Comparative effectiveness of BNT162b2 and ChAdOx1 nCoV-19 vaccines against COVID-19

Background Both BNT162b2 (Pfizer–BioNTech) and ChAdOx1 nCoV-19 (Oxford–AstraZeneca) vaccines have shown high efficacy against COVID-19 in randomized controlled trials. However, their comparative effectiveness against COVID-19 is unclear in the real world. We evaluated the comparative effectiveness of the BNT162b2 and ChAdOx1 nCoV-19 vaccines against COVID-19 in the UK general population. Methods We emulated a target trial using IQVIA Medical Research Database (IMRD), an electronic primary care database from the UK (2021). We included 1,311,075 participants, consisting of 637,549 men and 673,526 women age≥18 years, who received vaccination with BNT162b2 or ChAdOx1 nCoV-19 between January 1 and August 31, 2021. The outcomes consisted of confirmed diagnosis of SARS-CoV-2 infection, hospitalisation for COVID-19 and death from COVID-19 in the IMRD. We performed a cox-proportional hazard model to compare the risk of each outcome variable between the two vaccines adjusting for potential confounders with time-stratified overlap weighting of propensity score (PS). Results During a mean of 6.7 months of follow-up, 20,070 confirmed SARS-CoV-2 infection occurred in individuals who received BNT162b2 vaccine (PS weighted incidence rate: 3.65 per 1000 person-months), and 31,611 SARS-CoV-2 infection occurred in those who received ChAdOx1 nCoV-19 vaccine (PS weighted incidence rate: 5.25 per 1000 person-months). The time-stratified PS weighted rate difference of SARS-CoV-2 infection for BNT162b2 group vs. ChAdOx1 nCoV-19 group was -1.60 per 1000 person-months (95% confidence interval [CI]: -1.76 to -1.43 per 1000 person-months), and the hazard ratio was 0.69 (95% CI: 0.68 to 0.71). The results were similar across the stratum of sex, age (<65 and ≥65 years), and study periods (i.e., alpha-variant predominance period and delta-variant predominance period). The PS weighted incidence of hospitalisation for COVID-19 was also lower in the BNT162b2 vaccine group than that in the ChAdOx1 vaccine group (RD: -0.09, 95%CI: -0.13 to -0.05 per 1000 person-months; HR: 0.65, 95%CI: 0.57 to 0.74). No significant difference in the risk of death from COVID-19 was observed between the two comparison groups. Conclusions In this population-based study, the BNT162b2 vaccine appears to be more efficacious than the ChAdOx1 nCoV-19 vaccine against SARS-CoV-2 infection and hospitalisation for COVID-19 but not death from COVID-19.