Second Installment of Preliminary Interrogation by Patricio V. Marquez, Albert Figueras, Allison Ross, Giovanni S. Marquez, and Jaime Bayona

“…government and municipal officials were putting their heads together. So long as each individual doctor has come across only two or three cases, not one had thought of taking action.  But it was merely a matter of adding up the figures and, once this had been done, the total was startling.  In a very few days the number of cases had risen by leaps and bounds, and it became evident to all observers of this strange malady that a real epidemic had set in.” 

Albert Camus, The Plague, 1947

In this post, we build upon the preliminary interrogation described in our previous post “Searching for Evidence in the COVID-19 Era: The BCG Case” (“Searching for Evidence in the COVID-19 Era: The BCG Case”, April 18, 2020, http://pvmarquez.com/bcgcovid19).  

The objective in this post is to present the results of a rapid statistical analysis that we conducted to assess a hypothesis that has been put forward by some researchers about the association between COVID-19 reported deaths and Bacillus Calmette–Guérin (BCG) vaccination policy in countries.

The BCG vaccine, first administered to a human in 1921, is one of the most widely used of all current vaccines, covering >80% of neonates and infants in countries where it is part of the national childhood immunization program. The BCG vaccine has a documented protective effect against meningitis and disseminated TB in children (disseminated TB is a contagious mycobacterial infection in which mycobacteria have spread from the lungs to other parts of the body through the blood or lymph system) (WHO 2020).  

Overall, BCG vaccine efficacy is estimated to be about 51 percent in preventing any TB disease and up to 78 percent in protecting newborns from disseminated or meningeal TB (Canadian Immunization Guide 2014). The BCG vaccine is also of proven efficacy in the control of leprosy.  

The duration of BCG vaccine protection is not well-established, although one study demonstrated a protective effect for as long as 60 years (Nguipdop-Djomo, Heldal, Cunha Rodrigues, Abubakar, and Mangtani 2015). The BCG vaccine, however, will not prevent the development of active TB in individuals who are already infected with M. tuberculosis.

If processes for early identification and treatment of latent TB infection are not available, it is recommended that BCG vaccine may be considered for workers (such as health care workers, laboratory workers, prison workers and those working in shelters for the homeless) who may be repeatedly exposed to persons with untreated, inadequately treated or drug-resistant active TB, in conditions where protective measures against infection are not feasible (Canadian Immunization Guide 2014). 

BCG vaccination policies vary across countries. Countries with high TB prevalence tend to have universal BCG vaccination policies, while those with lower TB burden tend to vaccinate only high-risk groups (SAGE Working Group on BCG Vaccines and WHO Secretariat 2017). Among 180 countries with available data, 157 countries recommend universal BCG vaccination, while 23 countries have either stopped BCG vaccination (due to a reduction in TB incidence), or never recommended mass BCG immunization and instead favored selective vaccination of “at risk” groups (Zwerling, Behr, Verma, Brewer, Menzies, Pai 2011). For example, BCG vaccination in the latter group of countries may be considered in exceptional circumstances, such as for infants in high risk communities, for persons at high risk of repeated exposure, for certain long-term travelers to high prevalence countries, and in infants born to mothers with infectious TB disease.

As noted in our previous post, different studies involving the BCG vaccine have found that it provides heterologous or "nonspecific immunological effects" by altering immune response to pathogens other than Mycobacterium tuberculosis.  The BCG vaccine has been found to reduce non-TB respiratory infection, sepsis, and all-cause mortality among young children, and it may prevent upper respiratory tract infections among the elderly. Although the mechanism, magnitude, and duration of these nonspecific effects remain uncertain, some researchers speculate that BCG vaccination could moderate the severity of COVID-19 infections.

Results of an Additional Analysis

Data and Information Used.  Our team conducted an initial, limited, analysis of the available data and information.  Data on country population size were obtained from the World Bank Data Bank; data on COVID-19 deaths were from the Johns Hopkins University Center for Systems Science and Engineering (CSSE) database, and information on BCG vaccination policies, by country, was from the BCG World Atlas. COVID-19 deaths per capita was calculated using Johns Hopkins CSSE death counts and World Bank population estimates; the data were then logarithmically transformed to ensure normality.

Summary of Results

• An analysis of variance (ANOVA), including both small states and low-income countries, was conducted. Countries were divided into four groups according to four different types of BCG vaccination policy (no policy at all, universal BCG vaccination policy, previous BCG vaccination policy, and policy to vaccinate certain high-risk groups). The analysis showed that there is a statistically significant difference in the mean number of deaths per 1,000 people between country groups (p < .01).

• A second ANOVA was conducted excluding small states and low-income countries. Countries were also divided into four groups according to the four different types of BCG vaccination policy (no policy at all, universal BCG vaccination policy, previous BCG vaccination policy, and policy to vaccinate certain high-risk groups). The analysis showed that there is a statistically significant difference in the mean number of deaths per 1,000 people between country groups (p < .01).

• A two-sample t-test assuming equal variances was conducted, testing the difference in mean COVID-19 deaths per 1,000 population between countries with a universal BCG vaccination policy and countries without it (including no policy, only high-risk groups, and previous universal policy) was conducted. The test included small states and low-income countries and showed that there is a statistically significant difference in the mean number of deaths per 1,000 people between the two groups (p < .01).

• Finally, another two-sample t-test assuming equal variances, testing the difference in mean COVID-19 deaths per 1,000 population between countries with a universal BCG vaccination policy and countries without (including no policy, only high-risk groups, and previous universal policy) was conducted. In this case, small states and low-income countries were excluded. The results similarly showed that there is a statistically significant difference in the mean number of deaths per 1,000 people between the two groups (p < .01).

In conclusion, the results of our initial analyses show statistically significant correlations, both when small states and low-income countries were included and when they were excluded from the analyses.

Despite the interesting results that we obtained, substantial doubts arise and should be taken into account when further assessing the plausibility of the hypothesis:

• At present, there are a lot of questions about how different countries are reporting COVID-19 deaths. It seems that some countries only count deaths occurring in hospitals but not those happening in nursing homes, prisons, or at home (because testing was not done to confirm the diagnosis).  Moreover, large number of countries are facing supply-side challenges related to the availability of test kits for case detection and contact tracing.  The latter may have a confounding effect on the observed results.

• BCG vaccination policies and practices vary across time and countries.  For example, it is possible that a country with an 'only high-risk group' vaccination policy at present had a 'universal vaccination policy' up to 1960. In cases like this, there are differences in previous BCG vaccination coverage according to age groups. A proper analysis should take into account these inter- and intra-country differences.  In addition, the use of different BCG vaccine strains in different countries and even within the same countries may also have a confounding effect.  Currently, five main strains account for more than 90 percent of the vaccines in use worldwide with each strain possessing different characteristics, and there is no standardized production of BCG vaccine between manufacturers (WHO 2012).

• The role of additional factors that have not yet been assumed, investigated, or confirmed. For example, since the primary function of the respiratory system--to supply oxygen to all the parts of the body--is especially affected among COVID-19 infected individuals (probably because of multiple clotting in the lungs), does the decreased respiratory capacity observed among cigarette smokers play a major role in the clinical evolution of COVID-19 disease, increasing the risk of disease severity and death among COVID-19 infected individuals? As discussed on a previous post, cigarette smoking makes chronic lung diseases more severe and increases the risk for respiratory infections (“Does Tobacco Smoking Increases the Risk of Coronavirus Disease (Covid-19) Severity? The Case of China” of March 3, 2020, http://pvmarquez.com/Covid-19).

The Way Forward

Many clinical trials with different drugs treat COVID-19 disease patients are currently being conducted in many countries. Antivirals (lopinavir/ritonavir, lopinavir), monoclonal antibodies (tocilizumab) and other agents affecting the immune response (hydroxychloroquine and chloroquine) are being tested. Different treatments are also being tested as preventive measures in non-positive patients.

It should be clear, however, that research to find an effective vaccine is one of the most promising ways to suppress the pandemic (not the present wave, but future waves that may happen).

At this stage, it is crucial to try to understand sets of factors and their interactions, e.g., risk factors, protective factors, and infection periods.  Within this framework, the BCG hypothesis could be useful. There is a plausible hypothesis: vaccinated people could show some immunity advantage over non-vaccinated people (based on previous studies on the nonspecific effects of BCG on the immune response as described in our above-mentioned previous post of April 18, 2020).

Conducting more accurate analyses, including evaluating differences in BCG vaccination among infected people, recovered patients, and positive but asymptomatic patients, should be of high interest, not only to understand how COVID-19 interacts with people but also to inform the development of a vaccine (or to understand potential failures or successes of future vaccines).  Another avenue of analysis that may offer some clarity on this inquiry is related to the fact that some medical personnel working in develop countries, such as the United States (without BCG), are actually coming from low-and middle-income countries where BCG policy is in place. Hence, a question that may need to be analyzed is whether there are differences in terms of severity of COVID-19 disease among infected doctors and nurses based on the country of origin of medical personnel that emigrated and those who emigrated and are working in medical facilities in developed countries?

At this point, it is crucial is to ensure that decisions taken on BCG vaccination are taken based on the available information for several reasons:

• First, because it takes several weeks to develop the alleged positives effects of BCG on immunity after vaccination (so, it would not be useful for the present wave).

• Second, because of the low availability of BCG vaccines and their low production rate (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5777639/).

• Third, and most important, because the hypothesis has to be tested with more appropriate epidemiological and clinical methods before a causal relationship can be ascertained.