UK study of health care workers provides evidence Pfizer/BioNTech vaccine protects against infection

A preprint was posted today on the Lancet site present results of a study investigating efficacy of Pfizer/BioNTech vaccine in health care workers in the UK.

This is part of the SARS-CoV-2 Immunity and Reinfection Evaluation (SIREN) study that followed a large cohort of UK healthcare workers from >100 hospitals during the pandemic to assess the effect of prior infection on protection agains re-infection, and was amended to look at vaccine efficacy in January 2021.

In this report, there are a total of 23,324 participants from 104 hospitals that participated. Of these, 8,203 (35%) had been previously infected based on antibody or previous PCR tests.

By February 5, 19,384 (83%) had already received one dose of the the Pfizer/BioNTech vaccine, and 1,605 (7%) received both doses, with the median length of time between doses of 23 days. 6% others received the AstraZeneca/Oxford vaccine and were not included in this analysis.

This analysis tracks COVID-19 cases arising between December 7, 2020 and February 5, 2021, with regular asymptomatic PCR testing of all participants supplemented by additional testing if reporting symptoms in between scheduled asymptomatic PCR testing dates.

They did a rigorous statistical analysis to assess the hazard of infection (symptomatic or asymptomatic) and compare between vaccinated and unvaccinated groups, while adjusting for various factors that might contribute to risk of infection and thus bias the comparison including infection rate by week, age, sex, hospital, ethnicity, co-morbidities, region, job, and patient contact. This makes the comparison of vaccinated to non-vaccinated more representative of the potential causal effect of vaccination.

Summary of their results:

  • Almost 90% of HCW received the vaccine, so there was not evidence of substantial vaccine hesitancy in this population.

  • There were at total of 977 new infections (rate of 14 per 10,000 person-days), with only 71 occurring 21 days after 1st dose (8 per 10,000 person-days) and only 9 occurring at least 7 days after second dose (4 per 10,000 person-days).

  • This resulted in estimates of vaccine effectiveness in reducing risk of infection (symptomatic or asymptomatic) of 70% starting 21 days after first dose, and 85% starting 7 days after second dose. They did not specify the effectiveness after first dose but before second dose, which would be <70% but not clear how much lower.

  • They note that the vaccine efficacy was higher in the group not previously infected (72% and 86%), suggesting the vaccine efficacy was lower in the previously infected group -- but there were so few infections in either vaccinated or unvaccinated in previously infected group that there is not enough data to reliably estimate the efficacy for previously infected groups.

  • They found that the previously infected group overall had 90% lower infection rates than those who had not been previously infected, across both vaccinated and unvaccinated groups.

Before this, the only previous study demonstrating vaccine efficacy in reducing total infections = symptomatic + asymptomatic infections was a subset of the AstraZeneca study that demonstrated 52% total efficacy.

This paper provides results for the Pfizer/BioNTech that demonstrated strong efficacy against total infections of 85% after second dose and 70% starting 3 weeks after first dose, which further validates the positive results for this vaccine from the phase 3 study and other population level analyses done in Israel. This also provides evidence that the Pfizer/BioNTech vaccine has efficacy against the UK B.1.1.7 variant that came to dominate the UK starting in December 2020, which is encouraging since the Pfizer/BioNTech phase 3 study was completed before its emergence so had no information on efficacy vs. this fast-spreading variant that appears to cause more severe disease.

Interestingly, they found previous infection to be strongly protective, as well, as there were very few reinfections among vaccinated or non-vaccinated, and making it difficult to assess the degree of benefit conferred by the vaccine for previously infected in the short time frame of this study.

This reinforces what is known from a careful reading of the data and literature: reinfection within 6 months of infection is very rare, but sometimes happens, and recovered individuals retain a reasonable level of immune protection vs. reinfection. This raises questions of why previous infection status is not a primary criterion for determining vaccine priority in most of the USA, and apparently the UK as well.

While the uncertainty of durability of immunity means that it is reasonable to recommend vaccination even to those who were previously infected, during a time of limited vaccine supply it is likely prudent to do what Israel is doing and put those who have recovered from previous infection in a later vaccination priority group, at least if they have clear evidence of SARS-CoV-2 antibodies. It is unclear why this strategy has not been pursued in the USA or UK.

But overall, this study provides another brick in the wall of evidence for vaccine effectiveness against infection, which of course would protect against transmission since those who are not infected would not transmit to others. The vaccines prevent a high proportion of infections that would have occurred sans vaccine, but not all, which is why it is still advisable for vaccinated people to wear masks and practice recommended distancing, at least while the levels of community viral spread are reasonably high.

Figure 2a from paper presenting the hazard ratio of vaccinated:non-vaccinated. One minus this hazard ratio gives an estimate of vaccine efficacy. It can be seen that there is substantial efficacy (>70%) from days 21+ after first vaccine dose, and even more (>85%) from days 7+ after second dose. The day 0-3 result is likely an artifact of deferred effect bias, meaning that those experiencing symptoms would likely defer vaccination thus reducing cases in the few days after vaccination.

574 views0 comments

©2020 by Covid Data Science. Proudly created with