Study population and setting
This study from Public Health England estimated the effectiveness of the Pfizer-BioNTech and Oxford-AstraZeneca vaccines in preventing infection with the delta variant (B.1.617.2) of SARS-CoV-2 compared to the dominant alpha variant (B.1.1.7) in the UK. The authors used a test negative case-control design to estimate odds ratios for vaccine effectiveness against the two variants, and also compared the proportion of cases with the delta variant by vaccination status. The study considered all positive PCR tests for SARS-CoV-2 among people aged 16 years and older in the UK between October 26, 2020 and May 16, 2021. For the case-control analysis, cases were classified as the delta variant either through sequencing or if they were S-gene target positive on the TaqPath PCR assay; they were identified as the alpha variant via sequencing or if they were S-gene target negative. For the second analysis, only those cases with whole genome sequencing were used (an increasing proportion of cases were sequenced during this period, ranging from 10% in February 2020 to 60% in May 2021). A total of 12,675 sequenced cases were included in the study, and all negative community test results during this period among people presenting with symptoms in the prior ten days (n=99,439) were included as controls. Vaccination status was categorized as unvaccinated, one dose (for symptom onset 21 days or more after first dose), and two doses (14 days or more after second dose). Data from vaccinated individuals, test results, and sequencing were obtained from national databases. Logistic regression was used to estimate odds ratios for vaccine effectiveness in the test negative case control study, adjusted for calendar time, region, and a range of demographic and behavioral covariates.
Summary of Main Findings
In the adjusted case control analysis, after two doses of the Pfizer-BioNTech vaccine, there was a slight reduction in estimated vaccine effectiveness when comparing the alpha variant to the delta variant, from 93.4% (95% CI: 90.4 to 95.5) to 87.9% (78.2 to 93.2). After two doses of the Oxford-AstraZeneca vaccine, estimated vaccine effectiveness was also slightly higher against the alpha variant [66.1% (54.0 to 75.0)] than against the delta variant [59.8% (28.9 to 77.3)]. Estimated effectiveness (for any vaccine) with one dose was lower against the delta variant [33.5% (20.6 to 44.3)] than against the alpha variant [51.1% (47.3 to 54.7)]. In the second analysis, the adjusted odds ratio for infection with the delta variant vs. the alpha variant after one dose was 1.38 (1.10 to 1.72), and after two doses was 1.60 (0.87 to 2.97).
The test negative case-control design can be an effective tool for reducing bias arising from inappropriate selection of controls. The study drew from several linked national databases and was able to control for a wide array of demographic, regional, and temporal variables. The agreement between variant identification from whole genome sequencing and from the TaqPath PCR assay was reasonably high (of those tested with both, 87.5% of S-gene positive cases were correctly identified as delta variant, and 99.7% of S-gene negative cases were correctly identified as alpha variant).
Whole genome sequencing was done for only a subset of positive test results. If there were systematic differences by vaccination status and variant in those that were not sequenced, the results of this study would not be valid. If vaccination affected the probability of seeking care for symptoms consistent with COVID-19, results could be biased (e.g., if infected vaccinated individuals had less severe symptoms, they may be less likely to seek testing, which would inflate apparent vaccine effectiveness). Misclassification of variants is possible, particularly for the delta variant when identified by S-gene positivity, which could have affected study estimates. Finally, unmeasured confounding is a possibility given that this was an observational study.
This is the first study to date that estimates vaccine effectiveness against the SARS-CoV-2 delta variant, which is rapidly rising in prevalence throughout the world.