n = 184,566 adults who were at risk of Omicron BQ.1/CH.1.1/XBB.1.5 reinfections were included in the models. Time from previous infection was a time-updated covariate categorised as 120–180, 180–240, 240–300, 300–360, 360–420, 420–480, 480–540, 540–600, 600–660, 660–720, >720 days and its effect modelled as a trend over these categories (see Supplementary Fig. 6 for categorical effects). Risk is presented versus a reference category of 120–180 days from an infection in the wave starting ~6 months before the current wave (BA.2 for BQ.1/CH.1.1/XBB.1.5 waves). Line type and width represent the sequence of variants for better comparisons across waves (thick solid line represents the previous variant, thin solid line represents the penultimate variant, and dashed lines represent earlier variants). The 95% CIs are calculated as exponent of estimates ± 1.96 × standard error of the estimates. Adjusted (Supplementary Table 3 ) for time-fixed covariates age, sex, ethnicity, reporting working in healthcare, reporting having a long-term health condition, deprivation percentile, infection variant, region, number of previous infections, symptoms in most recent infection whether any previous infection had Ct < 30 or was LFD positive; and time-updated time from most recent vaccination (Fig. 3 ) and background infection prevalence. Results remain similar in sensitivity analyses without adjustment for background infection prevalence (Supplementary Table 4 ).
