Page 87 - Age of onset of disruptive behavior of residentially treated adolescents -Sjoukje de Boer

(see Table 3), with odds ratios ranging from 2.28 to 3.02, corresponding with effect sizes ranging of .02 to .07. The dropouts and completers differed neither on types nor in the number of types of disruptive behaviour.
Table 2
Pre-treatment disruptive behaviour characteristics by discharge status
Total Dropout Completer Completer
1 Number of patients for whom information about * p ≤ .05 (two-tailed)
** p < .01 (two-tailed, Holm-Bonferroni correction)
Table 3
(n=224) (n=77) (n=147)
N1 n(%)/M(SD) n(%)/M(SD) n(%)/M(SD) p-value
Cannabis usage (prior to treatment) Disruptive behaviour reported at school Disruptive behaviour (framework Frick)
Aggressive behaviour Oppositional behaviour Status offences Property violations
Number of disruptive behaviours
Early-onset (< age 12) disruptive behaviour
224 197
224 206 213 211 186 195
vs dropout 5 145 (64.7%) 60 (77.9%) 85 (57.8%) .003**
154 (78.2%)
189 (84.4%) 98 (47.6%) 196 (92.0%) 187 (88.6%) 3.1 (0.9) 139 (71.3%)
58 (86.6%)
69 (89.6%) 40 (55.6%) 69 (94.5%) 67 (89.3%)
3.3 (0.8) 59 (84.3%)
96 (73.8%)
120 (81.6%) 58 (43.3%) 127 (90.7%) 120 (88.2%) 3.0 (0.9) 80 (64.0%)
.041*
.118 .093 .331 .810 .058
.003**
the characteristic was available
Logistic regression analysis of the associations between client characteristics and completion and dropout
N OR Early-onset (< age 12) of disruptive behaviour 195 3.02
Model 2
95% CI
Model 1
95% CI R2 OR
(1.44-6.32)* .07 2.93
Cannabis usage (prior to treatment)
Disruptive behaviour reported at school
224 2.57
197 2.28
.06 2.14 (1.02-5.10)* .03
(1.39-6.18)* (1.10-4.16)*
(1.37-4.84)*
Model 1: univariate; model 2: multivariate. Multivariate model: n=195; OR = odds ratio; 95% CI = 95% confidence interval; R2 = Nagelkerke R2
* p <.05; Nagelkerke R2 model 2: 0.10
Binary logistic regression (dropout and completion)
The logistic regression analysis was performed to test the predictive value of the variables on (the dichotomous dependent variable) dropout. The multivariate model showed two significant independent predictors of dropout: early-onset of disruptive behaviour and cannabis usage prior to treatment (Table 3; χ2 (2, N=195) =
85