Background
Investigating outcomes of hospitalised COVID-19 patients throughout the pandemic is crucial to understand the impact of different SARS-CoV-2 variants. We compared 28-day in-hospital mortality of Wild-type, Alpha, Delta, and Omicron variant infections. Whether the difference in risk by variant varied by age was also evaluated.

Methods
We conducted a cohort study including patients ≥18 years, hospitalised between 2020 and 02-01 and 2022-10-15 with a SARS-CoV-2 positive test, from nine countries. Variant was classified based on sequenced viruses or from national public metadata. Mortality was compared using the cumulative incidence function and subdistribution hazard ratios (SHR) adjusted for age, sex, calendar time, and comorbidities. Results were shown age-stratified due to effect measure modification (P < 0.0001 for interaction between age and variant).

Findings
We included 38,585 participants: 19,763 Wild-type, 6387 Alpha, 3640 Delta, and 8795 Omicron. The cumulative incidence of mortality decreased throughout the study period. Among participants ≥70 years, the adjusted SHR (95% confidence interval) for Delta vs. Omicron was 1.66 (1.29–2.13). This estimate was 1.66 (1.17–2.36) for Alpha vs. Omicron, and 1.34 (0.92–1.95) for Wild-type vs. Omicron. These were 1.21 (0.81–1.82), 1.21 (0.68–2.17), and 0.98 (0.53–1.82) among unvaccinated participants. When comparing Omicron sublineages, the aSHR for BA.1 was 1.92 (1.43–2.58) compared to BA.2 and 1.52 (1.11–2.08) compared to BA.5.

Interpretation
The herein observed decrease in in-hospital mortality seems to reflect a combined effect of immunity from vaccinations and previous infections, although differences in virulence between SARS-CoV-2 variants may also have contributed.

Funding
European Union’s Horizon Europe Research and Innovation Programme.

Keywords
SARS-CoV-2COVID-19Variants of concernIn-hospital mortality