May 13 2023
Even on LinkedIn, you still see posts and comments asserting that the COVID-19 vaccines aren’t “real” and alleging that they do more harm than good. This is usually based on articles of questionable value and the author’s brother-in-law catching COVID-19 while vaccinated. Public health, however, warrants serious research and is not a matter of anecdotes.
The real question is whether the administration of these vaccines to large populations was effective in curbing the pandemic and saving lives. Today, we can answer yes to some of these questions with simple methods applied to data from the US government. Specifically, we can analyze US CDC and census data on state-by-state Excess Deaths and fractions of the population vaccinated for 2021 and 2022. The results are obvious but, sometimes, the obvious needs belaboring.
Excess Deaths: An Objective Metric
In a previous post, I recommended counts of Excess Deaths as a metric of the impact of COVID-19. These counts compare a model of pre-pandemic mortality with actual death counts during the pandemic when they exceeded the fluctuations compatible with the model. It’s more objective than “COVID-19 deaths” because (1) these are counted only for people known to be infected, and (2) different authorities have different criteria to impute a death to COVID-19 instead of a co-morbidity or an unrelated event.
Data from 2021-2022
We don’t need to consider deaths from 2020 because vaccine distribution started in 2021. We can also ignore 2023 data because (1) the counts are not complete yet and (2) new, “bivalent” vaccines are being used. The data for 2021 and 2022 are complete, and about the same vaccines. The different states in the US have had varying policies on vaccine distribution, and we know the percentage of the population in each state that had received all the shots needed to be fully vaccinated by 12/2022. If the vaccine made no difference, we should see no difference in mortality between states where almost everyone was vaccinated and states where only half the population was.
To check this out, we retrieved Excess Deaths and Vaccination Rates from the CDC. To account for population differences between states, we also retrieved population counts by state from the Census Bureau, to get Excess Deaths per 100,000 inhabitants.
Excess Deaths versus Vaccination Rates
The scatterplot is for 50 states, the District of Columbia, and Puerto Rico:
The purple lines are the medians of each variable. If the vaccines had no effect, we should see roughly the same number of points in each quadrant. The 38 points in the top-left and bottom-right quadrants on the one hand, and the 14 points in the bottom-left and top-right quadrants on the other should be like the result of 52 coin flips. The probability of up to 14 heads out of 52 coin flips is 0.0006, which is strong evidence against the hypothesis that the vaccine has no effect.
In other words, we can conclude that there is a clear negative correlation between Excess Deaths and Vaccination rates. Within the US, high vaccination rates are clearly associated with low Excess Deaths and vice-versa.
Scatterplot with Territory Names
To visualize these groups, we can label (almost) every state or territory on the map:
The states for which we couldn’t fit labels on the map are Hawaii, Maine, Maryland, New Hampshire, Pennsylvania, and South Carolina. The positions of individual states on this map are not always easy to explain. All but one of the states In the Low-Excess Deaths/Low Vaccination Rate quadrant have low population densities. This could account for the results but not for the exception of Michigan.
In the High-Excess Deaths/High Vaccination Rate quadrant, Vermont stands out. It has a low population density but the median age of 43 is the 6th-highest in the country. It could be an explanation, but other territories with an even older population, like Puerto Rico, did better.
The pandemic is now officially over. The US data validates the decision to distribute the vaccines early in 2021 in terms of lives saved. All it takes to see it is downloading data from the US CDC and Census Bureau and applying basic statistics.
Assessing the effect of the vaccines on the reproduction rate of the virus is more complex. Several authors have done it the UK as early as 2021, and in the US and Brazil in 2022.
Chen, X., Huang, H., Ju, J. et al. Impact of vaccination on the COVID-19 pandemic in U.S. states. Sci Rep 12, 1554 (2022).
Percio, J., Cabral, C.M., Fantinato, F.F.S.T. et al. Effect of vaccination against Covid-19 one year after its introduction in Brazil. Trop Dis Travel Med Vaccines 8, 25 (2022).
Pritchard, E., Matthews, P.C., Stoesser, N. et al. Impact of vaccination on new SARS-CoV-2 infections in the United Kingdom. Nat Med 27, 1370–1378 (2021). https://doi.org/10.1038/s41591-021-01410-w