Validation of French PHQ-4 and PC-PTSD-5 Screening Scales

A target of 10 patients per item of the tested scales was set. The PHQ-4 includes four items and the PC-PTSD-5 includes five items, corresponding to required sample sizes of 40 and 50 patients, respectively. Although the same patient sample completes both scales, these sample sizes were combined to account for potential loss to follow-up and to optimize statistical power, resulting in a total inclusion target of 90 patients. Descriptive analyses of sociodemographic and clinical characteristics will be conducted for all included patients. Quantitative variables will be summarized using means and standard deviations, and qualitative variables using frequencies and percentages. To address the primary objective, convergent validity will be assessed using Pearson's correlation coefficients between scores obtained on the tested scales (PHQ-4 and PC-PTSD-5) and those obtained on their respective gold standard instruments (PHQ-9, GAD-7, and PCL-5). Convergent validity will also be examined by inclusion center through comparison of correlation coefficients across centers. Secondary objectives will be addressed through: * Exploratory factor analysis to examine the dimensional structure of each scale and to compare it with that of the corresponding gold standard. * Assessment of internal consistency using Cronbach's alpha coefficient. Test-retest reliability will be evaluated using the intraclass correlation coefficient on a subsample of at least 20 participants, depending on the final total sample size, who will complete the PHQ-4 and PC-PTSD-5 scales twice. * Evaluation of diagnostic performance by calculating sensitivity and specificity for each scale among patients undergoing a psychiatric diagnostic assessment, with receiver operating characteristic (ROC) curves generated for each scale. * Assessment of feasibility based on the time required to complete both scales. All statistical tests will be two-sided and considered statistically significant at an alpha level of 0.05. No statistical imputation methods will be used for the management of missing data.