Development and validation of a prediction index for future probable major depression in adolescents.

BACKGROUND

No prediction indices for adolescent probable major depression have been developed and validated among Chinese adolescents.

METHODS

A 12-month cohort study was conducted among Hong Kong adolescents. Data were collected by a structured anonymous questionnaire in school settings. Participants were randomly assigned to develop (50 %) and validate (50 %) the adolescent probable major depression prediction index (AMDPI). The generalized linear mixed model (GLMM) with both forward and backward stepwise selection was used to identify the optimal model. The moderating roles of age and gender were tested. The AMDPI's performance was evaluated by C-index, sensitivity, specificity, positive and negative predictive values. The observed incidence was compared across six risk groups based on the 5th, 25th, 50th, 75th, and 95th percentiles of the AMDPI score.

RESULTS

The AMDPI included background factors of age, gender, the depression score at baseline, and predictors of negative coping, psychological resilience, problem behaviour, academic stress, and family support. Psychological resilience was the strongest predictor (OR (95%CI) = 0.795 (0.710, 0.891). The C-index was 0.72. At the cut-off of 0.30, the AMDPI's sensitivity, specificity, positive and negative predictive values were 0.74, 0.58, 0.46, and 0.82, respectively. The probable major depression incidence in the highest risk group (level 6: 63.0 %) was higher than that in the lowest group (level 1: 8.2 %) in the test sample. Age and gender did not show moderating effects.

CONCLUSIONS

The AMDPI is an effective tool for predicting probable major depression among Chinese adolescents in school settings. It is applicable across different ages and genders.