key: cord-0802021-zmde35xc
authors: Sobieszek, A.; Lipniacka, M.; Lipniacki, T.
title: Vaccine hesitancy strongly correlates with COVID-19 deaths underreporting
date: 2022-03-01
journal: nan
DOI: 10.1101/2022.02.27.22271579
sha: 65aa4ef4bc8a00014791bae5b382b8759fc20ddd
doc_id: 802021
cord_uid: zmde35xc

Vaccine acceptance is a key factor in achieving high immunization coverage and reducing the death toll of COVID-19 pandemic. Analyzing data from Europe and Americas we demonstrated that vaccine hesitancy strongly correlates with underreporting of COVID-19 deaths and cases. This correlation cannot be explained by the differences in economic indexes: GDP and Gini coefficient (measure of income inequalities). There is no correlation of vaccination percentage and Gini coefficient and the correlation with GDP is decreasing in time. The most striking is the comparison of Eastern European and South American countries; the latter group of countries shows significantly higher vaccination percentage while having a lower or comparable GDP and higher Gini coefficient. The analysis suggests that timely and reliable information about the COVID-19 cases and the associated deaths plays a key role in achieving population-wide vaccine acceptance.

Vaccinations were proved to be very effective in reducing the death toll and medical costs of COVID-19 pandemic [1] [2] [3] . Despite the fact that effectiveness of vaccines in reducing COVID-19 deaths was proved both by clinical trials and further studies during subsequent pandemic waves associated with new variants of concern 3 , vaccine hesitancy is still a widespread attitude in some countries. The vaccine acceptance and hesitancy were investigated in numerous survey studies in which people were asked about their willingness to vaccinate 4-7 . In the early study (June, 2020), Lazarus et al. surveyed people in 19 countries, that comprises about 55% of world population, to find differences in willingness to vaccinate rates that range from more than 85% (in China and Brazil), about 75% in India and the US, to less than 60% (in France, Poland and Russia) 8 . The high willingness to vaccinate in China (85%), Brasil (88%) and UK (89%) and the low willingness to vaccinate in Russia (42%) was confirmed by a later (February 2021) IPSOS study 4 . Lindholt et al. studying Denmark, France, Germany, Hungary, Sweden, Italy, UK and USA between September 2020 to February 2021 found large variations in vaccine acceptance ranging from 83% in Denmark to 47% in France and Hungary 6 . However, according to the data retrieved from Our World in Data 5 in France the significant reduction in vaccine hesitancy from 47% to 16% was observed during 2021, resulting in the actual 80% first dose vaccination rate (and nearly 50% booster rate) in the beginning on February 1, 2022.

Arce et al. reports a survey performed in the second half of 2020 across seven low-income countries (Burkina Faso, Mozambique, Rwanda, Sierra Leone and Uganda), five lower-middle-income countries (India, Nepal, Nigeria and Pakistan) and one in an upper-middle-income country (Colombia) 7 . The average acceptance rate across these countries was found equal 80.3% (95% CI 74.9-85.6%) considerably higher than in US 64.6% (95% CI 61.8-67.3%, high income) and Russia 30.4%, ( 95% CI 29.1-31.7%, upper-middle income).

In summary, the survey studies indicate high differences in vaccine acceptance, but without clear relation between the acceptance rate and the income per capita. People in China, Brazil and West European countries exhibited high willingness to vaccinate, while people in Eastern European countries (Russia, Poland, Hungary) had the lowest willingness. The survey studies are a natural method to investigate vaccine acceptance in the time period in which people do not have a full access to vaccines. When vaccines become widely available in a given country, however, the vaccination rate is a more straightforward measure of vaccination willingness, as it reports actual people behavior. As the vaccine availability correlates with the gross domestic product (GDP) per capita the vaccination rate reflects both the people's willingness to vaccinate and the country's economic standing. In European and American countries, considered in this study, the importance of the second factor is decreasing in time. Keeping in mind the possible influence of economic factors we will consider the proportion of people that obtained at least one vaccination dose as a measure of vaccine acceptance.

Surveys indicate that respondents reporting higher levels of trust in information from government sources were more likely to accept a vaccine 6, 8 . It is thus important to analyze how reliability of COVID-19 pandemic reporting correlates with vaccination rates and so vaccine acceptance. Numerous studies indicate that in significant proportion countries the number of excess deaths is substantially higher than the number of COVID-19 confirmed deaths, suggesting a significant COVID-19 deaths underreporting 9, 10 . Similarly, the ratio of cumulative COVID-19 confirmed cases to cumulative COVID-19 associated deaths (defined as the maximum of the COVID-19 confirmed deaths and excess deaths) varies significantly between countries (also before the vaccination campaign) suggesting underreporting of COVID-19 cases. This indicates that some countries report only a relatively small fraction of COVID-19 deaths and cases 9 .

We use the COVID-19 deaths and cases reporting as a measure of reliability of disclosed information about epidemic and show that it strongly correlates with vaccination rates in Europe and Americas. Then focusing on Eastern Europe and South America we show that generally higher vaccination rates and more reliable epidemic reporting in the second region may not be explained by GDP PPP (purchase power parity) per capita or Gini coefficient (that measures economic inequalities with a country).

We analyze the relation between the reliability of COVID-19 deaths reporting and the vaccination rates in Europe and Americas. As measure of reliability of the COVID-19 death reporting we chose the ratio F of the cumulative COVID-19 confirmed (and reported) deaths to the cumulative excess deaths, both calculated since the country's first 50 COVID-19 confirmed deaths until a given time tdeaths, based on the Economist database 10 . Intuitively, the highest is the ratio F, the better is COVID-19 death reporting.

In most analyzed countries F < 1, there are however some countries for which F > 1, as well as three countries (Bahamas, Norway, Uruguay) with negative numbers of excess deaths. It is important to note that Australia (between April 6, 2020 and November 28, 2021) and New Zealand (between February 3, 2020 and January 30, 2022), not considered in this study, despite having small numbers of COVID-19 confirmed deaths, have overall negative excess of mortality equal to 0,05% 10 . This is considered to be a consequence of non-medical anti-epidemic interventions that reduced deaths associated with other infectious diseases such as influenza 11 . The possible reduction of non-COVID-19 deaths suggests that in countries with reliable reporting, in which nearly all deaths associated with COVID-19 infections are reported, F may exceed 1. However, the increase of F to the values much greater than 1 is unlikely associated with further reporting improvement.

In Fig. 1a , we thus assume an arbitrary threshold (varied in Fig. 1e ) Fthreshold = 1.2 and assign it to countries for which F > Fthreshold or to countries for which the number of excess deaths is negative.

In order to measure vaccine acceptance rather than the availability of vaccines in considered countries we defined the vaccination rate S as the share of individuals that received at least one vaccination dose before a given time tvacc. For Fig. 1a we set tdeaths = March 31, 2021 and tvacc = September 30, 2021, to find that ρS-F, the population-weighted Pearson correlation between S and F, is ρS-F = 0.78. To support this finding we calculate also ρ * S-F, the population-weighted Pearson correlation between rank values of S and F, and obtain ρ * S-F = 0.79. Correlation between rank values is an appropriate measure of relation also between variables that not are not normally distributed. Of note, the non-weighted Pearson correlation on ranks is known as Spearman correlation. Our choice of tdeaths < tvacc allow to study the . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 1, 2022. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 1, 2022.

Since the nominal values of tdeaths, tvacc as well as Fthreshold chosen for Fig. 1a , are to some extent arbitrary in Figs. 1c, 1d and 1e we analyze dependence of ρS-F as well as ρ * S-F on these three parameters. Generally, we may notice that ρS-F is close to ρ * S-F indicating linear relation between S and F. Both ρS-F and ρ * S-F weakly depend on tdeaths (in range December 31, 2020 -July 30, 2021) and on Fthreshold (in range 1.0-1.5). Not surprisingly both correlation coefficients grow with tvacc between April 30, 2021 and (Fig. 2b and Fig. 2d) suggests no correlation between them, but allows us to distinguish three groups of countries: Eastern . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The analysis in Fig. 2 suggests that higher vaccination rate may be explained by the higher GDP PPP only in the early phase of vaccination campaign when in part of the considered countries the vaccines were not widely available. Surprisingly, the vaccination rate is not anti-correlated with Gini coefficient; i.e. in the considered countries the economic inequalities are not associated with the lower vaccination rates. In further analysis we focus on South American and Eastern European countries that show surprising differences in vaccination rates.

. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 1, 2022. ; . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 1, 2022. ; Comparison of South America and Eastern Europe is important because a higher vaccination rate in South America with respect to Eastern Europe may not be explained by the higher GDP PPP per capita. As shown in Fig. 4a , although in each region, separately, vaccination rate increases with GDP, when both regions are considered jointly the population weighted Pearson correlation ρS-GDP and ρ * S-GDP becomes negative for tvacc ≥ August 31, 2021 (Fig. 4c) . The weighted correlation is strongly influenced by the two most populous countries: Russia and Brazil. As of September 30, 2021, Russia having GDP PPP per capita more than twice higher than Brazil had more than twice lower vaccination rate (Fig. 4a) .

Strikingly, in the epidemic period till September 30, 2021, 87% of excess deaths in Brazil were reported . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 1, 2022. ; https://doi.org/10.1101/2022.02.27.22271579 doi: medRxiv preprint as COVID-19 deaths, while this proportion is 19% in Russia (Fig. 3a) . Similarly, the ratio P(tcases = tdeaths = September 30, 2021) of COVID-19 confirmed cases to COVID-19 associated deaths exceeded 30 in Brazil and was less than 10 in Russia (Supplementary Figure 2a) , showing a much more reliable epidemic reporting in Brazil than in Russia. Finally, we may notice that all South American countries which on average have higher vaccination rates than Eastern European countries have also higher Gini coefficients, which results in surprisingly positive correlation between these two variables ( Fig. 4b and   Fig. 4d ). Higher Gini coefficient and lower GDP PPP per capita in South America in relation to Eastern Europe rules out the possibility that higher vaccination rate and more reliable epidemic reporting are jointly a consequence of better economic standing of South America with respect to Eastern Europe.

Considering European and American countries we showed a significant correlation between reliability of COVID-19 pandemic reporting and vaccine acceptance. The reliability of reporting is assessed based on two measures: ratio, F, of cumulative COVID-19 confirmed cases to excess deaths (in the same period of time) and ratio, P, of COVID-19 confirmed cases to COVID-19 associated deaths. In some countries F is close to 1, and P is of order 100 (which suggests reliable reporting), while in some countries F is below 0.25, and P is below 10, which suggests that only a small fraction of COVID-19 deaths and cases are confirmed and disclosed to the public. Our measure of vaccine acceptance is the proportion of the population that received at least one vaccination dose. Obviously, the vaccinated proportion reflects both willingness to vaccinate, and availability of vaccines; however, as the vaccines become more and more available the first factor becomes dominant.

Intuitively, correlation between reliability of reporting and vaccinated proportion can be a consequence of confounding factors related to economic standing. High GDP and low Gini coefficient can positively influence both epidemic reporting and vaccination rate as the richer countries have more resources for COVID-19 testing and their citizens have better access to vaccines. In fact, there exists correlation between GDP PPP per capita, however this correlation (for European and American countries) decreases below 0.33 after October 31, 2021, when vaccines become widely available in Americas and Europe. Surprisingly, we observe no negative correlation between vaccination rate and Gini coefficient. This implies that high correlation between reliability of reporting and vaccinated proportion may not be fully explained by the economic factors.

The most striking is the comparison of South American and Eastern European countries. For this selected subset of countries, we observe a high population-weighted Pearson correlation between vaccination rate and COVID-19 deaths reporting, ρS-F = 0.75 (and population-weighted Pearson correlation between rank values of S and F, ρ * S-F = 0.80) that may not be explained by economic factors. On average South American countries have a higher vaccination rate, lower GDP PPP per capita, and . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted March 1, 2022. substantially higher Gini coefficients than Eastern European countries. Better economic standing of Eastern European countries, suggest that vaccine hesitancy in these countries follows from unreliable and insufficient reporting of COVID-19 pandemic. In the case of low mortality diseases such as COVID-19, people may not base their risk assessment on information from their relatives or friends, instead they require reliable information about epidemic risk from medical or governmental authorities. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted March 1, 2022. ; https://doi.org/10.1101/2022.02.27.22271579 doi: medRxiv preprint

Effectiveness of COVID-19 booster vaccines against covid-19 related symptoms, hospitalisation and death in England

Deaths involving COVID-19 by vaccination status, England -Office for National Statistics

How do death rates from COVID-19 differ between people who are vaccinated and those who are not? Our World in Data

Global attitudes : COVID-19 vaccines

Willingness to get vaccinated against COVID-19

Public acceptance of COVID-19 vaccines: cross-national evidence on levels and individual-level predictors using observational data

COVID-19 vaccine acceptance and hesitancy in low-and middleincome countries

A global survey of potential acceptance of a COVID-19 vaccine

The pandemic's true death toll: millions more than official counts

The Economist's tracker for covid-19 excess deaths

Reduced mortality in New Zealand during the COVID-19 pandemic

World Bank

World GDP per capita

This study was supported by the National Science Centre Poland OPUS, grant 2018/29/B/NZ2/00668.The funding agency had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.

The authors have no competing interests.