-treatment maternal depression, controlling for pre-maternal depression. For EXT youth, externalizing change did not account for any additional variance in post-maternal depression, controlling for pre-maternal depression. These results support our second hypothesis that improvements in maternal depression are related to reductions in externalizing behaviour for MIXED, but not EXT youth. Table 2 presents the separate linear regression results for MIXED and EXT youth.
Table 2.
Linear regression results for externalizing change predicting post-treatment maternal depression in MIXED and EXT youth.
Variable MIXED (n = 42) EXT (n = 27)
B SE B β B SE B β
Model 1
Pre-maternal depression .22 .19 .24 .63 .13 .61**
Model 2
Pre-maternal depression .21 .15 .23 .63 .13 .61**
Pre- to post-externalizing Δ −.06 .02 −.64* .01 .02 .04
R2 (Model 1) .06 .38
F (Model 1) (1,40) 1.36 (1,23) 24.03**
R2 (Model 2) .46 .38
F (Model 2) (1,39) 16.44* (1,22) .11
Note. For externalizing change, a positive score reflects improvement from pre- to post- and a negative score reflects increased impairment. For maternal depression at pre- and post-, higher positive score reflects more impairment. Degrees of freedom for the F-statistic are presented in brackets. **p < .0001; *p < .001, two-tailed.
Table options
Supplementary analysis
Thus far, we have been defining improvements based on externalizing change because the adolescents in our study were initially referred for their externalizing problems. However, it seems important to examine changes in internalizing behaviour over treatment in MIXED youth given their co-occurring internalizing problems. Repeated measures ANOVA showed that internalizing behaviour improved significantly from pre- to post-treatment in MIXED youth, F(1, 23) = 36.49, p < .001, partial η2 = .61. Concerning the relationship between changes in maternal depression and internalizing outcome, linear regression analysis revealed a trend towards an association between a reduction in maternal depression and improvements in youth internalizing behaviour – internalizing change accounted for an additional 12% of the variance in post-maternal depression above and beyond pre-maternal depression, R2 change = .12, F(1, 21) = 3.15, p < .10. However, when externalizing change was entered into the regression model first, internalizing change did not account for any additional variance in post-maternal depression (controlling for pre-maternal depression), R2 change = .01, F(1, 20) = .51, ns.
