Exercising caution in correlating COVID-19 incidence and mortality rates with BCG vaccination policies due to variable rates of SARS CoV-2 testing

Janine Hensel; Daniel J McGrail; Kathleen M McAndrews; Dara Dowlatshahi; Valerie S LeBleu; Raghu Kalluri

As of April 7 th , 2020, 211 countries around the world had reported COVID-19 cases. Among those, 81 countries had more than 500 confirmed COVID-19 cases. Using a 500 confirmed COVID-19 cases cut-off, we ascertained whether COVID-19 incidence and associated mortality is correlated with BCG vaccination policies. Three countries (Bahrain, Serbia, and Iceland) did not have information on vaccination policies. Out of the remaining 78 countries ( Table 1) , 8% of the countries never had a universal BCG policy ('Never universal GCG policy', Figure 1A) . Belgium, Netherlands, Canada, Italy, Lebanon, and United States of America currently never had a universal BCG policy, yet all these countries may offer vaccination for high risk subgroups (healthcare professionals and children with parents from high risk countries). 23% of the analyzed countries had a universal BCG policy in the past, including many European countries, Australia, Ecuador, and Israel ('Universal BCG policy in the past', Figure 1A ). 69% of the countries currently have a universal BCG vaccination policy, including most countries in Central America, South America, Africa, and Asia ('Current universal BCG policy', Figure 1A ). The number of total COVID-19 cases (with a 500 cut-off) in each of the three groups is shown in Figure 1B . The number COVID-19 cases per 1 million (M) inhabitants was significantly reduced in countries with current universal BCG policy when compared to countries with a never-or in the past-universal BCG policy ( Figure 1C and D). The percent mortality per total COVID-19 cases was significantly reduced in countries with current and past universal BCG policy when compared to countries that never had a universal BCG policy (Figure 1E and F) . These analyses however indicated that other factors may confound the results, and included population density, median age, TB incidence, testing rate, and urban population. The following analyses addressed these potential confounders.