Investigating Salmon Bias among International Immigrants in Sweden. A Register-Based Cohort Study

Andrea Dunlavy, Stockholm University
Agneta Cederstöm, Stockholm University
Srinivasa Vittal Katikireddi, University of Glasgow
Mikael Rostila, Stockholm University
Sol P. Juárez , Stockholm University/Karolinska Institute

Migrants generally have lower mortality rates than host populations in high-income countries, with debate focusing on the extent to which this can be explained by unrecorded and unhealthy selective outmigration (salmon bias hypothesis). However, scarce research to date has been able to evaluate these issues for international immigrants. Applying cox regression models to Swedish register data, we aim to assess whether there is evidence of selective return migration, accounting for possible unrecorded outmigration. The study population consists of all foreign-born persons living in Sweden who immigrated between 1970 and 2001 and were of working age (defined as 18-65 years old) during the follow-up period (1990-2001). The primary outcome was first out-migration from Sweden, defined as a recorded date of migration by the authorities, or proxy measured when sources of income (e.g., job-earnings or social benefits) were not registered for two consecutive years. To assess health selection, we use the Charlson index of co-morbidity estimated during two years before the start of follow-up. Preliminary results show that, contrary to the salmon bias hypothesis, emigration was significantly lower among immigrants who experienced ill-health (HR: 0.79; 95%CI:0.74-0.83) relative to those who did not.

See extended abstract

 Presented in Session 68. Out-Migration: Measures, Causes and Consequences