Study population and setting
This retrospective cohort study analyzed individual-level Swedish register data combined from various state agencies on 7,943,843 individuals aged 20 years and above on March 12, 2020, including 3,126 COVID-19 deaths (all recorded COVID-19 deaths in Sweden up to May 7, out of 17,181 total deaths) between March 13 and May 7, 2020. The study examined sociodemographic risk factors associated with the risk of COVID-19 death for the entire population of Sweden, including age, sex, civil status, individual disposable income, region of residence, and country of birth. Cox proportional hazard regression models were estimated to obtain hazard ratios for how the risk of death from COVID-19 varied across those characteristics. The gradients in mortality across risk factors for COVID-19 mortality were also compared with those observed for all other causes of death.
Summary of Main Findings
The study found that for men and women alike, never married, divorced, and widowed individuals experience 1.5–2 times higher mortality from COVID-19 than those who were married. Both men (HR: 1.24; 95% CI: 1.07, 1.43) and women (HR: 1.51; 95% CI: 1.28, 1.79) with primary education experienced higher mortality than individuals with postsecondary education, and those with secondary education experienced 25% and 38% higher mortality, respectively, again relative to those with postsecondary education. Individuals in the lowest third of individual net income also experienced higher mortality than those in the highest income third. Further, immigrants from low- and middle-income countries from the Middle East and Northern Africa displayed more than three times higher mortality among men and two times higher among women as compared to those born in Sweden. Further, those living in Stockholm county had 4.5 times higher mortality than those living outside Stockholm.
A significant strength is the complete coverage of the total population and all deaths in Sweden, from both COVID-19 and other causes, eliminating any selection bias, and the high quality of the individual level socio-demographic data, which allows for robust comparisons.
It is likely that not all COVID-19 deaths were confirmed, and thus they may have been excluded from COVID-19 mortality, and mortality from other causes may also have been increased due to collateral impacts of COVID-19. Consequently, both the true mortality of COVID-19 and the differences in patterns between COVID-19 mortality and all-cause mortality may be underestimated. The study had a relatively short follow-up period of slightly less than two months.
The study stands out in its comprehensiveness, and in providing high-quality microlevel data linking death records to data on sociodemographic background characteristics associated with variation in the risk of death. The findings are useful for countries weighing COVID-19 strategies and most at-risk populations. The finding of elevated mortality among immigrants from low- and middle-income countries is notable in that it deviates from findings on other causes of death.
This review was posted on: 16 November 2020