Background
The COVID-19 pandemic has had several phases with varying characteristics. We aimed to compare severity outcomes in different phases in a population with limited bias from risk factors.
Methods
In a nationwide observational study of all unvaccinated first-time COVID-19 test positive individuals in Sweden aged 18–64 years without comorbidity, from week 45 of 2020 to week 5 of 2022, variant periods with certain variants constituting ≥ 92 % of all whole genome-sequenced cases nationwide, were compared regarding hospitalisation (with main discharge code of COVID-19), severe illness (use of high-flow nasal oxygen or admission to intensive care unit), and death due to COVID-19. Logistic regression was used to estimate odds ratios (ORs) for comparison of these outcomes between variant periods, using adjustments for variant period, age, sex, country of birth, place of residence, income, and education.
Findings
The study included 789,133 individuals, including 15,145 hospitalised individuals. Among all individuals, the adjusted ORs for hospitalisation were 1.7 for the Alpha period vs the Pre-variant period (week 45–52 2020), 1.8 for the Delta period vs the Alpha period, and 0.1 for the Omicron period vs the Delta period (all comparisons significant). Among hospitalised individuals, the adjusted ORs for severe illness were 1.4 for the Alpha period vs the Pre-variant period, 1.7 for the Delta period vs the Alpha period, and 0.5 for the Omicron period vs the Delta period (all comparisons significant), and the adjusted ORs for death were 1.1 for the Alpha period vs the Pre-variant period (non-significant), 1.8 for the Delta period vs the Alpha period (significant), and 0.1 for the Omicron period vs the Delta period (non-significant).
Interpretation
In this population with limited bias from risk factors, vaccination, and previous infection, disease severity increased from the pre-variant to the Delta period and then decreased with the Omicron period, among all individuals and among hospitalised individuals. These severity outcome differences should be considered when the pandemic is evaluated.