A recent study by Kang et al. 1 found that individuals infecte d with the SARS-CoV-2 BA. 5 Omicron sub-lineage exhibited more severe symptoms at the onset of symptomatic disease but had a shorter viable virus shedding period compared to individuals with BA. 1 or BA. 2. However, the study did not elucidate whether these differences reflected any variability in severe outcomes as the disease progressed. Additionally, data on BA. 4 severity compared to previous sub-lineages are limited. Studies have shown that both BA. 4 and BA. 5 have a spike protein mutation with the potential for immune evasion, 2 which may result in increased severity for BA. 4 and BA. 5 compared to previous sublineages. Herein, we used a case-control study to assess relative severity of BA. 4 and BA. 5 compared to BA. 2 by estimating the differences in risk of severe outcomes following presentation to emergency care. While this has been undertaken in other countries with similar demographics, these studies have been conducted in settings with low vaccine coverage. The study included individuals with a PCR confirmed SARS-CoV-2 infection (COVID-19) in England between 1 April 2022 and 1 August 2022 inclusive, where linked whole genome sequencing (WGS) results were available confirming infection with the sublineages BA. 2, BA. 4, or BA. 5, and who had attended an emergency department (ED) between one day before their positive test date and 14 days after their positive test date. Two case-control outcome definitions were considered. Outcome definition 1 included individuals whose ED attendance ended in hospital admission or transfer with a length of stay in hospital of 2 or more days; or whose ED attendance ended in death. They were compared with a control group of individuals whose ED attendance ended with discharge or a hospital admission of less than 2 days’ duration, and who did not die in the 2 days following ED attendance. Outcome definition 2 included COVID-19 patients who attended ED and received oxygen therapy. Controls were those who attended ED but did not receive oxygen therapy. This provided validation for definition 1 and a detailed metric for severity by using oxygen therapy as an indicator, as oxygen therapy has been frequently used in critical cases of COVID-19. 3

Odds ratios (OR) of the outcomes and 95% confidence intervals (CI) were estimated using conditional logistic regression models. The models were stratified for week of positive test, and adjusted for age, sex, vaccination status, prior infection status, socioeconomic deprivation, region of residence, and ED attendances during the July extreme heat event. Further information on data acquisition and statistical analysis can be found in the supplementary document.