Hester van Eeren

Comparative effectiveness of MST and FFT | 5 105 | Table 2. Variance ratio and 5-number summary of continuous covariates after PS application in full sample (N = 697) Variance ratio ‡ Minimum 25th percentile Median 75th percentile Maximum Age FFT 0.79 12.10 14.76 15.95 16.79 20.39 MST 11.07 14.80 15.83 16.72 18.34 CBCL Internalizing problems FFT 0.96 33.00 55.00 61.00 68.00 88.00 MST 33.00 55.00 62.00 69.00 82.00 Externalizing problems FFT 0.92 34.00 61.00 70.00 74.00 92.00 MST 34.00 63.00 69.00 75.00 88.00 Total behavioral problems FFT 0.87 24.00 60.00 68.28 71.00 85.00 MST 27.00 60.00 67.00 72.00 83.00 YSR Internalizing problems FFT 0.97 30.00 44.00 54.00 61.00 83.00 MST 27.00 44.00 50.00 58.00 85.00 Externalizing problems FFT 1.02 29.00 52.00 59.00 66.00 80.00 MST 29.00 51.00 58.00 66.00 93.00 Total behavioral problems FFT 1.06 28.00 47.00 56.00 62.00 77.00 MST 26.00 46.00 54.00 62.00 82.00 Parenting stress FFT 1.21 -1.40 0.61 1.98 3.34 7.78 MST -1.52 0.45 1.92 3.42 8.95 ‡ In the weighted sample the 2.5th and 97.5th percentiles of the F-distribution are 0.78 and 1.22 respectively. Repeating the analyses in the complete case sample of 361 adolescents and their families also showed no difference in externalizing problem behavior (CBCL: -0.16; 95% CI -4.32 – 4.14, YSR: -0.56; 95% CI -3.79 – 2.68), with, again, a small effect size of d = 0.01 and d = 0.05, respectively. While there was no difference in this sample between the treatment groups concerning the proportion of youth living at home and the proportion of youth engaged in school or work after treatment, it was significantly more likely that adolescents assigned to MST had had police contact during treatment than those assigned to FFT (RR 2.40; 95% CI 1.26 – 5.94, RD 15.3 %; 95% CI 4.7% - 24.5%) (Table 3). After this, covariates selected for the PS in the outcome regression model were added in the complete case sample to overcome possible misspecifications of the model. All covariates were selected except for level of education of the adolescent, level of education and employment of the primary caregiver, and internalizing behavioral problems reported by adolescents in a parsimonious model on the outcome. This model can be interpreted as an additional check on the results found earlier. Again, it showed no difference between the effect of MST and FFT on externalizing problem behavior (CBCL: -0.56; 95% CI -3.06 – 1.49, YSR: 0.35; 95% CI -1.58 – 2.31) (Table 3).