Implementation of a sequential screening process for depression in humanitarian crises: a validation study of the Patient Health Questionnaire in Syrian refugees

Abstract

Background: Despite the urgency of mental health surveillance and treatment in humanitarian emergencies, there has been little evaluation of instruments that keep response burden to a minimum. We aimed to evaluate a sequential screening process for the detection of major depressive disorder that improves the efficiency of screening-confirmation procedures using the two-item and eight-item Patient Health Questionnaires (PHQ-2 and PHQ-8, respectively) in a refugee population. Methods: We analysed data from a cross-sectional survey completed by Syrian refugees in a camp in Greece between January and February, 2017. All survey items were translated to and back-translated from Arabic. Face-to-face interviews took place in an area with audio privacy. We used logistic regression to measure the correlation between the PHQ-2 classification and the summed score of the remaining six items of the PHQ-8, and used empirical data to simulate the sequential screening process for major depressive disorder. The sequential process replicates the diagnoses captured if the PHQ-2 were used to narrow the population who would complete the full PHQ-8 assessment. First, respondents complete the PHQ-2; next, only respondents who score above a threshold are considered at risk for depression and complete the remaining six items. We compared the positive and negative percent agreement of the PHQ-2 and sequential screening process. Response burden was assessed through item and instrument completion. Findings: We included data from 135 respondents, representing 40% of the adult population in the refugee camp at the time of the survey. Mean age was 30 years (range 18–61 years); 41% of respondents were women. The PHQ-2 and PHQ-8 instruments were completed by 123 (91%) and 113 (84%) respondents, respectively (p=0·014). The scores of the PHQ-2 and remaining six items of the PHQ-8 were significantly correlated (p<0·001). Using a cut-off score of ≥2, the sequential screening process is 90% sensitive using the PHQ-8 classification as the reference standard with a 100% negative agreement, meaning there were no false negative classifications. The sequential screening process eliminated the completion of the full PHQ-8 instrument for 34 respondents (25%). The PHQ-2 model did not lack a predictor explaining a significant proportion of the variation (deviance information criterion: p=0·395). Interpretation: The sequential screening process detected 90% of people who had symptoms consistent with major depressive disorder. Concurrent validity of the PHQ-2 and PHQ-8 was shown. Our results suggest that this sequential screening approach is a useful strategy to streamline surveillance of major depressive disorder in humanitarian contexts with a scarcity of mental health specialists while simultaneously reducing the response burden. Funding: None.