key: cord-0705727-mlrn72yq
authors: Robertson, John F R; Sewell, Herb F
title: COVID-19 vaccine efficacy data: solid enough to delay second dose?
date: 2021-06-10
journal: Lancet
DOI: 10.1016/s0140-6736(21)01046-1
sha: 5885cd828da25a4c2b44607d752555ed7a04f248
doc_id: 705727
cord_uid: mlrn72yq

nan

It is well established that randomised controlled trials are the gold standard for estimating the effect of interventions. In such studies, outcomes in participants randomised to receive the new intervention are compared with those randomised to the control product, to produce unbiased estimates of the effect of the intervention.

E l i z a b e t h M c Q u a d e a n d James Platts-Mills show a new analysis of published data from our randomised controlled trial 1 of the ChAdOx1 nCoV-19 (AZD1222) vaccine to further explore the effect of vaccination on prevention of asymptomatic infections.

McQuade and Platts-Mills propose a counterfactual that is the combination of symptomatic cases prevented and asymptomatic cases observed, to represent the hypothetical situation that might occur if a vaccine solely had effect on symptomatic infections and had no ability to prevent asymptomatic infection. In this situation, an excess of asymptomatic cases in the vaccine group would be observed, which McQuade and Platts-Mills attempt to quantify.

Although the underlying assumptions might be too stringent to be realistic, the results they showed are interesting and highlight the difficulties of estimating true vaccine efficacy against asymptomatic infections. We showed efficacy against all cases-that is, the combination of symptomatic or asymptomatic cases. 1 The analysis of any nucleic acid amplification test-positive outcomes maintains the benefits of the randomisation within the trial while incorporating the people that might have otherwise been symptomatic were it not for vaccination. The highly significant efficacy seen for this endpoint shows the overall benefit of the vaccine in preventing opportunity for trans mission of SARS-CoV-2. infections is shown by the outcome that includes any nucleic acid amplification test-positive cases. However, there is still value in directly estimating the efficacy against asymptomatic infections. As estimated in table 1 of the Article, 1 this parameter is calculated using the rate ratio that compares asymp tomatic infections between the vaccinated and control groups. However, we argue that the rate of asymptomatic infections in the control group is not the appropriate comparator because it requires the implausible assumption that all prevented symptomatic cases were completely averted, rather than converted to asymptomatic infections. If it is instead assumed that all symptomatic infections prevented by the vaccine would become asymptomatic, the appropriate counterfactual would be the rate of asymptomatic infections in the control group plus the rate of symptomatic cases that was averted by vaccine. Using data presented in table 1 of the Article, 1 we calculated expected efficacy on the basis of this alternative counterfactual comparison to estimate an upper bound on the true efficacy against asymptomatic infections. We estimate efficacy against asymptomatic COVID-19 cases to be 61·9% (appendix), which is much higher than the reported estimate of 22·2%. This analysis provides further confidence that the ChAdOx1 nCoV-19 vaccine (AZD1222) provides substantial protection against asymptomatic infections. ETRM reports grants from the US National Institute of Allergy and Infectious Diseases. JAP-M declares no competing interests. Second, individuals documented to have had a SARS-CoV-2 viral infection have a substantially greater response to the first dose of vaccine (up to over a thousand times the neutralising antibodies before the first vaccine dose), which is much greater than the level of response seen through natural infection and also after the second dose of vaccine in individuals who are naive in terms of previous exposure to SARS-CoV-2 virus. 4, 5 Third, the reported 47% reduction in infection rates on days 1-14 does not mirror findings reported in any of the phase 3 randomised clinical trials including populations that were mostly virus-naive individuals [6] [7] [8] or in preliminary analyses 9,10 of much larger population-based cohorts than that described by Amit and colleagues. 1 The immunologically straight forward explanation of these results is that the first dose of BNT162b2 was given to a population of HCWs, a substantial proportion of whom had been exposed to the SARS-CoV-2 infection and therefore the first dose of vaccine was, in essence, equivalent to a booster dose generating a secondary immune response. How else could the infection rate reduction between days 1-14 be explained?

Of note, the difference on days 1-14 (47% vaccine efficacy) and days 15-28 (85%) is 38%, which lies within other estimates of vaccine efficacy for the much larger population cohorts from Israel (33-59% rate of reduction). 10 Additionally, the reporting days for infections in this study (days 15-28) are not only unusual, given that the second dose is scheduled for day 22, but also hide the fact that if only a single dose was given, the neutralising antibodies would be falling by days 22-28 (with decreasing immunity) rather than rising quickly as they do with a second dose of mRNA vaccine. 11 On this basis, we feel the authors should withdraw their Correspondence until they can provide a more substantive report with all the appropriate serology of these HCWs before the first vaccine dose.

The UK Government and Public Heath England, along with the general media, have seized these results to support and justify the UK policy of delaying the second dose of BNT162b2 to 12 weeks, but our concerns need to be addressed to ensure scientific rigour. As we have noted previously, 12 we have reason to believe delaying the second dose of BNT162b2 carries considerable personal and population risks.

JFRR holds shares and was a Director and Chief Scientific Officer, 2003-13, at Oncimmune, which was spun out from the University of Nottingham, UK, as a company for early diagnosis of cancer using detection of autoantibodies to cancer antigens. In 2020, Oncimmune used the technology platform for measuring antibodies to SARS-COV-2 antigens; JFRR had not been involved in this development. HFS declares no competing interests.

Faculty of Medicine and Health Sciences, University of Nottingham, University Park, Nottingham NG7 2RD, UK

Early rate reductions of SARS-CoV-2 infection and COVID-19 in BNT162b2 vaccine recipients

Seroprevalence of SARS-CoV-2 antibodies in healthcare workers at a London NHS Trust

Risk of COVID-19 among front-line health-care workers and the general community: a prospective cohort study

Binding and neutralization antibody titers after a single vaccine dose in health care workers previously infected with SARS-CoV-2

Antibody responses in seropositive persons after a single dose of SARS-CoV-2 mRNA Vaccine

Safety and efficacy of the BNT162b2 mRNA Covid-19 vaccine

Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARSCoV-2: an interim analysis of four randomised controlled trials in Brazil, South Africa, and the UK

Efficacy and safety of the mRNA-1273 SARS-CoV-2 vaccine

The effectiveness of the first dose of BNT162b2 vaccine in reducing SARS-CoV-2 infection 13-24 days after immunization: real-world evidence

Estimating real-world COVID-19 vaccine effectiveness in Israel

Durability of responses after SARS-CoV-2 mRNA-1273 vaccination

Delayed second dose of the BNT162b2 vaccine: innovation or misguided conjecture?

We thank John Robertson and Herb Sewell for their interest in our Correspondence. 1 We reported early rate reductions in SARS-CoV-2 infections and COVID-19 disease in health-care workers (HCWs) working at the Sheba Medical Center, Israel, receiving the BNT162b2 mRNA vaccine. 1 In Israel, individuals previously infected with SARS-CoV-2 were ineligible for vaccination at the time of our evaluation. Indeed, 538 (6%) of 9647 HCWs at the Sheba Medical Center were infected before vaccine roll-out and excluded from the analysis. 1 Moreover, serology screening before receiving the first dose was offered to HCWs in our hospital. Overall, 5835 HCWs, none of whom were known to have been infected previously, were tested. 59 (1%) tested positive for antibody and consequently did not receive the vaccine. Therefore, we considered the HCW cohort included in this analysis mostly antibody-negative. Theoretically, a small proportion of HCWs could have been antibody-positive due to unrecognised past infection and were not tested before receiving the first dose; however, extrapolating from the large proportion of tested HCWs, these numbers should be negligible. Moreover, a proportion of antibodypositive people will be included in most vaccine efficacy evaluations and therefore reflect real-life settings. 2, 3 We report adjusted rate reductions of 30% (95% CI 2-50) in all SARS-CoV-2 infections and of 47% (17-66) in symptomatic COVID-19 during days 1-14 after first dose of BNT162b2. Two biases should be considered in interpreting this estimate of vaccine efficacy after first dose period (days 1-4): referral bias, where symptoms developing after the first dose were attributed to