Fig. 3: Relative risk of hospital attendance, admission or death, by vaccination status and reinfection status.

Risk of hospitalisation and mortality by vaccination status and reinfection status, for COVID-19 cases with Omicron lineage BA.2 compared to BA.1 in England, 01 December 2021–2025 March 2022. The central measures are adjusted hazard ratios and the error bars are the corresponding 95% confidence intervals (CI) from Cox regression models with an interaction term between variant (BA.2 vs BA.1), and vaccination (A total n = 1,243,212; unvaccinated or ≤28 days after vaccination = 247,748; ≥28 days after first dose = 64,694; ≥14 days after second dose = 294,071; ≥14 days after third dose = 537,822) or reinfection (B total n = 1,243,212; reinfection = 125,239; first infection = 1,117,973) status. The models were stratified for exact specimen date, area of residence, age group and vaccination status, and additionally using regression adjustments for within-age-group residual differences in exact age, sex, ethnicity, index of multiple deprivation (IMD) quintile and within-IMD-quintile residual differences in exact IMD rank, and reinfection status. LRT P values from two-sided tests for interaction between vaccination status and variant status for attendance, admission, and death are 0.65, 0.87, and 0.11 respectively. LRT P values from two-sided tests for interaction between reinfection status and variant status for attendance, admission, and death are 0.34, 0.70, and 0.16, respectively. These explorative tests were not adjusted for multiple comparisons. The adjusted hazard ratio estimates and 95% confidence intervals that the figure is based on are included in Table S4.