A total of 16 826 hospitalisations meeting the definition of ARI diagnosis with a valid SARS-CoV-2 laboratory test during the proxy variant periods were included (Fig. 1). The proportion testing positive for SARS-CoV-2 was highest during the delta period (70.9%) and decreased with subsequent waves (63.5% during BA.1 and 41.5% during BA.4/5 period). The proportion unvaccinated was high during delta (94.1% of cases and 85.2% of controls), as this surge occurred early during the vaccine rollout, but decreased during BA.1 (43.7% of cases and 38.8% of controls) and BA.4/5 (35.3% of cases and 28.3% of controls). The number of hospitalisations having received booster doses was low for all three proxy variant periods (0, 8 and 291 individuals during delta, BA.1 and BA.4/5 respectively). These hospitalisations were excluded from further analysis as numbers were insufficient to calculate VE for booster doses.
Of the 16 527 hospitalisations included (excluding those with booster doses), 6669 (40.4%) were between the ages of 18–49, 4260 (25.8%) were 50–59 years old, 2638 (16.0%) were 60–69 years old and 2960 (17.9%) were 70 years or older (Table 1). Age distribution was similar by proxy variant period and cases were older than controls for all periods. The majority of hospitalisations were amongst females, with males making up 36.6% (6047) of overall hospitalisations which was similar by proxy variant period (37.7%, 35.9% and 31.4% males for delta, BA.1 and BA.4/5, respectively). Males made up a higher percentage of cases than controls overall and for all proxy variant periods, except during BA.4/5 when there was no significant difference between proportion of males in cases or controls (p = 0.247). The majority (59.7%) of hospitalisations were in individuals residing outside of the main metropolitan areas. Most hospitalisations (48.8%) were amongst individuals in the lowest income level (< R13000 per month). Only 3.7% (618) of hospitalisations had a previous documented SARS-CoV-2 infection with the proportion being significantly higher for controls than for cases throughout the variant periods (p < 0.001 for delta and BA.1 and p = 0.005 for BA.4/5). A higher proportion of cases compared to controls were unvaccinated for all proxy variant periods (94.1% versus 85.2%, p < 0.001 for delta; 43.8% versus 38.8%, p = 0.003 for BA.1; 40.9% versus 32.9%, p < 0.001 for BA.4/5). Characteristics of included hospitalisations by SARS-CoV-2 vaccination status are provided in supplementary information (Table S1).
BNT162b2 showed a significantly protective VEs against COVID-19-related hospitalisation for all variant periods (89.3% (95% CI, 85.9–91.9) during the delta period, 31.4% (95% CI, 19.1–41.9) during the omicron BA.1 period and 22.7% (95% CI, 2.2–38.9) during the omicron BA.4/5 period) (Fig. 2). VE estimates for Ad26.COV2.S were also significantly protective against COVID-19-related hospitalisation for all proxy variant periods, however the estimates were lower than for BNT162b2 for delta and BA.1 (48.8% (95% CI, 39.6–56.5) and 19.8% (95% CI, 5.8–31.6)) and higher for BA.4/5 (45.0% (95% CI, 29.8–57.0)). Only estimates for vaccination < 3 months ago were calculated for delta period since delta was early during the vaccine rollout. When stratified by time since vaccination, BNT162b2 VE estimates during BA.1 remained significantly protective < 3 months, 3–5 months and ≥ 6 months since last vaccination (VE of 48.3% (95% CI, 29.8–61.9), 17.4% (95%, 0.2–31.7) and 62.1% (95% CI, 46.9–72.9) respectively). During BA.4/5 VE estimate remained significantly protective < 3 months since last vaccination (68.1% (95% CI, 24.3–86.8), however did not remain significantly protective 3–5 months or 6 + months after vaccination (14.6% (95% CI, -41.5-48.4) and 19.9% (95% CI, -3.1-37.7) respectively). For Ad26.COV2.S VE estimates were not significantly protective < 3 month or 3–5 months since vaccination during BA.1 or BA.4/5 (p = 0.345, p = 0.324, p = 0.053, p = 0.455 respectively). This is likely a result of small sample size for these time periods. Ad26.COV2.S was significantly protective after 6 months since last vaccination during BA.1 and BA.4/5 periods (53.0% (95% CI, 40.6–62.8) and 45.4% (95% CI, 29.7–57.5) respectively).
BNT162b2 conferred significant protection against COVID-19-related hospitalisation for 18–59 year olds and for ≥ 60 year olds during the delta period (88.9% (95% CI, 66.0-96.9) and 86.8% (95% CI, 82.4–90.1) respectively) and during the omicron BA.1 period (29.8% (95% CI, 11.4–44.4) and 32.9% (95% CI, 14.5–47.3) respectively; Fig. 3). During the omicron BA.4/5 period, BNT162b2 conferred significant protection against COVID-19-related hospitalisation in those over the age of 60 years (49.1% (95% CI, 26.3–65.0), however in the younger age group this protection was no longer significant (1.9% (95% CI, -35.0-5.3). Ad26.COV2.S conferred significant protection against COVID-19-related hospitalisation for those between 18–59 years during the delta period (50.5% (95% CI, 41.4–58.1)) and during the omicron BA.4/5 period (27.6% (95% CI, 5.3–44.7)), however VE was not significant during the omicron BA.1 period (12.0% (95% CI -4.8-26.2). In the older age group (≥ 60), VE was only significantly protective during the omicron BA.4/5 period (81.3% (95% CI, 57.1–91.9) and was not significant during the delta (-11.5% (95% CI, -227.5-62.1)) or omicron BA.1 periods (33.6% (95% CI, -9.6-59.7)).
BNT162b2 offered significant protection in HIV-uninfected individuals during all three proxy variant periods with the highest VE during delta (89.40%, 95% CI, 85.90–92.00), decreasing during BA.1 (32.20%, 95% CI, 19.30–43.10) and decreasing further during BA.4/5 (23.00%, 95% CI, 1.40–39.80) (Fig. 3). Amongst PLWH, BNT162b2 offered a significantly protective VE during delta which was comparable to the HIV-uninfected group (89.60%, 95% CI, 57.00-97.50), however the VE did not remain significant during the omicron variant periods. Ad26.COV2.S offered a significantly protective VE in HIV-uninfected individuals during delta (48.90%, 95% CI, 37.80–58.00) and a comparatively high VE during BA.4/5 (45.80%, 95% CI, 29.30–58.50). The VE during BA.1 was not significant. Amongst PLWH, Ad26.COV2.S offered a significantly protective VE during delta (42.50%, 95% CI, 21.80–57.70) and BA.1 (44.40%, 95% CI, 18.70–62.00), however the VE was not retained during BA.4/5 (29.10, 95% CI -37.20-63.40).