Study population and setting
This observational study assessed protection from symptomatic PCR-confirmed SARS-CoV-2 infection conferred by vaccination and/or prior infection among 13,109 healthcare workers (HCWs, 74% female, median age 39 years) in a hospital network in the UK from April 23, 2020 to February 28, 2021. HCWs were classified by their vaccination status (1 dose vs. 2 doses of either the Pfizer-BioNTech or Oxford-AstraZeneca vaccines) and antibody status (any previous positive test for anti-spike IgG antibodies). Person-time was classified into five groups: 1) unvaccinated, seronegative; 2) unvaccinated, seropositive; 3) vaccinated once, seronegative; 4) vaccinated twice, seronegative; and 5) vaccinated once or twice, seropositive. HCWs could contribute person-time at risk to multiple exposure groups. Vaccinated follow-up time began 14 days after vaccination, while seropositive follow-up time began 60 days after initial positive antibody result. Asymptomatic HCWs were offered voluntary PCR testing for SARS-CoV-2 infection every two weeks and antibody testing every two months beginning April 23, 2020. Risks of infection were compared across vaccination and antibody status, and were compared between the B.1.1.7 variant and non-B.1.1.7 variants. The B.1.1.7 variant was identified via PCR test results indicative of S gene target failure (SGTF) as of November 16, 2020; all positive samples starting on December 1, 2020 were sequenced to identify the viral lineage. The authors used Poisson regression to test for differences in incidence rates in each exposure group, adjusted for month, age, sex, ethnicity, occupational role, amount of patient contact, and whether the HCW worked in a non-ICU ward caring for COVID-19 patients. Sensitivity analyses were conducted to test for differences by vaccine type, by time since vaccination, by SGTF, and by B.1.1.7 vs. non-B.1.1.7 lineage.
Summary of Main Findings
A total of 13,109 healthcare workers (HCWs) contributed 2,835,260 person-days of time to the study. Over the study period, 1,273 HCWs tested positive for IgG anti-spike antibodies; 8,285 HCWs received the Pfizer-BioNTech vaccine and 2,738 received the Oxford-AstraZeneca vaccine. There was an increase in the incidence rate above baseline levels, even after adjustment, during the first 14 days after vaccination; these cases were excluded from analysis. There were 294 symptomatic infections among unvaccinated seronegative HCWs during follow-up, 32 among vaccinated HCWs (31/32 were previously seronegative), and 1 in an unvaccinated seropositive HCW. Relative to unvaccinated seronegative HCWs, the adjusted incidence rate ratio (IRR) for seropositivity was 0.02 (95% CI: <0.01 to 0.18). Among previously seronegative HCWs, the adjusted IRR for a single vaccine dose was 0.33 (95% CI: 0.21 to 0.52), and there were no infections among HCWs who had received a second dose. There were 16 COVID-19 hospitalizations among unvaccinated seronegative HCWs, and no hospitalizations among vaccinated HCWs. When considering any SARS-CoV-2 infection regardless of symptoms, the IRR for seropositivity was 0.15 (95% CI: 0.08 to 0.26); among seronegative HCWs, the IRR was 0.36 (0.26 to 0.50) for the first vaccine dose and 0.10 (95% CI: 0.02 to 0.38) for the second dose. There were no statistically significant differences in incidence by vaccine type. Viral loads were highest among unvaccinated previously seronegative cases, followed by vaccinated previously seronegative cases, and lowest among unvaccinated seropositive cases. There was no evidence that SGTF or B.1.1.7 lineage altered the incidence rate among seropositive or vaccinated HCWs, though the confidence intervals were wide.
Frequent testing of HCWs allowed consideration of COVID-19 incidence across person time for a wide range of exposures (defined by antibody and vaccination status). A high proportion of infections was assessed for SGTF and/or sequenced to determine viral lineage.
Follow-up time after vaccination, particularly the second vaccination, was short and few events were observed. There were increases in incidence during the first two weeks after vaccination, but this person-time and these cases were not included in analyses: if these HCWs were systematically different from those remaining in the risk set, the ensuing selection bias would overestimate vaccine protection. Any prior positive antibody test was used as a proxy for previous infection, which is subject to several limitations. First, the frequency of testing among HCWs was not reported, and it appears that all person-time before a positive antibody test was classified as seronegative; thus, misclassification of person-time is likely and incidence among seropositive HCWs may be underestimated. If HCWs with prior symptoms were more likely to seek serological testing, and if asymptomatic infection confers less protection against re-infection, then this issue would be exacerbated. On the other hand, false positive results are not uncommon even with highly specific assays in low-prevalence populations, which would bias estimates of incidence among seropositive HCWs in the other direction. HCWs with positive PCR results for SARS-CoV-2 infection were removed from all risk sets; it is not clear why these events were not used as a better indicator of prior infection. Seropositive and vaccinated HCWs were less likely to be tested for SARS-CoV-2 infection, which may have underestimated incidence rates of asymptomatic infection in those groups.
This study took advantage of frequent PCR and serology testing among HCWs in an environment of high transmission to directly compare SARS-CoV-2 incidence rates by vaccination and antibody status, and was able to compare estimates of protection against the B.1.1.7 variant vs. non-B.1.1.7 lineages.
This review was posted on: 5 April 2021