OBJECTIVETo investigate the measurement invariance and the score distribution characters of the 9-item Patient Health Questionnaire （PHQ-9）， and to test its feasibility in comparing depression among adolescents of different sexes and ages.
METHODSFrom September 2018 to May 2019， 12 241 adolescents aged 10 to 19 years old across 22 primary and secondary schools in Hangzhou were included based on cluster sampling method， and they were investigated by PHQ-9. The measurement invariance was tested by multi-group confirmatory factor analysis. Poisson regression analysis and Logistic regression analysis were adopted to explore the score distribution of PHQ-9 by sex and age.
RESULTSThe configural， metric， scalar and strict invariances （χ2=2 492.527， df=79， P<0.01， RMSEA=0.071， SRMR=0.031， CFI=0.942， TLI=0.947， ΔCFI=0.004， ΔTLI=0.002 compared with the scalar model） of PHQ-9 across sex were all acceptable， and the configural， metric and scalar invariances （χ2=3 311.991， df=414， P<0.01， RMSEA=0.076， SRMR=0.055， CFI=0.928， TLI=0.937， ΔCFI=0.010， ΔTLI=0.002 compared with the metric model） across age were acceptable. The detection rate of mild depressive symptoms in girls was higher than that in boys （52.99% vs. 46.65%， χ2=48.344， P<0.01）. The PHQ-9 total scores of girls aged 13 to 17 were higher than those of boys （D=0.092~0.144， P<0.01）. There was an interaction between sex and age （χ2=32.800， df=1， P<0.01， OR=0.895）. Except low self-evaluation and suicide or self-injury， the detection rate of assessed symptom on each item of girls was higher than that of boys （P<0.05）.
CONCLUSIONIt is discovered that PHQ-9 has measurement equivalence across sex and age in adolescents， and girls and older adolescents scored higher in PHQ-9.