Testing the EPDS and PHQ-9 Among Peripartum Adolescents in Nairobi, Kenya: a Mixed-Methods Study

Background: Cognitive interviewing is a well-recognised methodology to identify clinical and commonsensical relevance of mental health questionnaire items by our research participants. Depression is amongst the most common condition impacting pregnant and parenting adolescents in sub-Saharan Africa (SSA). In Kenya, studies have reported depression prevalence estimates of 12-50% in peripartum adolescents. While young people prefer using English, there has not been enough data to point to how well they respond to Kiswahili translations of the commonly used tools. Method: Thirty-two participants between ages 14-18 years were approached and through informed consent for them to participate. We used Edinburgh Postnatal Depression Scale (EPDS) and Patient Health Questionnaire-9 (PHQ-9) in English and Kiswahili versions to carry out the interviews and were coded along with 4 domains of comprehension, retrieval, judgement, and patient response. The interviews were discussed within the team and tables were shared between 3 raters for further consensus. The interview transcripts were analyzed manually using thematic analysis. We used the findings from the interviews to make modifications to the existing English versions of EPDS and PHQ-9 and to modify Kiswahili versions too. Discussion: We found that adolescents had challenges in fully understanding items in both sets of tools however, EPDS was better received than PHQ-9. Psychometrically, the EPDS English version and Kiswahili versions fared better than the PHQ-9 English and Kiswahili versions. PHQ-9 presented considerable issues with regards to semantic clarity however had simpler response options in comparison to EPDS which was more experientially appropriate, but response options were not simple. On our thematic analysis, we felt that the adolescents were significantly challenged by the new, unanticipated pregnancy or motherhood experience. Poverty, poor partner support, discord with parents and distress in the family, and traumatic end of supportive and helpful relationships were the notable problems. We believe the cultural and linguistic modifications made on these tools would make them more suitable to be validated with a sample of pregnant adolescents. Conclusion: While sensitive cross-cultural translations for the commonly used open access depression tools is critical, it is also pertinent to understand whether these are developmentally appropriate. Our participants had multiple psychosocial and material challenges that necessitate measures sensitive to their age, social context, and health experiences. Additionally, we felt that both the tools may not be suitable for self-administration and may need the involvement of a case worker or community health worker.