Introduction: Red blood cell (RBC) alloimmunization is one of the most common complications of RBC transfusion. Despite the fact that alloimmunization impacts many transfusion recipients, the lack of standardized alloimmunization registries in the US has prevented large studies from being completed. Through the development of a uniform electronic research database, we investigated alloimmunization rates and diagnoses in inpatients and outpatients across the US.

Methods: The REDS-III recipient database, established in 2013, houses demographic, diagnostic, transfusion, laboratory, and other information on inpatients and outpatients at 12 hospitals in 4 geographic regions. For this study, an antibody "responder" (R) was defined as an individual who had a putative transfusion (or pregnancy)-associated red cell antibody identified by screen at any point during the study period, independent of documented RBC exposure or pregnancy during the study period. An antibody "non-responder" (NR) was defined as an individual who was transfused with at least one unit of RBCs during the study period, and had a negative antibody screen at least 15 days post-transfusion. Anti-D antibodies were excluded from studied females, given an inability to distinguish passively administered anti-D in the database from authentic alloantibodies. Demographics, blood type, ICD9/10 codes, and antibody screen results were compared for R and NRs. Odds ratios were calculated for the likelihood of being a R or NR, based on the presence or absence of select chronic disease states.

Results: 612,417 patients were recorded in the REDS recipient database from 2013-2016. 318,918 unique patients in the database had an RBC antibody screen during the study period. Of these, 6597 had at least one transfusion-associated RBC alloantibody identified by routine antibody screen (overall alloimmunization rate among all screened patients independent of RBC exposure = 2.07%). 63.8% of the R were female (alloimmunization rate 2.97%) and 36.2% were male (alloimmunization rate 1.35%). 30,468 unique patients (49.8% female and 50.2% male) were NR; the mean RBC exposure during the study period was similar between R and NR (mean, s.d., and median of 8.8, 18.2, and 4 units versus 9.3, 14.3, and 5 units). Approximately 60-70% of the antibodies in R were pre-existing, and the remainder were generated during the study period. Few children under 10 years of age were R compared to older age groups; an equation of R/R+NR = 26/26 + 467 = 5.3%. There were no differences in R vs NR by A/B/O blood type, though Rh negative patients (R/R+NR of 1175/1175 + 3901 = 23.1%) were significantly more likely to be R than Rh positive patients (R/R + NR of 5365/5365 + 26544 = 16.8%). .A total of 8892 clinically significant antibodies were detected, including 4749 (53.4%) in the Rh family (2588 anti-E, 780 anti-C, 556 anti-c, 105 anti-e, 525 anti-D), 1937 (21.8%) in the K family, 912 (10.3%) in the Kidd family, and 669 (7.5%) in the Duffy family, among others. Patients with sickle cell disease (n=554, OR 2.6, p<0.0001), sickle cell trait (n=117, OR 1.8, p=0.02), `coronary atherosclerosis or heart disease (n=12,403, OR 1.2, p<0.0001), lupus (n=969, OR 1.3, p=0.002), rheumatoid arthritis (n=1213, OR 1.2, p=0.01), and myelodysplastic syndrome (n=1162, OR 1.3, p=0.0008) were more likely to be R than NR. In contrast, patients with leukemia (n=2801, OR 0.5, p<0.0001), solid tumors (n=17,513, OR 0.6, p<0.0001), and solid organ transplant (n=1491, OR 0.7, p=0.0001) were less likely to be R than NR.

Conclusions: Data collected in this multi-center recipient database provide the largest RBC alloimmunized patient cohort ever studied in the US. The 2.07% overall rate of alloimmunization is similar to that reported in other retrospective studies, with females having higher alloimmunization rates than males. Rh negative individuals were significantly more likely to be alloimmunized than Rh positive patients, despite the exclusion of anti-D antibodies in females. Other unexpected findings include the predisposition to alloimmunization observed in patients with sickle cell trait and coronary artery disease. Ongoing studies are investigating the timing of RBC exposure to new antibody formation in this patient cohort, as well as potential physiologic explanations for the chronic disease associated findings.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.