Prediction of adolescent psychotherapy outcomes using youth- and caregiver-reported symptoms data.

Objective: We used longitudinal youth- and caregiver-reports of adolescent psychological symptoms from three samples of youth receiving mental health services in routine treatment settings to derive expected change trajectories and identify cases at risk for treatment failure.

Method: Participants were 1906 youth (1053 caregivers) receiving treatment in community mental health settings, merged across three samples. The Symptoms and Functioning Severity Scale (SFSS) was used as an indicator of weekly clinical change. Multilevel modeling methods were used to develop expected change trajectories and identify cases at risk for treatment failure (not on track; NOT). Logistic regression was used to predict client improvement as a function of NOT status.

Results: The SFSS was a reliable indicator of therapeutic change according to youth-reported symptoms. Caregiver reports were not as robust. Whereas predictive accuracy of NOT status yielded moderately high sensitivity in detecting improvement according to youth report, caregiver reports were not as predictive.

Conclusions: The youth-reported version of the SFSS-based algorithm seems appropriate for implementation in clinical care. Future studies should search for similarly predictive measures for caregivers.