Issues in COVID-19 research and statistical analyses (Part VIII)

Re-analyses on vaccine efficacy in COVID-19 research

Much of the news these days focuses on vaccine efficacy rates being re-updated due to various new COVID-19 variants coming out often, originating in different countries, like delta (B.1.167.2) from India and lambda (C.37) from Peru. These vaccine rates are being re-updated and re-analyzed, but not very much is discussed about how these rates are constantly analyzed and assessed. What population of data ends up being used for re-analyses? If the data is used from another country, how can the results be generalized to the United States?

The Saul et al article below uses a three segment, two knot linear regression to model and handle the separate periods of vaccinations. While the modeling seems fine where the authors re-analyzed the Pfizer Phase 3 vaccine trial data, their conclusions that a 2nd vaccination did not increase efficacy appears to go against what has already been found. Would their analyses have changed if they included more people who have been vaccinated now to have more heterogeneity? Also, how would it have been different not doing a meta-analyses on summarized data but on person level data?

In another re-analysis by Hunter and Brainard using meta-analyses, they also found that in re-analysis of real-world Pfizer vaccination data from Israel that a single dose can provide strong efficacy by 21 days and that the 1st and 2nd dose can be spaced out even more than thought, but they do not exclude the 2nd dose. This approach seems reasonable. However, given that the Israeli population may have exposure to different variants than other parts of the world, is this result generalizable to other populations?

These are the various questions that we should be asking about these various re-analyses which reassess published data with meta-analyses or other techniques. Assessing how the analyses were done statistically and whether the results are not only meaningful but also generalizable are all important questions to assess. These are the questions we should continue to ask and not ignore.

Keywords: vaccine efficacy, statistical analysis, COVID-19, Usha Govindarajulu

Written by:

Usha Govindarajulu

July 7, 2021

References:

Saul A, Drummer HE, Scott N, Spelman T, Crabb BS, and Hellard M (March 1, 2021). “Reanalysis of the Pfizer mRNA BNT162b2 SARS-CoV-2 vaccine data fails to find any increased efficacy following the boost: Implications for vaccination policy and our understanding of the mode of action”. MedRΧiv. doi: https://doi.org/10.1101/2021.02.23.21252315

Hunter PR and Brainard J (February 3, 2021). “Estimating the effectiveness of the Pfizer COVID-19 BNT162b2 vaccine after a single dose. A reanalysis of a study of ‘real-world’ vaccination outcomes from Israel” MedRΧiv. doi: https://doi.org/10.1101/2021.02.01.21250957

https://www.cdc.gov/mmwr/volumes/70/wr/social-media/mm7013e3_COVIDVaccineFieldEffectiveness_IMAGE_29Mar21_v2_1200x675-medium.jpg

Usha Govindarajulu is a writer and biostatistician . www.UshaGovindarajulu.com