Decision Tree Algorithm−Based Prediction of Vulnerability to Depressive and Anxiety Symptoms in Caregivers of Children With Glaucoma

Abstract

Development and validation of a decision tree model (DTM) for prediction of mental health status in Korean caregivers of children with glaucoma. Cross-sectional study. Caregivers of children younger than 18 years with diagnosed primary childhood glaucoma (n=42), secondary childhood glaucoma (n=51), and glaucoma suspect (GS; n=36) were prospectively enrolled at Seoul National University Children's Hospital, Seoul, Korea. The participants completed 2 questionnaires, the 9-item Patient Health Questionnaire (PHQ-9) and the 7-item Generalized Anxiety Disorder (GAD-7) Assessment. A DTM analysis for discrimination of those with moderate-to-severe depressive (PHQ-9 score ≥10) and anxiety (GAD-7 score ≥11) symptoms was performed with recursive partitioning algorithms based on the obtained clinical, demographic, and socioeconomic data. The mean PHQ-9 and GAD-7 scores did not significantly differ among the 3 groups (P=.823 for PHQ-9 score; P=.730 for GAD-7 score). The DTM's balanced accuracies were 0.875 (95% CI=0.778-0.972) for the PHQ-9 score and 0.880 (95% CI=0.800-0.960) for theGAD-7 score. The DTM of the PHQ-9 revealed that in caregivers of children with glaucoma, depressive symptoms should be strongly suspected when (1) the child has undergone more than 2 glaucoma surgeries; or (2) the visual acuity (VA, converted to logarithm of minimum angle of resolution [logMAR]) in the better eye is worse than 0.4 if the child has had only 1 or no surgery. Evaluation of the number of glaucoma surgeries and VA in the better eye can be a useful decision support tool in predicting mental illness in caregivers of children with glaucoma.