Our study population included health care personnel who had been tested for SARS-CoV-2. Participants were enrolled from December 28, 2020 (2 weeks after the introduction of a Covid-19 vaccine), through May 19, 2021, at 33 sites across 25 U.S. states, representing more than 500,000 health care personnel (Table S1 in the Supplementary Appendix, available with the full text of this article at NEJM.org). The majority (68%) of the participating facilities were acute care hospitals (with or without affiliated outpatient and urgent care clinics), and 32% were long-term care facilities. Covid-19 vaccines were introduced at the participating facilities in December 2020, and the vaccine coverage among health care personnel at these facilities reached 55 to 98% for the receipt of at least one dose of vaccine and 51 to 94% for the receipt of two vaccine doses during the study period.
The study protocol was reviewed by the Centers for Disease Control and Prevention and the institutional review board at each participating medical center and was conducted in accordance with federal laws and institutional policies. The authors vouch for the accuracy and completeness of the data reported and for the fidelity of the study to the protocol.
We conducted a test-negative case–control study involving health care personnel, a group that comprised all paid and unpaid health care personnel with the potential for direct exposure to patients or the potential for indirect exposure to infectious materials at the workplace.13 Testing for SARS-CoV-2 was based on occupational health practices at each facility and was leveraged to identify cases and controls for this study. Case participants were defined as health care personnel who had at least one Covid-19–like symptom and a positive result for SARS-CoV-2 on polymerase-chain-reaction (PCR) testing, other nucleic acid amplification testing, or antigen-based testing.14 The index test date (date that the specimen was obtained) for cases was the first SARS-CoV-2–positive test for the episode of Covid-19–like illness for which case participants were enrolled. The illness was defined as symptomatic if the participant had at least one of the following symptoms present within 14 days before or after the index test date: fever (a body temperature documented at ≥38°C or subjective fever), chills, cough (dry or productive), shortness of breath, chest pain or tightness, fatigue or malaise, sore throat, headache, runny nose, congestion, muscle aches, nausea or vomiting, diarrhea, abdominal pain, altered sense of smell or taste, loss of appetite, or red or bruised toes or feet.
Persons who tested negative on PCR or other laboratory-based nucleic acid amplification testing, regardless of symptoms, were eligible for inclusion as controls. Control participants were matched to case participants according to site of enrollment and week of test date. Within any given week and study site, any participants who tested positive for SARS-CoV-2 (cases) and those who tested negative (controls) and agreed to complete a survey or to be interviewed were matched, with a target ratio of three controls per case. Persons with previous infection, defined as a positive SARS-CoV-2 test (on PCR or antigen testing) that had occurred more than 60 days before the index test date, were excluded.
Information on the participants’ demographic characteristics, symptoms of Covid-19–like illness, underlying conditions and risk factors associated with severe Covid-19,15 and medical care received was collected by means of interviews or participant-completed surveys. The interviews and surveys also included information on potential confounders related to workplace and community behaviors. Medical records were reviewed in order to collect information about the SARS-CoV-2 test, including the date, test type, and result, and about the medical care sought during the Covid-19–like illness. Information on Covid-19 vaccination dates and products received was obtained from occupational health clinics, vaccine cards, state registries, or medical records.
Vaccination status of the participants was determined at the time of their SARS-CoV-2 test date. Participants were considered to be unvaccinated if they had not received any dose of Covid-19 vaccine as of the test date. We defined the interval from days 0 through 13 after receipt of the first dose as the time before effectiveness from a single dose is expected. We further stratified this interval to evaluate for a potential early effect of the first dose by measuring vaccine effectiveness at 0 to 9 days and at 10 to 13 days after receipt of the first dose, on the basis of the cutoff when vaccine effectiveness after the first dose was measured both in this study and in clinical trials.1,7
The effectiveness of a single vaccine dose was measured from 14 days after receipt of the first dose through 6 days after receipt of the second dose (partially vaccinated). We conducted a sensitivity analysis to evaluate the effectiveness of a single vaccine dose before receipt of the second dose to exclude potential early effects after receipt of the second dose. In an additional sensitivity analysis that evaluated the potential influence of vaccine-related reactions leading to the testing of health care personnel, we excluded participants who had been tested within 0 to 2 days after receipt of the second dose. The effectiveness of two doses of vaccine was measured at 7 days or more after receipt of the second dose (complete vaccination), which was consistent with the Pfizer–BioNTech clinical trial.7 In a sensitivity analysis, we also evaluated the effectiveness of two doses of vaccine at 14 days or more after receipt of the second dose, which was consistent with the Moderna trial.8
We used conditional logistic regression to estimate vaccine effectiveness as 1 minus the matched odds ratio (×100%) for partial vaccination or complete vaccination as compared with no vaccination. We evaluated the influence of age, race and ethnic group, presence of underlying medical conditions or risk factors for severe Covid-19, and other factors related to community and workplace behaviors, such as the use of personal protective equipment and receipt of influenza vaccine during the current respiratory season, as potential confounders for vaccine effectiveness by including each variable with vaccination status in the model and then retaining variables that resulted in a change of more than 10% in the model estimate for vaccination status.
In the final model, we adjusted for age, race and ethnic group, presence of at least one underlying condition or risk factor for severe Covid-19, and close contact with patients with Covid-19 in the workplace or with persons with Covid-19 outside the workplace. We evaluated vaccine effectiveness according to vaccine product and in subgroups defined according to participants’ age (<50 years or ≥50 years), race and ethnic group, presence of underlying conditions, health care job categories, and clinical case definitions that were consistent with those used in the clinical trials. We examined the adjusted vaccine effectiveness according to 2-week intervals of follow-up after receipt of the second dose (as compared with unvaccinated participants) to assess for waning of vaccine effect. All the statistical analyses were conducted with the use of SAS software, version 9.4 (SAS Institute).