Internal migration and depressive symptoms: Exploring selection and outcomes in a South African cohort

Abstract

Introduction

Investigations of migration effects on mental health conditions, including depression are sparse in low- and middle-income countries (LMIC), yet mental health may play a role in a decision to migrate, and migration in turn can impact on mental health outcomes.

Methods

This paper uses two waves of data from the Migrant Health Follow-Up Study, a young adult cohort of 2967 internal migrants and residents of the Agincourt study site in rural northeast South Africa to explore the relationship between internal migration and depressive symptoms, as measured on the Center for Epidemiological Studies Depression (CES-D) scale. We employ logistic regression analysis to investigate selectivity of migrants in relation to depressive symptoms, and we fit generalized linear models to analyse depressive symptoms (CES-D scores) as a function of migration status and sociodemographic and health characteristics, accounting for temporal sequence.

Results

Although we observe systematically low reporting of depressive symptoms, average CES-D scores are lower among migrants (comprising approximately 53% of the cohort) compared to Agincourt residents at both survey timepoints. We do not find evidence of a selection effect in relation to mental health among those newly migrating between Wave 2 and 3 (n = 1393). In analyses of the CES-D score outcome, the significant influence of migration status on depressive symptoms is reduced with the inclusion of controls in the models. Consistent employment and higher levels of education are associated with lower CES-D scores, while diagnosis of a chronic condition is associated with higher scores.

Conclusion

The relationship between migration and depressive symptoms is influenced by factors preceding a migration and destination-place characteristics and experiences. Further examination of the role of migration at different stages of the process, along with continuing attention to psychosocial measurement considerations for LMIC subpopulations, can improve our understanding of these complex interrelationships and contribute to evidence.