This study examines a previously observed association between the proxy measures of immunity from BCG of countries and the morbidity and mortality from COVID-19. We explored this association further by adopting the TB incidence in different age-groups as a surrogate of the BCG coverage, as no age-specific BCG coverage data are currently available. Our multivariate analysis, in summary, identified the presence of a universal BCG vaccination policy to negatively affect both the cases and deaths from COVID-19. COVID-19 deaths were further affected negatively by the BCG coverage of population and positively by the incidence of TB in > 65-year olds.
The impact of BCG vaccination on transmission of Mycobacterium tuberculosis is limited, as it does not prevent the reactivation of latent pulmonary infection, which is the principal source of bacillary spread in the community. There is significant variation in the effectiveness of BCG vaccination, and prior exposure to environmental mycobacteria is thought to influence this (17). This variation may also explain the inconclusive evidence shown on the protection of BCG vaccination against COVID-19. In contrast, reduced incidence of TB is a more direct indicator of the efficacy of BCG vaccination. This may explain our observations.
The presence of a universal BCG vaccination policy showed a stronger correlation than the BCG coverage, for both cases and deaths from COVID-19. In concurrence, in the study by Sala et al. (9), BCG coverage only accounted for 12.5% of the variance observed. The results of Miller et al. (18) suggest that factors increasing the fraction of the elderly population being vaccinated against BCG, like an earlier year of initiation of a universal BCG vaccination policy, has a good correlation with COVID-19 cases and deaths.
BCG vaccination is believed to confer a non-specific increase in immunity (19). It acts via both innate and adaptive immune responses, and in the latter, via both T and B cells. This theory is applied when BCG is used as immunotherapy in the management of melanoma and bladder cancer (20). BCG induces long-lived memory B-cells (21). BCG also induces and maintains tissue-resident CD4 T cells in the lungs (22). The COVID-19 is presumed to enter the body through the lungs (21). These tissue-resident CD4 cells may therefore influence the entry into the body and therefore contribute to lower the cases and deaths in countries with BCG vaccination. A clinical trial has already begun testing the utility of BCG for boosting immunity against COVID-19 (23).
Over 90% of the TB cases in the elderly are due to reactivation of primary infection (24). Rarely, previously infected older persons who have eliminated the viable tubercle bacilli may revert to a “naïve” immunologic status and are at risk of reinfection (25). These mechanisms may imply the loss of the initial protection from BCG vaccination. The better correlation of COVID-19 cases and deaths with the TB incidence in older ages groups than the younger groups also support this. The changes occurring in the immune system with ageing may also contribute to this (26, 27). The reduced incidence of TB could also represent an overall improvement in preventive care. The results we observed may be related to an unknown organism or antigen closely correlated to both TB and COVID-19, similar to the hygiene hypothesis (28).
The mortality from COVID-19 is reportedly higher in patients who are older and comorbid (1, 29). In the study by Zhou et al. (1), the interquartile range (IQR) of the age of survivors and non-survivors was 45–58 and 63–76 years, respectively. This shows a clear demarcation with possibly no-overlap. Our analysis only identified the TB incidence in > 65-year olds to be significantly negatively correlated with deaths. The incidence in the total population and the 15-64-year group showed smaller correlations We used the incidence of TB is a surrogate marker of the immunity offered by the BCG vaccination. Lesser deaths observed in populations where there is a lesser number of TB in the > 65-year olds support this hypothesis. There is no worldwide age-specific case and mortality data for COVID-19 and we were therefore unable to investigate this hypothesis further.
The relationship between the income and COVID-19 cases is paradoxical to observations during previous epidemics (30, 31). High-income countries generally depend on larger budgetary allocations for secondary preventive strategies to contain an epidemic within the health system capacity. Their outcomes during the COVID-19 pandemic has remained worse than countries with lower income due to a multitude of reasons. The income level remained independently associated with the case number in our multivariate analysis and has been reported by other researchers as well (3, 8, 18, 32). In the subgroup analysis, we observed that, for high-income countries, the correlation between cases and deaths and the degree of BCG coverage was higher. Most high-income countries do not have a universal BCG vaccination policy, but income level and BCG coverage were independent predictors in our multivariate analysis. One possible contributing factor to the higher cases and deaths in the high-income countries could be the higher median age of the population in these countries (33). A positive correlation between the median age of the population and COVID-19 morbidity and mortality has been reported (32).
We observed minimal correlation between the CFR and the variables studies. The CFR can be affected by the lack of testing to identify cases. There is a significant difference in the testing per-capita among countries, i.e. high-income countries performing more tests per million population (34). There is also a negative correlation between the percentage of the population tested and positive rate (32). This may have caused the apparent reduction in CFR compared to the deaths per 100,000 population in the high- and upper-middle-income countries. The lag-between the deaths and the cases may also contribute to this (8), as the pandemic is still evolving. The CFR, therefore, may be unreliable in the COVID-19 pandemic.
The main limitation of this analysis is the inability to link the exposures relate to BCG vaccination to the occurrence of COVID-19 in the same person (ecological fallacy) as the data considered are at population level. Further, there may be other factors between countries that are associated with the exposure, which may account for the differences, such as comorbidities (1) and other epidemiological characteristics of the vulnerable groups, which have not been considered in the model. Several countries who do not have a universal BCG vaccination at present have stopped the vaccination in the last 20-years. Their most recent estimated BCG coverage is still high. We have also not included other variables like the temperature and humidity, which have already been proven to be associated with COVID-19 (35).