ABO Blood Groups and the Incidence of Complications in COVID-19 Patients: A Population-Based Prospective Cohort Study

After a COVID-19 outbreak in the Falles festival of Borriana (Spain) during March 2020, a cohort of patients were followed until October 2020 to estimate complications post-COVID-19, considering ABO blood groups (ABO). From 536 laboratory-confirmed cases, 483 completed the study (90.1%) carried by the Public Health Center of Castelló and the Emergency and Microbiology and Clinical Analysis of Hospital de la Plana Vila-real. The study included ABO determination and telephone interviews of patients. The participants had a mean age of 37.2 ± 17.1 years, 300 females (62.1%). ABO were O (41.4%), A (45.5%), B (9.1%), and AB (3.9%). We found no difference in the incidence of COVID-19 infections. A total of 159 (32.9%) patients reported one or more post-COVID-19 complications with divergent incidences after adjustment: O (32.3%), A (32.6%), B (54.1%), and AB (27.6%); B groups had more complications post-COVID-19 when compared with O group (adjusted relative risk [aRR] 95% confidence interval [CI] 1.68, 95% CI 1.24–2.27), and symptoms of fatigue (1.79, 95% CI 1.08–2.95), myalgia (2.06, 95% CI 1.10–3.84), headache (2.61, 95% CI 1.58–4.31), and disorder of vision (4.26 95% CI 1.33–13.60). In conclusion, we observed significant differences in post-COVID-19 complications by ABO, with a higher incidence in B group. Additional research is justified to confirm our results.


Introduction
Ongoing symptomatology in COVID-19 patients after the acute phase of the illness is frequent. It has been characterized as a syndrome of long or persistent COVID, affecting both children and adults [1][2][3]. The research on the risk and protective factors of infection and severity of COVID-19 is crucial in the second year of the pandemic. Thus far, many potential factors have been considered [4,5].
Among these factors, the ABO blood groups (ABO) are being studied with intensity [6][7][8][9], and it has been recommended to determine the ABO of COVID-19 patients to improve medical care [10].
There are, however, conflicting results about the role of ABO and the risk of COVID-19 disease. Some studies found the A and B blood groups related to a higher risk of infection, complications, and mortality [11][12][13][14][15]. In contrast, other studies have not found differences in disease severity among the O and A groups [16], whereas other studies have found the O group as protective, compared to the non-O blood groups. Finally, large and well-designed studies have not found that the ABO blood groups are a risk factor for COVID-19 infection or severity [17,18].
Several conditions could explain this situation of conflicting results, including diversity of sample size, effect size, multiple confounders, publication bias, or even chance distributions [17,19,20]. Complications considering ABO and the follow-up of patients with a mild illness have been less studied [21][22][23][24].
This study aimed to estimate the frequency of persistent symptoms and complications in COVID-19 patients and its associations with ABO blood groups.

Materials and Methods
After a COVID-19 outbreak in the Falles festival with several mass gathering events (MGEs) in Borriana (Spain) during March 2020 [25], we followed a population-based prospective cohort of patients until October 2020 [26] to estimate disease evolution and incidence of complications post-COVID-19, and its relationship with participants ABO blood group. The study was carried by the Public Health Center of Castelló, and the Emergency and Microbiology and Clinical Analysis Services of Hospital de la Plana Vila-real (Spain).
In October 2020, health staff of the Hospital de la Plana Vila-real, and health centers of Borriana, Vila-real, Onda, and La Vall d'Uixo conducted a telephone interview of each participant to obtain information about their health situation, medical assistance, illness's evolution, symptoms post-COVID-19, and duration. In addition, we acquired data from the May-June questionnaire on age, sex, weight, height, body mass index (BMI) (kg/m 2 ), occupation, level of physical exercise, smoking habits, consumption of alcohol, chronic illness, and COVID-19 exposures. COVID-19 exposition included the following features: see a person with a cough at MGEs, attendance MGEs ≥ 2, contact with a COVID-19 case, and family with COVID-19 case.

Statistical Methods
The expected incidence of COVID-19 infection by the ABO was estimated from the distribution of ABO in three sources: two blood donor studies of Catalonia and Navarra [13,29] and one study of an active general population of the Spanish Mediterranean zone [30]. We compared the observed ABO distribution with the expected distribution through Fisher's exact test. Distributions of ABO from the three sources were applied to the 1338 participants in the COVID-19 outbreak study [25] to obtain an incidence rate by each ABO. We used the O group as the reference to estimate relative risks (RR) and 95% confidence interval (CI) by Poisson regression.
Reported complications and symptoms were the dependent variables, and ABO was the predictive variable. We used the Chi2, Fisher's exact test, Wilcoxon matched-pairs signed-ranks test, and Kruskal-Wallis tests for unadjusted comparisons. After a review of medical literature, we used directed acyclic graphs [31], and identified age, sex, lifestyle, and COVID-19 exposures as confounding factors [32], and adjusted our models accordingly using inverse probability weight regression [33]. We performed all calculations with the statistical program STATA ® version 14.
Following up this cohort was part of the public health surveillance as a prolongation of the COVID-19 outbreak control measures [25] and the response of the COVID-19 pandemic. It was exempted from Ethics Review Board approval's protocol according to the Spanish legislation. The study was approved by the director of the Public Health Center of Castelló and the management of the Health Department of La Plana. All participants or the parents of the minors provided their informed written consent to be included in the study.

Results
The participation rate was 90.1% (483/536) considering the patients with laboratoryconfirmed COVID-19 in the COVID-19 outbreak. The mean age was 37.2 ± 17.1 years (rank 1-81), with 300 females (62.1%) and 183 males (37.9%). ABO blood groups distribution was O (41.4%), A (45.5%), B (9.1%), and AB (3.9%). The subjects' characteristics by ABO are shown in Table 1. We found no appreciable differences among ABO by demographic, lifestyle, and COVID-19 exposure.  We found no difference in the incidence of SARS-CoV-2 when comparing the observed proportion of ABO and the expected proportion from blood donors of Catalonia (p = 0.247) and Navarra (p = 0.089), and an active general population (p = 0.366) ( Table 2). However, RRs of no-O groups presented a higher incidence than the estimated O group with a range of 1.06-2.06 and was significant when applying the two blood donor sources to the participants in the COVID-19 outbreak but not when using the active general population.

Variables ABO Blood Groups n = 483 p-Value
Symptoms and duration of the COVID-19 illness among ABO are shown in Table 3. B group experienced a higher proportion of symptomatic disease, medical consultation, hospitalization, and a longer duration of illness. B group reported a higher presence of symptoms such as diarrhea, vomits, weakness, headache, and myalgia. Fever and loss of smell/taste presented significant differences with the other blood groups, (p = 0.020) and (p = 0.016), respectively. There was no difference in habitual health status before and after the COVID-19 considering the ABO blood group. We found significant differences in the O and A groups between the before and after the illness in health status (p = 0.000) and (p = 0.001); in contrast, little to no difference for the B group (p = 0.180) and AB group (p = 0.414). Table 2. Observed distribution of COVID-19 cases of ABO blood groups, incident rate, and relative risk comparing with blood donors, and an active general population from the Spanish Mediterranean zone as the reference population. Borriana COVID-19 cohort 2020.

Variables
ABO Blood Groups n = 483      (Table 4) with a mean duration of 160.9 ± 45.6 days. The incidence of complications was higher in the B group (50%), compared to the other ABO (31.2%) with a marginally non-significant difference (p = 0.072). In October 2020, 81.8% of patients (395/483) had recovered to their former health, but recovery was less frequent in the B group (70.5%), compared to the other ABO (82.9%) (p = 0.047). In addition, 83.2% (402/483) reported that their health status was the same as before the illness; this was also lower in subjects of the B group (70.5%) when compared to the subjects with other ABO (84.3%) (p = 0.025).
In October 2020, 53.4% of patients reported at least one symptom (258/483) without difference among ABO (p = 0.586) ( Table 4). Reported symptoms of fatigue, abdominal pain, muscle pain, loss of smell/taste, headache, the difficulty to solve simple math operations, and skin lesions was higher in the B group. The AB group had a lower incidence of complications and reported symptoms than the B group, and compared with the O group, had high recovery and return health status as before.  We show adjusted incidence rate (aIR) and adjusted RR (aRR) of complications and reported symptoms in Table 5. The B group presented higher aIR than the other ABO in complications, lower recovery, and return health status as before with aRR of 1.68 (95% CI 1.24-2.27), 0.86 (0.70-1.03), and 0.85 (0.71-1.04).  33-13.60), and medical consultation in the acute illness (56.7% versus 42.5% aRR = 1.33 95% CI 1.01-1.75). Loss of the smell/taste was higher in the B group but with a marginally non-significant difference (26.0% versus 14.8% aRR = 1.75, 95% CI 0.95-3.23). Brain fog, as a summary of mental symptoms (Table 5), was higher in the B group but not significant (18.1% versus 12.3% aRR = 1.46 95% CI 0.75-2.73). On the other hand, the AB group's small size prevented obtaining valid estimations.

Discussion
Our results suggest that ABO had not an appreciable impact on the incidence of COVID-19 infection, with the B-group subjects experiencing a higher incidence of complications and reported symptoms than the O group at six months after the acute COVID-19 illness.
Regarding the incidence of SARS-CoV-2 infection in ABO, our results are in line with several studies where no effect of ABO could be found with suitable controls for comparison [17,18,34,35]. The protection of the O group among ABO, which could be found in several studies [29,36,37], is overcome following the Ellis model of ABO incompatibility and SARS-CoV-2 transmission [38]; in the first stages of the epidemic, the non-O groups are more infected, but after, when the majority of the population is infected, the O group has a similar infection rate.
Most ABO studies have been focused on the incidence and severity of COVID-19 [17], and few studies on symptoms and complications post-COVID-19. Some studies found that the O group offered protection against COVID-19 infection, whereas A-group subjects experienced a higher risk of adverse outcomes [12,13,39,40], with more cardiovascular complications in the A-group subjects [41], whereas subjects in the B group experienced more severe illnesses and higher fatality rates [11,15,22,41,42], but no ABO blood group differences in mortality were found in other studies [43][44][45].
In our study, the B-group subjects reported more symptoms and severity of the disease than the other groups. This pattern has been associated with post-COVID-19 complications [46,47]. Complications and reported symptoms here are concordant with the post-COVID-19 disease despite a mild illness in most patients. Hair loss, fatigue, smell/taste loss, headache, muscle pain, insomnia, and anxiety symptoms agree with the post-COVID-19 disease definition [48,49]. In general, physical symptoms (fatigue, muscle pain, headache, lost smell/taste, and vision disorder) had a higher incidence in the B group than mental symptoms (anxiety, depression, insomnia) when compared with other ABO. No difference in complications and reported symptoms were found between O and A groups, in contrast with other studies [23,[50][51][52].
Le Pendu et al. [53], in a review of the association between ABO and COVID-19, concluded that the O group provided some protection in comparison with non-O groups, and this could be mediated either by natural anti-A and anti-B antibodies or by a lower efficiency of furin cleavage in the O group [54,55]. In addition, Bloch et al. [56] found that blood donors of group B had higher neutralizing antibody titers than the other ABO. According to this study [56], a possibility is cross reactivity of the virus and the B antigen, and this increases antibody production, but the cause is unknown. In our previous study [26], the O group presented lower persistence of anti-SARS-CoV-2 antibodies than the non-O groups, but we found no difference in antibody levels between ABO blood groups.
Concerning infectious diseases, the O group could have special protection derived from anti-A/anti-B actions of humoral innate immunity [57,58]. In addition, the O group had some physiological advantages such as endurance running, compared with non-Ogroups [59]. ABO differences in respect to health and disease are not entirely understood. Still, some studies found that O group subjects have lower thrombosis and more hypertension disease than no-O groups, and the A group has more risk of cardiovascular disease [60][61][62].
The study had some important aspects, including a population-based prospective cohort design, high participation rate, a follow-up of 6 months, and control of potential confounding factors, considering that the exposure to the virus and the duration time could play a crucial role in the clinical course of the disease [63].
The study has some limitations; measured manifestations were reported by questionnaire, so we could not discard information, recall bias, and residual confounding. COVID-19 is a new disease, and some aspects could not be considered in advance. Most patients had a mild illness, and they may not be representative of the COVID-19 patients. A small sample of the AB group prevented definitive estimations in this blood group. The study was focused on COVID-19 patients, and we could not compare our findings with the COVID-19 negative member of the cohort.
Our results support the active medical follow-up of COVID-19 patients considering the high level of symptoms persistence [64]. In our study, the B group appears associated with a higher risk of prolonged symptoms, whereas the O group subjects experienced lower affectation. Considering that there are few studies on ABO blood groups and COVID-19 complications, further research is justified to improve our understanding of the ABO relationship with COVID-19 [65][66][67].

Conclusions
ABO blood group patients presented significant differences in post-COVID-19 complications with a more severe course observed in the B group. Additional research is justified to confirm our results.