Paradoxical Case Fatality Rate dichotomy of Covid-19 among rich and poor nations points to the “hygiene hypothesis” ======================================================================================================================= * Bithika Chatterjee * Rajeeva Laxman Karandikar * Shekhar C. Mande ## Abstract In the first six months of its deadly spread across the world, the Covid-19 incidence has exhibited interesting dichotomy between the rich and the poor countries. Surprisingly, the incidence and the Case Fatality Rate has been much higher in the richer countries compared with the poorer countries. However, the reasons behind this dichotomy have not been explained based on data or evidence, although some of the factors for the susceptibility of populations to SARS-CoV-2 infections have been proposed. We have taken into consideration all publicly available data and mined for the possible explanations in order to understand the reasons for this phenomenon. The data included many parameters including demography of nations, prevalence of communicable and non-communicable diseases, sanitation parameters etc. Results of our analyses suggest that demography, improved sanitation and hygiene, and higher incidence of autoimmune disorders as the most plausible factors to explain higher death rates in the richer countries Thus, the much debated “hygiene hypothesis” appears to lend credence to the Case Fatality Rate dichotomy between the rich and the poor countries. **Significance** The current COVID-19 epidemic has emerged as one of the deadliest of all infectious diseases in recent times and has affected all nations, especially the developed ones. In such times it is imperative to understand the most significant factor contributing towards higher mortality. Our analysis shows a higher association of demography, sanitation & autoimmunity to COVID-19 mortality as compared to the developmental parameters such as the GDP and the HDI globally. The dependence of sanitation parameters as well as autoimmunity upon the mortality gives direct evidences in support of the lower deaths in nations whose population do not confer to higher standards of hygiene practices and have lower prevalence of autoimmune diseases. This study calls attention to immune training and strengthening through various therapeutic interventions across populations. Keywords * COVID-19 * Risk factors * Computational modelling ## Introduction Higher GDP and improved Human Development Index (HDI) has led to improved sanitation and consequently reduction in the load of communicable disease burden in many countries (1). Interestingly, disease burden of non-communicable diseases now occupies areas of major concern in the higher HDI countries, whereas these are of lesser significance in the low and low-middle income countries. Thus, a distinct correlation between HDI status of a country, and the prevalence of specific diseases has emerged in recent history (2). About half of the world’s population lives in low and low middle income countries (3). Typically access to healthcare facilities, hygiene and sanitation is poorer in these countries and is often believed to be the contributing factor of higher incidence of communicable diseases in these countries. Thus, it is not unexpected if infectious disease pandemics, such as that due to SARS-CoV-2, have catastrophic consequences in the low and low-middle income countries. Yet on the contrary, the disease prevalence and the Case Fatality Ratio (CFR) during the Covid-19 pandemic shows a contrasting opposite trend in the low and low-middle income countries when compared to that of the high income countries. It is fascinating to explore reasons that would explain higher prevalence and deaths due to Covid-19 among richer nations. An interesting relationship between severity of Covid-19 outcome and several non-communicable disorders such as diabetes, hypertension, cardiovascular disorders has been noted (4–6). However, the non-communicable disease burden of a country and its apparent relation to CFR due to Covid-19 has not been explored in detail yet. As a large population with these disorders lives in the high HDI countries, co-morbidities with diabetes, hypertension, cardiovascular disorders and respiratory disorders might have emerged as important determinants of CFR due to Covid-19 in these countries. Similarly, people above the age of 65 are also believed to be at a greater risk (7), with the percentage of such people being significantly more in the higher HDI countries (8). Thus, co-morbidities with non-communicable diseases and the fraction of people living above the age of 65, being skewed towards the high-income countries, offers possible explanations to the perplexing observation of CFR dichotomy among nations. Among the many parameters that lead to non-communicable diseases, autoimmunity occupies an important place. Interestingly, a correlation between autoimmune disorders and HDI has been proposed (9). Since, one of the primary manifestations of Covid-19 has been a severe autoimmune reaction in the later phase of the disease (10), susceptibility to autoimmunity in SARS-CoV-2 infection is a possibility to consider. One of the reasons of rising prevalence in autoimmune disorders in the western countries has been proposed to be that related to the “hygiene hypothesis” (11). The hypothesis postulates that exposure to pathogens early in life protects people from allergic diseases later in their lives. Moreover, improvement in hygiene practices such as better sanitation, availability of safe drinking water, hand washing facilities etc. reduces the impact of communicable diseases. On the contrary such a reduction to the exposure to infectious agents might be related to higher prevalence of autoimmune disorders (12). We wished to explore if epidemiological data supports correlation between autoimmune prevalence, the “hygiene hypothesis” and how much of the variation in death per million due to Covid-19 is explained by the same. We have chosen more than 25 parameters and attempted to find correlation between these and CFR, if any. These parameters include GDP, HDI, prevalence of various diseases, demographic parameters, various sanitation parameters etc. Our findings fascinatingly reveal that the apparent correlation between GDP of a nation and Covid-19 related deaths, or that between GDP and CFR, is a manifestation of other parameters. ## Results Analysis was carried on reported numbers of total 10112754 cases 501562 deaths as on the 29th June 2020. Data from well-known corona virus global deaths tracking sources (13–15) show that more than 70 percent Covid-19 mortality has occurred in high income countries like Italy, Spain, UK, France and USA (16). Our analyses also reveal a similar trend (Fig. 1). To explore the dichotomy of rich and poor countries in relation to Covid-19 deaths, we considered several potential explanatory parameters. The factors included percentage of older population, population density, incidence of communicable and non-communicable diseases etc. The data shows a positive correlation between log (GDP per capita) and log (deaths per million), abbreviated as LDM (Fig. 3, SI Appendix, Tables S1, Fig. S1b), which reinforces the information of richer countries being most burdened by the disease. ![Figure 1.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/08/04/2020.07.31.20165696/F1.medium.gif) [Figure 1.](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/F1) Figure 1. The number of Deaths per Million inhabitants in each country due to Covid-19. Color gradient indicates the country’s log (GDP). It is clearly seen that higher GDP countries appear to have higher prevalence and CFR due to Covid-19. ![Figure 3.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/08/04/2020.07.31.20165696/F2.medium.gif) [Figure 3.](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/F2) Figure 3. Plot showing correlation matrix of every variable with each other. The color key is depicted as −1(red) showing inverse correlation, 0 showing no correlation, and +1(blue) showing maximum positive correlation. The sizes of the circles range from small(low correlation) to larger(high correlation) ### Demographics Among the many distinctive features of higher GDP countries, are its ageing population and a larger fraction of urbanization. We observe that while proportion of population above 65, obesity and the percentage of urban population are positively correlated with the Covid-19 deaths (Fig. 3, SI Appendix, Tables S1, Fig.S1c, Fig. S1e, Fig. S1g) the gender ratio and population density have negligible negative correlation (Fig. 3, SI Appendix, Tables S1, Fig.S1d, Fig. S1f). Interestingly, whereas at an individual level gender does seem to have an impact on Covid-19 related death as numerous studies have shown (17, 18), at the aggregate level the gender ratio varies little across countries. Likewise, urbanization has an impact while population density does not seem to have an impact. We observe a positive correlation between LDM and percentage of population with age over 65, which is expected from the reports showing that older population is more susceptible to the virus (7). Thus, the overall demographic parameters yield a positive correlation with Covid-19 deaths per million (Fig. 2a). ![Figure 2.](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/08/04/2020.07.31.20165696/F3.medium.gif) [Figure 2.](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/F3) Figure 2. The actual values of log (deaths per million) or LDM plotted against their predicted values. The explanatory variables have been grouped, as described in the text, into a) demographics, b) sanitation, c) autoimmune diseases, d) demographics & sanitation, e) sanitation & autoimmune diseases and f) demographics, sanitation and autoimmune diseases. Regression analysis were done with LDM as dependent variable and the combinations of different variables. The R2 of each graph is depicted in Table 1. ### Safe Sanitation Lack of sanitation and poorer hygiene practices are known to be responsible for the higher communicable disease burden in the low GDP countries (1). It is therefore reasonable to expect that parameters describing safe sanitation and safe drinking water to be correlated negatively with the Covid-19 deaths. Surprisingly we find a contrary observation, where different sanitation parameters are correlated positively with the Covid-19 outcome (Fig. 2b). It is therefore perplexing to note positive correlation of sanitation parameters, as described in methods, with the Covid-19 CFR (Fig. 3, SI Appendix, Tables S1, Fig. S1h-m). View this table: [Table 1:](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/T1) Table 1: Showing R-the correlation coefficient of combinations of variables with log (Deaths/Million) as well as correlation of the residues with log (GDP) and HDI. Also showing the adjusted R2. ### Communicable Diseases Several studies have reported the protective role of parasitic/bacterial infections in bolstering human immunity, also referred to as immune training (19, 20). We observe that the prevalence of communicable diseases such as Malaria and Tuberculosis as well as parasitic diseases such as schistosomiasis, Onchocerciasis have a weak negative correlation with LDM (Fig. 3, SI Appendix, Tables S1, Fig. S1t-z1) These intuitive observations related to communicable diseases with respect to Covid-19 deaths are suggestive of the role of “immune training” and the final disease outcome. ### BCG vaccination Recent studies have suggested that the countries in which the majority population was immunized by BCG vaccine had lower LDM due to trained immunity (21). We find negligible of correlation between BGC vaccination in different countries with LDM (Fig. 3, SI Appendix, Tables S1, Fig. S1n). A possible explanation for this difference could be due to the fact that comparisons in the previous studies were made with socially similar nations while we consider a diverse group of countries. Moreover, the distribution showing two distinct clusters for BCG vaccination (SI Appendix, Fig. S1n), and with no apparent correlation within the clusters also makes it difficult to draw any conclusions on the effect of BCG vaccination on CFR at the population level. It is also pertinent to note that we find significant negative correlation of mean BCG vaccination of a country and prevalence of autoimmune disorders (Fig. 3). ### Autoimmunity One of the manifestations of better hygiene and safe sanitation practices in high GDP countries is the increased incidence in autoimmune disorders (22). It has been postulated that better hygiene practices could lower a person’s immunity and make the person susceptible to autoimmune diseases. We therefore tested if any of the autoimmune diseases showed an association with the LDM variable. We interestingly find a strong positive correlation with Multiple Sclerosis, Type 1 Diabetes Mellitus, Rheumatoid Arthritis and Psoriasis and a weaker correlation with Asthma (Fig. 2c, Fig. 3, SI Appendix, Tables S1, Fig. S1o-s). ### Regression Analysis We now combined many related factors as described in methods such as demographic parameters, sanitation, communicable and non-communicable diseases. These combinations were arrived at after calculating correlations among all different parameters under consideration (Fig3, SI Appendix, Tables S1). In order to inspect whether development indices like HDI and GDP of countries had an actual impact on our variable of LDM, we ran regression models using combinations of explanatory factors outlined above and checked the correlation between the residues, i.e. difference in actual values of the dependent variable and its predicted values from predictor variables, to GDP and HDI. After testing various combinations of these groups, we observe that demographic variables, sanitation parameters and incidence of autoimmune disorders together explain most of the observed variation in LDM and development variables do not seem to count once the effect of these explanatory variables is accounted for (SI Appendix, Tables S2). More precisely, if one uses multiple regression to predict death rates, then the residues are uncorrelated with GDP and HDI. The R value for this model is 0.74 and the adjusted R2 value is 0.48. The correlation of this model’s residue with GDP and HDI came to 0.04 and −0.03 respectively (SI Appendix, Tables S2). ## Discussion Various corona virus global deaths tracking sources (13–15, 23) point towards uneven distribution of Covid-19 deaths with richer countries having higher CFR. In order to understand the reasons behind this paradoxical observation we analyzed publicly available data on various parameters. In this context, it is well known that when we have multiple factors correlated with dependent variable and these factors are themselves mutually correlated, it is not possible to determine as to which of these may be the causal variables using purely statistical techniques. To address this, we could collect data from a controlled experiment or look for data on individuals rather than countrywide aggregated data. However, in this instance many variables make sense only at the aggregate level ruling out both these options. In such challenging situations, it is desirable to use domain knowledge and identify the variables whose correlation with the dependent variables has a logical explanation. In the present context, older population having higher death rate can be explained biologically. Similarly, since Covid-19 spreads rapidly via person-to-person contact, population density is likely to have a big impact. Our observation of the weak negative correlation of Covid-19 LDM with communicable diseases, and its positive correlation with incidence of autoimmune disorders in the high GDP countries, is perplexing. As the parasite and bacterial disease burden is high in low and low-middle income countries, this can best be inferred upon by the “immune training” in the population of these countries due to chronic exposure to communicable diseases. As mentioned earlier, the observed correlation between incidence of autoimmune disorders and LDM can be explained as Covid-19 seems to lead to severe autoimmune reaction. Paradoxically, better sanitation leads to poorer “immune training” and thus could be leading to higher LDM. Thus, the observed correlation of demographics, autoimmune disorders and sanitation with LDM does have explanation coming from the domain. The transfer rate for COVID-19 being high, we do not see any correlation between incidence of communicable diseases and LDM due to COVID-19. As can be seen in (SI Appendix, Tables S2), the correlation of GDP and HDI with each model’s residue is almost negligible. Both GDP and HDI seem to have a dominant role in determining the deaths per million yet both become insignificant when our linear regression models considered other notable and explainable confounding factors such as the demography, sanitation and the pre-covid prevalence’s of autoimmune and tropical diseases. The parameters of sanitation and autoimmune diseases consistently contributed significantly to the R2 no matter which combination of variables were chosen. Thus, our study seems to give empirical evidence supporting the “hygiene hypothesis (24)”. It has not escaped out attention that there could be other factors such as the country’s stage of the epidemic and lower reporting/testing in less developed countries that could also affect the mortality numbers. However, the statistical evidence given by us should give a head start to investigate the role of hygiene practices and autoimmunity on Covid-19 susceptibility at individual patient’s level. We note that statistical analyses such as this attempt to provide evidence of practices of on a macroscopic scale. Individual level variations among people, or communities, typically get masked in such analyses. Thus, although we provide a possible explanation based on “hygiene hypothesis” on the CFR differences among economically stronger and weaker countries, this should not be inferred as our advocating a move towards weaker hygiene practices for handling future pandemics. Rather this analysis opens up avenues to consider “immune training” with possibilities of microbiome therapies to supplement improved hygiene and sanitation practices. ## Methods The country-wise Covid-19 deaths per million data was collected from [https://ourworldindata.org/coronavirus](https://ourworldindata.org/coronavirus)(25) with the original source: European Centre for Disease Prevention and Control (ECDC) on the 29th June 2020. We chose deaths per million as compared to the confirmed reported cases as it gives a more reliable parameter to assess the current pandemic. The parameters on GDP per capita, population density, age profile, gender ratio, obesity, HDI, urban population percentage was sourced from the WHO (26) and the World Bank organization (27). One of our postulated factors for Covid-19 deaths being ‘hygiene’, we collected all possible variables that quantify the cleanliness parameters in countries. This included the percentage of access to handwashing facilities, access to basic drinking water, availability of basic sanitation, prevalence of open defecation, safe drinking and safe sanitation. Most of these variables were available through WHO and UNICEF water and sanitation surveillance projects (28). The prevalence percentages of different autoimmune disorders such as Multiple Sclerosis, Type1 Diabetes Mellitus, Psoriasis, Rheumatoid Arthritis, Asthma, and communicable diseases such as Schistosomiasis, Onchocerciasis, Lymphatic Filariasis, Ascaris, Hook worm, Malaria and Tuberculosis were downloaded from the [http://ghdx.healthdata.org/gbd-results-tool](http://ghdx.healthdata.org/gbd-results-tool) (29). The diseases were classified into two broad categories as communicable and non-communicable diseases. The Mean BCG vaccination coverage in percentage was downloaded from [http://www.bcgatlas.org/index.php](http://www.bcgatlas.org/index.php) (30). The 2017 GDP gross domestic products per capita based on purchasing power parity converted to 2011 international dollars and Human Development Index 2018 was used for analysis. For details of sources of each variable consider Dataset S1. We further filtered countries that had at least 10 deaths in total and population size of at least 100,000. Moreover, we restricted consideration of the countries to have at least 4 deaths/million. For the countries missing corresponding values for our variables, we either imputed them with values smaller than the smallest values in the other countries or dropped the countries from further analyses. For the missing values in variables related to drinking water, sanitation and wash we predicted them using a regression model as we noticed a high correlation amongst each other. For missing values of HDI, we used GDP data to impute values for HDI. After filtration and imputation of missing values we were left with 106 countries with values corresponding to more than 25 variables. It is customary to consider log of GDP instead of GDP for analyses in many disciplines to brings it closer to normal distribution. We also considered log of deaths per million instead of deaths per million for all our calculations. We compared individual (Pearson) correlation coefficients of different variables with the log of deaths per million (LDM) as well as with log (GDP) and HDI. This helped in assessing the important variables to be used in our linear regression model. The country-wise data on various parameters: GDP, HDI, sanitation parameters, prevalence of various communicable and non-communicable diseases were clubbed into development variables (GDP, HDI), demographics (percentages of older population, urban population and obesity among adults), Sanitation (percentage of access to handwashing facilities, basic drinking, basic sanitation, no open defecation, safe drinking and safe sanitation), Tropical diseases (Schistosomiasis, Onchocerciasis, Lymphatic Filariasis, Ascaris, Hook worm, Malaria and Tuberculosis) and Autoimmune Diseases (Multiple Sclerosis, Type1 Diabetes Mellitus, Psoriasis, Rheumatoid Arthritis and Asthma). ### Statistical Analysis used To explore which of these combined variables are important, we tried multivariate linear regression model using R with several combinations of variables to obtain insights into our understanding of epidemiology of Covid-19. LDM was assessed for its association with our predictor variables grouped as described above into development variables, demographic variables, sanitation, tropical diseases and autoimmune disorders. Our final objective was to select the best combination of these groups of variables that would yield the largest adjusted-R2 value. We have refrained from reporting p-values since these are typically derived under the assumption of normality of distribution, whereas the distributions of several variables as seen in the scatter plots (SI Appendix, Fig. S1) are far from normal. ## Data Availability dataset is available for data. ## Supplementary Information for ![](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/08/04/2020.07.31.20165696/F4/graphic-5.medium.gif) [](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/F4/graphic-5) ![](http://medrxiv.org/https://www.medrxiv.org/content/medrxiv/early/2020/08/04/2020.07.31.20165696/F4/graphic-6.medium.gif) [](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/F4/graphic-6) Fig S1. Plot showing the scatter plot of each variables with log (deaths per million) View this table: [Table S1.](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/T2) Table S1. Showing the (R)correlation coefficient of each variable with log (Deaths/Million) as well as correlation with log (GDP) and HDI View this table: [Table S2.](http://medrxiv.org/content/early/2020/08/04/2020.07.31.20165696/T3) Table S2. Showing the (R) correlation coefficient of each combinations of variables with log (Deaths/Million) as well as correlation of their residues with log (GDP) and HDI. Also showing the R2-adjusted. ## Acknowledgements BC gratefully acknowledges financial support by the National Centre for Cell Science, Pune. Authors declare no competing financial interests. * Received July 31, 2020. * Revision received July 31, 2020. * Accepted August 4, 2020. * © 2020, Posted by Cold Spring Harbor Laboratory This pre-print is available under a Creative Commons License (Attribution-NonCommercial-NoDerivs 4.0 International), CC BY-NC-ND 4.0, as described at [http://creativecommons.org/licenses/by-nc-nd/4.0/](http://creativecommons.org/licenses/by-nc-nd/4.0/) ## References 1. 1.GBD 2017 Disease and Injury Incidence and Prevalence Collaborators, Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 392, 1789–1858 (2018). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/S0140-6736(18)32279-7&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=30496104&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F08%2F04%2F2020.07.31.20165696.atom) 2. 2. H.-Y. Pan, et al., National incidence of autoimmune liver diseases and its relationship with the human development index. Oncotarget 7, 46273–46282 (2016). 3. 3.World Bank Country and Lending Groups – World Bank Data Help Desk. [https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups](https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country-and-lending-groups). Accessed 29 June 2020. 4. 4. B. Li, et al., Prevalence and impact of cardiovascular metabolic diseases on COVID-19 in China. Clin Res Cardiol 109, 531–538 (2020). [PubMed](http://medrxiv.org/lookup/external-ref?access_num=http://www.n&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F08%2F04%2F2020.07.31.20165696.atom) 5. 5. J. Yang, et al., Prevalence of comorbidities in the novel Wuhan coronavirus (COVID-19) infection: a systematic review and meta-analysis. International journal of infectious diseases (2020). 6. 6. A. Emami, F. Javanmardi, N. Pirbonyeh, A. Akbari, Prevalence of underlying diseases in hospitalized patients with COVID-19: a systematic review and meta-analysis. Archives of academic emergency medicine 8 (2020). 7. 7.COVID-19 and Older Persons: A Defining Moment for an Informed, Inclusive and Targeted Response, United Nations Department of Economic and Social Affairs. [https://www.un.org/development/desa/ageing/news/2020/05/covid19/](https://www.un.org/development/desa/ageing/news/2020/05/covid19/). Accessed 2 July 2020. 8. 8. M. Roser, Human development index (HDI). Our World in Data(2014). [https://ourworldindata.org/humandevelopmentindex#:~:text=The%20Human%20Development%20Index%20(HDI)%20provides%20a%20single%20index%20measure,a%20decent%20standard%20of%20living](https://ourworldindata.org/humandevelopmentindex#:~:text=The%20Human%20Development%20Index%20(HDI)%20provides%20a%20single%20index%20measure,a%20decent%20standard%20of%20living). Accessed 2 July 2020 9. 9. J. Grosse, et al., Incidence of diabetic ketoacidosis of new-onset type 1 diabetes in children and adolescents in different countries correlates with human development index (HDI): an updated systematic review, meta-analysis, and meta-regression. Hormone and Metabolic Research 50, 209–222 (2018). 10. 10. C. Galeotti, J. Bayry, Autoimmune and inflammatory diseases following COVID-19. Nature Reviews Rheumatology 16, 413–414 (2020). 11. 11. H. Okada, C. Kuhn, H. Feillet, J.-F. Bach, The ‘hygiene hypothesis’ for autoimmune and allergic diseases: an update. Clinical & Experimental Immunology 160, 1–9 (2010). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1111/j.1365-2249.2010.04139.x&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=20415844&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F08%2F04%2F2020.07.31.20165696.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000275443000001&link_type=ISI) 12. 12. J.-F. Bach, L. Chatenoud, The hygiene hypothesis: an explanation for the increased frequency of insulin-dependent diabetes. Cold Spring Harbor perspectives in medicine 2, a007799 (2012). [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6MTQ6ImNzaHBlcnNwZWN0bWVkIjtzOjU6InJlc2lkIjtzOjExOiIyLzIvYTAwNzc5OSI7czo0OiJhdG9tIjtzOjUwOiIvbWVkcnhpdi9lYXJseS8yMDIwLzA4LzA0LzIwMjAuMDcuMzEuMjAxNjU2OTYuYXRvbSI7fXM6ODoiZnJhZ21lbnQiO3M6MDoiIjt9) 13. 13.WHO Coronavirus Disease (COVID-19) Dashboard. [https://covid19.who.int](https://covid19.who.int). Accessed 18 July 2020 14. 14.COVID-19 situation update worldwide. European Centre for Disease Prevention and Control. [https://www.ecdc.europa.eu/en/geographical-distribution-2019-ncov-cases](https://www.ecdc.europa.eu/en/geographical-distribution-2019-ncov-cases). Accessed 18 July 15. 15.COVID-19 Map. Johns Hopkins Coronavirus Resource Center. [https://coronavirus.jhu.edu/map.html](https://coronavirus.jhu.edu/map.html). Accessed 18 July 16. 16. P. Schellekens, D. Sourrouille, Tracking COVID-19 as Cause of Death: Global Estimates of Relative Severity (2020). 17. 17. J.-M. Jin, et al., Gender Differences in Patients With COVID-19: Focus on Severity and Mortality. Front. Public Health 8 (2020). 18. 18. L. A. Walter, A. J. McGregor, Sex-and Gender-specific Observations and Implications for COVID-Western Journal of Emergency Medicine 21, 507 (2020). 19. 19. M. G. Netea, J. Quintin, J. W. M. van der Meer, Trained Immunity: A Memory for Innate Host Defense. Cell Host & Microbe 9, 355–361 (2011). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1016/j.chom.2011.04.006&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21575907&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F08%2F04%2F2020.07.31.20165696.atom) [Web of Science](http://medrxiv.org/lookup/external-ref?access_num=000291289500005&link_type=ISI) 20. 20. S. M. Quinn, et al., Anti-inflammatory Trained Immunity Mediated by Helminth Products Attenuates the Induction of T Cell-Mediated Autoimmune Disease. Front. Immunol. 10 (2019). 21. 21. L. E. Escobar, A. Molina-Cruz, C. Barillas-Mury, BCG vaccine protection from severe coronavirus disease 2019 (COVID-19). PNAS (2020) [https:/doi.org/10.1073/pnas.2008410117](https://doi.org/10.1073/pnas.2008410117) (July 21, 2020). 22. 22. J.-F. Bach, The hygiene hypothesis in autoimmunity: the role of pathogens and commensals. Nature Reviews Immunology 18, 105 (2018). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1038/nri.2017.111&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=29034905&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F08%2F04%2F2020.07.31.20165696.atom) 23. 23. P. Schellekens, D. M. Sourrouille, “Covid-19 Mortality in Rich and Poor Countries: A Tale of Two Pandemics?” (Social Science Research Network, 2020) (July 18, 2020). 24. 24. M. Yazdanbakhsh, P. G. Kremsner, R. van Ree, Allergy, Parasites, and the Hygiene Hypothesis. Science 296, 490–494 (2002). [Abstract/FREE Full Text](http://medrxiv.org/lookup/ijlink/YTozOntzOjQ6InBhdGgiO3M6MTQ6Ii9sb29rdXAvaWpsaW5rIjtzOjU6InF1ZXJ5IjthOjQ6e3M6ODoibGlua1R5cGUiO3M6NDoiQUJTVCI7czoxMToiam91cm5hbENvZGUiO3M6Mzoic2NpIjtzOjU6InJlc2lkIjtzOjEyOiIyOTYvNTU2Ny80OTAiO3M6NDoiYXRvbSI7czo1MDoiL21lZHJ4aXYvZWFybHkvMjAyMC8wOC8wNC8yMDIwLjA3LjMxLjIwMTY1Njk2LmF0b20iO31zOjg6ImZyYWdtZW50IjtzOjA6IiI7fQ==) 25. 25. M. Roser, H. Ritchie, E. Ortiz-Ospina, J. Hasell, Coronavirus pandemic (COVID-19). Our World in Data (2020). [https://ourworldindata.org/coronavirus](https://ourworldindata.org/coronavirus). Accessed 29 June 2020 26. 26.GHO. [https://www.who.int/data/gho](https://www.who.int/data/gho). Accessed 29 June 2020 27. 27.World Bank Open Data | Data. [https://data.worldbank.org/](https://data.worldbank.org/). Accessed 29 June 2020 28. 28.Water, sanitation and hygiene - exposure. [https://www.who.int/data/gho/data/themes/water-sanitation-and-hygiene](https://www.who.int/data/gho/data/themes/water-sanitation-and-hygiene). Accessed 29 June 2020 29. 29.GBD Results Tool | GHDx. [http://ghdx.healthdata.org/gbd-results-tool](http://ghdx.healthdata.org/gbd-results-tool). Accessed 29 June 2020 30. 30. A. Zwerling, M. Pai, The BCG world atlas: a new, open-access resource for clinicians and researchers. Expert review of anti-infective therapy 9, 559–561 (2011). [CrossRef](http://medrxiv.org/lookup/external-ref?access_num=10.1586/eri.11.71&link_type=DOI) [PubMed](http://medrxiv.org/lookup/external-ref?access_num=21819321&link_type=MED&atom=%2Fmedrxiv%2Fearly%2F2020%2F08%2F04%2F2020.07.31.20165696.atom)