Study Design
We used a test-negative case–control design to estimate vaccine effectiveness against symptomatic Covid-19 caused by the omicron variant as compared with the delta variant in persons 18 years of age or older.17 The odds of vaccination in persons with symptomatic, PCR-positive cases of SARS-CoV-2 infection were compared with those in symptomatic persons who tested negative for SARS-CoV-2 in England.
Data Sources
Covid-19 Testing Data
PCR testing for SARS-CoV-2 in England is undertaken by hospital and public health laboratories (Pillar 1) as well as by community testing (Pillar 2). Pillar 2 testing is available to anyone with symptoms consistent with Covid-19 (high temperature, new continuous cough, or loss or change in sense of smell or taste), anyone who is a contact of a person with a confirmed case, care home staff and residents, and persons with a positive rapid lateral-flow antigen test. Lateral-flow tests are freely available to all members of the population for regular home testing. Data on all positive PCR and lateral-flow tests, and on negative Pillar 2 PCR tests from persons with a date of onset of Covid-19 symptoms after November 25, 2020, were extracted up to January 12, 2022 (Fig. S1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). Persons who reported symptoms and were tested in Pillar 2 between November 27, 2021, and January 12, 2022, were included in the analysis.
Any negative tests taken within 7 days after a previous negative test, and any negative tests for which the symptom-onset date was within the 10 days after a previous symptom-onset date for a negative test, were dropped because these probably represented the same episode. Negative tests taken within 21 days before a subsequent positive test were also excluded because chances were high that these were false negatives. Positive and negative tests within 90 days after a previous positive test were also excluded; however, when participants had later positive tests within 14 days after a positive test, preference was given to PCR tests and tests from symptomatic persons. For persons who had more than one negative test, one test was selected at random in the study period. Data were restricted to persons who had reported symptoms and gave a symptom-onset date within the 10 days before testing to account for reduced PCR sensitivity beyond this period in an infection event. Only positive tests with sequencing or genotyping information or information on spike gene (S) target–negative status (indicative of probable omicron infection) were included in the final analysis. A small number of positive tests were excluded when sequencing showed neither the delta nor the omicron variant. Finally, only samples obtained on November 27, 2021, or after were retained for analysis because this corresponded to the period when S target–negative status was predictive of the omicron variant.
Vaccination Data
The National Immunization Management System (NIMS) contains demographic information on all persons residing in England who are registered with a general practice physician in that country and is used to record all Covid-19 vaccinations.29 The NIMS was accessed on January 18, 2022, for dates of vaccination and vaccine manufacturer, sex, date of birth, race or ethnic group, and residential address. Addresses were used to determine the index of multiple deprivation (a national indication of level of deprivation that is based on small geographic areas of residence, assessed in quintiles) and were also linked to Care Quality Commission–registered care homes with the use of the unique property-reference number. Data on geographic region (NHS region), clinical risk-group status, status of being in a clinically extremely vulnerable group, and health and social care worker status were also extracted from the NIMS. Clinical risk groups included a range of chronic conditions as described in the Green Book,30 whereas the clinically extremely vulnerable group included persons who were considered to be at the highest risk for severe Covid-19, including those with immunosuppressed conditions and those with severe respiratory disease.31 Booster doses were identified as a third dose given at least 175 days after a second dose and administered after September 13, 2021. Persons with four or more doses of vaccine, a heterologous primary schedule, or fewer than 19 days between their first dose and second dose were excluded.
Identification of Variants and Assignment to Cases
Sequencing of PCR-positive samples was undertaken through a network of laboratories, including the Wellcome Sanger Institute. Whole-genome sequences were assigned to U.K. Health Security Agency definitions of variants on the basis of mutations.32,33 S target status on PCR testing is an alternative approach for identifying each variant because the omicron variant has been associated with S target–negative results on PCR testing with the TaqPath assay, whereas the delta variant almost always has an S target–positive result.26 Approximately 40% of Pillar 2 community testing in England is carried out by laboratories using the TaqPath assay (Thermo Fisher Scientific). Cases were defined as being due to the delta or omicron variant on the basis of whole-genome sequencing, genotyping, or S target status, with sequencing taking priority, followed by genotyping. When subsequent positive tests within 14 days included sequencing or genotyping information or information on S target–negative status, this information was used to classify the variant. A priori, we considered that S target–negative status would be used to define the omicron variant when the variant accounted for at least 80% of S target–negative cases. Beginning on January 10, 2022, delta cases were identified by sequencing and genotyping only because the positive predictive value of S target–negative status to identify the delta variant had decreased and could no longer be used.
Testing data were linked to the NIMS on January 18, 2021, through combinations of the unique individual NHS number, date of birth, surname, first name, and postal code with the use of deterministic linkage. A total of 91.8% of eligible tests could be linked to the NIMS.
Statistical Analysis
Logistic regression was used, with the PCR test result as the dependent variable and case participants being those testing positive (stratified in separate analyses as being infected with either the omicron or delta variant) and controls being those testing negative. Vaccination status was included as an independent variable, and effectiveness was defined as 1 minus the odds of vaccination in case participants, divided by the odds of vaccination in controls.
Vaccine effectiveness was adjusted in logistic-regression models for age (18 to 89 years in 5-year bands, then everyone ≥90 years), sex, index of multiple deprivation (quintile), race or ethnic group, history of foreign travel, geographic region, period (day of test), health and social care worker status, clinical risk-group status, status of being in a clinically extremely vulnerable group, and previously testing positive. These factors were all considered potential confounders and so were included in all models.
Analyses were stratified according to primary immunization course (ChAdOx1 nCoV-19, BNT162b2, or mRNA-1273 vaccine). Any heterologous primary schedules were excluded.
Vaccine effectiveness was assessed for each primary course in intervals of 2 to 4, 5 to 9, 10 to 14, 15 to 19, 20 to 24, and 25 or more weeks after the second dose. Vaccine effectiveness was assessed at 2 to 4, 5 to 9, and 10 or more weeks after a BNT162b2 or mRNA-1273 booster after a ChAdOx1 nCoV-19 or BNT162b2 primary course. In addition, the ChAdOx1 nCoV-19 booster was assessed after a ChAdOx1 nCoV-19 primary course in these postvaccination intervals. In persons with an mRNA-1273 primary course, vaccine effectiveness was assessed after BNT162b2 or mRNA-1273 booster vaccines after 1 week and after 2 to 4 weeks.