Results of the current study suggest that patients with blood group 'B' have a four-fold increased risk of developing severe infection. In addition, it also reiterates the observation that 'O' blood group was significantly associated with a decreased risk of severe malaria. Other blood groups ('A' and 'AB') did not show any association.
The role of ABO blood group in malaria has been investigated in various populations, but robust data is limited . This study was an attempt to analyse the association of ABO blood group in large number of adult patients and healthy controls in Odisha, a state, highly endemic for falciparum infection . Higher prevalence of blood group 'O' was observed in uncomplicated cases, an indication of its possible protective property against severity as indicated in previous reports [[1, 11–13]. The mechanism of protection is not clearly understood. It is postulated that the phenomenon of rosetting is one of the mechanisms that contributes to disease severity . This rosetting capacity varies among different blood groups . Lowest rosette formation is observed in blood group 'O' individuals. This study also clearly highlights the association of severe malaria with blood group 'B'. Interestingly, this association was valid across all grades of severity. Although the number of patients who died (n = 20) was small, frequency of blood group 'B' was higher in these subjects (80%) in comparison to survivors (54%). Previously, studies on patients from Zimbabwe , Gabon  and Ethiopia  showed a significant association of 'A' blood group with severe malaria. Blood group 'AB' has also been reported to be associated with severity in Sri Lanka , Mali  and Ethiopian populations . There are no reports implicating blood group 'B' with severity. The variability of observations made with regard to different blood groups may be attributable to different rosetting capacity, heterogenous population groups and varied infective strains . Blood group 'A' in Uganda and Gambia [4, 7], 'B' group in Thailand  and 'AB' group in Kenya  have been associated with increased rosetting phenomenon.
It is presumed that the prevalence of blood group 'O' would be higher in malaria endemic areas due to its capacity to confer protection. An analysis of blood group in healthy controls revealed a distribution of 'O' (42%), which was much higher compared to 'A' (21%), 'B' (29%) and 'AB' (8%). A community-based study in a tribal population of Odisha, where malaria is endemic, also showed higher prevalence of blood group 'O' . Significantly, a lower prevalence 'O' blood group has been observed in other malaria non-endemic states like Maharashtra [26, 27] and Uttar Pradesh  in India, indicating a selective advantage of this blood group in endemic localities. This hypothesis is further supported by higher prevalence of 'O' blood group worldwide where malaria infection is prevalent .
Meta-analysis combines results of several similar studies to produce a single estimate of the major effect with enhanced precision . Current analysis revealed significant association of blood group 'O' with protection against severe malaria (OR = 0.45) In contrast, blood groups 'A' and 'AB' were associated with susceptibility to severity: Blood group 'A' and 'AB' conferred 1.27- and 1.74- fold higher risks respectively. Although in the current study, 'B' blood group was significantly associated with severe malaria on meta-analysis the association was insignificant. This variation may be a population specific phenomenon.
There are however limitations in the present study. Several RBC polymorphisms, including those linked to glucose-6-phosphate dehydrogenase, pyruvate kinase, complement receptor-1 and haemoglobinopathies, have a role in the clinical outcome of malaria , but were not included for analysis in the present study. Such polymorphisms have only been reported from the Western belt of Odisha among tribal communities [32, 33] and not in areas from which patients and HC in the current study were enrolled.