The morbidity and mortality of COVID-19 are associated with ABO and Rh blood groups.

Dear Editor

We read with interest the paper of Dai (2020) about the association of ABO blood groups with coronavirus disease (COVID-19) severity and cardiovascular diseases. Examining morbidity and mortality measures of COVID-19 from various countries reveals a considerable geographical variation. We have suggested that the genetic background of the populations might be one of the responsible factors in this context. The author hypothesized that individuals with A blood group are more likely to develop severe COVID-19 once infected, compared with patients having O phenotype. He has also suggested that ABO blood groups are not risk factors for acquiring the SARS-CoV-2 infection. We conducted an ecologic study to determine whether the ABO and Rh phenotypes are associated with the increased risk of getting COVID-19 and/or severity of the disease. Prevalence, mortality (both per 106 population) and fatality (per 100 infected cases) data (on 23 May 2020) as the main epidemiologic measures of morbidity and mortality of COVID-19 were obtained for 86 Asian, European, African and American countries, in conjunction with the number of diagnostic tests which have been performed (per 106 population) in each country (https://www.worldometers.info/coronavirus/#countries). World distribution of ABO and Rh blood groups was also determined (http://www.rhesusnegative.net/themission/bloodtypefrequencies). There are considerable differences in overall health and economic situation of the countries. These differences could affect the capacity for testing, recording and reporting the disease in each country and consequently might be considered as potential confounders. Therefore, data for potential confounding variables including life expectancy at birth in years, gross national income (GNI) per capita, density of medical doctors, density of nursing and midwifery personnel, and age-standardized prevalence of tobacco smoking among persons aged 15 years and older in each country were collected (www.who.int/countries/en/).

The data (Supplementary Material Table S1 online) were checked for normality using the Kolmogrov–Smirnov test and non-normally distributed variables including epidemiologic measures of disease (prevalence, mortality and fatality), the total number of diagnostic tests and GNI were log-transformed. Univariable and multivariable linear regression analysis were used for analyzing the data. Variables with the p-value ≤0.10 in the univariable analysis were introduced into the multivariable model. Epidemiologic measures of disease were considered as outcome variables, and blood group frequencies as well as the above-mentioned potential confounders were introduced into the model as explanatory variables. Three distinct models (one for each epidemiologic measure) were constructed. A backward elimination procedure was used for model construction and a two-tailed p-value less than 0.05 was considered significant in the final model.

Descriptive statistics for study variables (Supplementary Material Table S2) indicate considerable variation in morbidity and mortality measures as well as in blood groups distribution in various countries. In univariable analysis (Supplementary Material Table S3), A, B and Rh– phenotypes had significant associations with all disease measures (p ≤ 0.004; ). The direction of association was positive for the A and Rh– blood groups while it was negative for the B phenotype. Results for the final multivariable linear regression models are displayed in . For log-fatality of COVID-19, only frequency of A blood group remained in the final model, indicating a partial correlation coefficient of 0.39, identical to the univariable analysis (). In the multivariable regression model for log-mortality, partial correlation coefficients for Rh– and B phenotypes, adjusted for life expectancy at birth, were 0.31 and –0.31, respectively.

Figure 1

Association of mortality and morbidity measures of COVID-19 (on 23 May 2020) with frequency of ABO and Rh-negative blood groups in 86 countries.

Table 1

Multivariable linear regression analyses for association of epidemiologic measures of COVID-19 with frequencies of A, B and Rh– blood groups in various countries in the world.

	Unstandardized coefficients		Standardized coefficients	Partial correlations	t	p
Variables	B	Standard error
Log-prevalence as dependent variable
Constant	1.213	0.290	–	–	4.175	<0.001
Log-test	0.487	0.060	0.630	0.686	8.165	<0.001
B phenotype	–2.243	0.585	−0.296	−0.405	−3.835	<0.001
Log-mortality as dependent variable
Constant	–2.195	0.927	–	–	–2.367	0.021
LE	0.047	0.012	0.382	0.426	4.054	<0.001
B phenotype	–2.571	0.912	–0.266	–0.312	–2.821	0.006
Rh– phenotype	3.445	1.213	0.263	0.313	2.839	0.006
Log-fatality as dependent variable
Constant	–0.144	0.174	–	–	–0.832	0.408
A phenotype	1.840	0.488	0.388	0.388	3.769	<0.001

The first model was significant with F = 58.82; df = 2, 75; p < 0.001; adjusted R2 = 0.574. The second model was significant with F = 27.740; df = 3, 74; p < 0.001; adjusted R2 = 0.510. The third model was significant with F = 14.208; df = 1, 80; p < 0.001; adjusted R2 = 0.140. Log-test: log-number of performed COVID-19 test per 106 population.

LE: life expectancy at birth.

Fatality is an index of the severity of the COVID-19 infection and the results of the present study reveal that the severity of disease is higher in countries where A phenotype is more frequent. Prevalence, on the other hand, is an index of risk for acquiring the infection and, based on our results, the risk is higher in countries where B blood group is less common while no association with A phenotype was observed. Both of these observations are consistent with the hypothesis proposed by Dai.

The association of mortality of COVID-19 with Rh– phenotype is a novel finding which needs further consideration. In order to investigate whether A and Rh– phenotypes are highly correlated with each other or Rh– could be considered as an independent risk factor for COVID-19 mortality prognosis, further analysis was done. The correlation between A and Rh– phenotypes was significant (r = 0.73, p < 0.001). To explore the possible independent role of Rh– phenotype with mortality, the associations of A– and A+ phenotypes with log-mortality were determined. Interestingly, correlation was higher for A– (r = 0.62, p < 0.001) compared with A+ phenotype (r = 0.40, p < 0.001), which is in favor of the independent role of Rh– blood group. Nevertheless, due to inherent limitation of the ecologic studies in which outcome and risk factors are measured on aggregate level, no definite conclusion could be reached at this time. In future studies based on individual data for COVID-19, ABO and Rh blood groups, it is possible to obtain a more conclusive result about the Rh– phenotype.

Generalization of the results of an ecologic study directly to individuals is an ecologic fallacy. However, the results of an ecologic study could be used for generation of hypotheses. It has been reported that COVID-19 patients without a history of cardiovascular disease are at risk for cardiovascular complications and hemostatic abnormalities. The association of ABO blood groups with vascular disorders and thrombosis has a long history and partly attributed to the higher level of von Willebrand factor and consequently of factor VIII. Therefore, in addition to the author hypothesis which specifically referred to the cardiovascular events in hypertensive individuals with COVID-19 infection, a more holistic approach to vascular disorders in this novel infection is warranted in future research. Indeed, the field for pursuing new research hypotheses and comprehensive research for determination of the role of ABO and Rh blood groups in the pathogenesis of these events in COVID-19 patients is completely open.

Supplemental Material

Supplementary material is available at European Journal of Preventive Cardiology online.

Author contribution

Both authors contributed equally to this work.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.