# Psychometric validation of the Patient Health Questionnaire-9 in Chinese adolescent and adult psychiatric inpatient populations

## TL;DR

This study validates the PHQ-9 depression screening tool for Chinese psychiatric inpatients, finding age-specific cutoffs and highlighting differences between adolescents and adults.

## Contribution

The study provides the first psychometric validation of the Chinese PHQ-9 in psychiatric inpatient populations and identifies age-specific diagnostic cutoffs.

## Key findings

- The PHQ-9 showed good internal consistency in both adolescents (α=0.876) and adults (α=0.883).
- Age-specific optimal cutoff scores were identified: 15.5 for adolescents and 14.5 for adults.
- The two-factor model was preferred for adults, but not for adolescents due to small sample size limitations.

## Abstract

Depressive disorder represents a major public health burden globally, yet the validity of the Patient Health Questionnaire-9 (PHQ-9)—a widely used depression screening tool—remains underexplored in Chinese psychiatric inpatient populations, particularly in age-stratified analyses. This study aimed to (1) validate the Chinese version of the PHQ-9 in Chinese psychiatric inpatients (contrasting with community-based findings) and (2) compare its psychometric properties between adolescent and adult inpatients.

This cross-sectional study enrolled 485 psychiatric inpatients (including 105 adolescents) from Shanxi Bethune Hospital. Participants completed the Chinese version of the PHQ-9. Analyses encompassed confirmatory factor analysis (CFA), Gaussian Graphical Model-based network analysis, and receiver operating characteristic (ROC) curve analysis to determine optimal diagnostic cutoff scores.

Results showed the PHQ-9 had good internal consistency: Cronbach’s α = 0.876 (adolescents) and 0.883 (adults). CFA revealed no significant difference in fit between the unidimensional and two-factor (cognitive-affective vs. somatic) models in adolescents (Δχ²=0.79, p=0.374), with both models showing marginal fit (likely affected by small sample size). In adults, the two-factor model was preferred (Δχ²=6.49, p=0.011). The Network Comparison Test found no significant differences in network structure (M = 0.211, p=0.598) or global strength (S = 0.262, p=0.186) between age groups, but the adolescent network had poor stability (correlation stability coefficient = 0), limiting interpretation. ROC analysis identified age-specific optimal cutoffs exceeding the conventional threshold of 10: 15.5 for adolescents (sensitivity=0.84, specificity=0.47) and 14.5 for adults (sensitivity=0.79, specificity=0.66). Notably, 64.7% of the total sample scored ≥15 on the PHQ-9, while only 43.7% had a primary diagnosis of depressive disorder (ICD-11 6A7), indicating comorbid depressive symptoms contributed to higher cutoffs.

The findings of this study validate the structural validity and diagnostic validity of the PHQ-9 among Chinese adult psychiatric inpatients, while emphasizing that the interpretation of its factor structure in the adolescent population requires caution. The age-related symptom topological patterns indicated by network analyses are highly likely to be influenced by the insufficient size of the adolescent sample and need to be confirmed by subsequent studies. The results of the ROC curve highlight the clinical significance of formulating population-specific diagnostic cutoffs; however, the impact of comorbidity on the findings of this study must be taken into consideration.