Abstract
The present research assessed the psychometric properties of the Patient Health Questionnaire-9 (PHQ-9) through an examination of its internal structure, invariance analysis, and standardization. Social distribution analyses of the measure were conducted using linear and binomial logistic regression. The sample consisted of 10,069 adults from all 27 states in Brazil. The data were obtained through four collections across different years of the COVID-19 pandemic (2020, 2021, 2022, and 2023), using independent samples. Confirmatory Factor Analysis (CFA) indicated that the measure is unidimensional with satisfactory fit indices. The model was invariant in relation to the variables investigated at four different levels (configural, metric, scalar, and strict). The standardization supported hypothetical cut scores indicating the severity of depressive symptoms, categorized as very low (0 to 6), low (7 to 13), moderate (14 to 19), high (20 to 23), and very high (≥ 24). We found that sex/gender, skin color/ethnicity, age, education level, and year of the pandemic were predictors of depressive symptoms in the adjusted linear regression analysis. The logistic regression showed variables with higher chances for a positive screening diagnosis of depression, with adjusted Odds Ratio as follows: years 2021 (OR(adj) = 1.275) and 2023 (OR(adj) = 1.409), women (OR(adj) = 1.900), Pardos individuals (OR(adj) = 1.252), education up to high school (OR(adj) = 1.272), being a northeast region resident (OR(adj) = 2.127), and younger people (OR(adj) = 1.716). The findings of this study indicate the suitability of the PHQ-9 for assessing depression in the population and recommend its use for monitoring depressive symptoms in the coming years in Brazil. Clinical implications include developing interventions to address the psychological impact of this and any future health crises.