Investigating the salmon bias effect among international immigrants in Sweden: a register-based open cohort study

BACKGROUND: Studies of migration and health have hypothesized that immigrants may emigrate when they develop poor health (salmon bias effect), which may partially explain the mortality advantage observed among immigrants in high-income countries. We evaluated the salmon bias effect by comparing the health of immigrants in Sweden who emigrated with those who remained, while also exploring potential variation by macro-economic conditions, duration of residence and region of origin. METHODS: A longitudinal, open cohort study design was used to assess risk of emigration between 1992 and 2016 among all adult (18+ years) foreign-born persons who immigrated to Sweden between 1965 and 2012 (n = 1 765 459). The Charlson Comorbidity Index was used to measure health status, using information on hospitalizations from the Swedish National Patient Register. Poisson regression models were used to estimate incidence rate ratios (RRs) with 95% confidence intervals (CIs) for emigrating from Sweden. RESULTS: Immigrants with low (RR = 0.83; 95% CI: 0.76–0.90) moderate (RR = 0.70; 95% CI: 0.62–0.80) and high (RR = 0.62; 95% CI: 0.48–0.82) levels of comorbidities had decreased risk of emigration relative to those with no comorbidities. There was no evidence of variation by health status in emigration during periods of economic recession or by duration of residence. Individuals with low to moderate levels of comorbidities from some regions of origin had an increased risk of emigration relative to those with no comorbidities. CONCLUSIONS: The study results do not support the existence of a salmon bias effect as a universal phenomenon among international immigrants in Sweden.