Do healthy people migrate more? A 21-year follow-up of a rural cohort in Bangladesh.

Demographers have long suspected that health influences whether a person migrates-a "healthy migrant effect" -but this has rarely been tested for a longer period with high-quality, longitudinal data. This study aimed to assess which measures of health are associated with subsequent migrations among young adults from a rural community in Bangladesh, adjusted for socio-demographic characteristics, and how long these associations persist. The 1996 Matlab Health and Socioeconomic Survey (MHSS) characterized health (by self-reported chronic, and acute morbidity symptoms in the past 12 months and one month, respectively, and self-rated health status) of adults within the Matlab Health and Demographic Surveillance System (HDSS) cohort. Analyses included 3756 (M = 1,496, F = 2260) adults aged 18-34 years (the age when migration peaks) to study the effect of health on migration. Cox Proportional Hazards models were estimated to describe associations between health status and subsequent out-migration in 1996-2017, controlling for age, sex, education, religious affiliation, and household asset quintiles. Discrete-time logistic models were estimated to assess the sustained effects of health status measures on out-migrations. Results reveal that self-reported chronic morbidity, neither acute morbidity nor self-rated health status, inhibited subsequent migration. More reported chronic morbidity symptoms were associated with a lower migration (hazard ratio, HR = 0.82, CI = 0.74-0.92 for one symptom and HR = 0.73, CI = 0.63-0.84 for ≥2 symptoms relative to no symptoms). The differences diminished but persisted over time. Socio-demographic variables inhibiting migration were female sex, older age, and lower-level education. In conclusion, healthy young rural adults were more likely to migrate than their counterparts with symptoms of chronic morbidity, and the effect of chronic morbidity on subsequent migrations waned but not eliminated over time.