Background

Transfusions that are ABO compatible but not group identical (mismatched) are given for a variety of reasons including inventory availability, avoiding wastage from outdating, and clinical urgency. A recent observation at our centre suggested that patient outcome was different for those patients that received a transfusion of units with a compatible but mismatched ABO group compared to those receiving ABO group identical blood. Hence, we performed a retrospective hospital registry study to explore the association between mismatched blood and in-hospital mortality in transfused patients.

Study Design

Our patient/blood utilization database included 35,487 transfused hospitalized patients from 3 acute care academic centres from April 1, 2002 to October 31, 2011. Information on transfused RBCs included duration of storage (days) and ABO type. Patient data included: sex; age; hemoglobin; creatinine; diagnosis; interventions; ABO blood group and hospital discharge status. Factors associated with mismatched blood and in-hospital mortality were examined using generalized estimating equations to account for the potential serial dependence over multiple transfusions. The effect of exposure to ABO mismatched blood on in-hospital death was examined through Cox regression with time-dependent strata defined by: year of first admission; disease group; and the cumulative number of units transfused (≤ 7 days of storage; > 7 days but ≤ 28 days storage; and, >28 days of storage); and, controlling for available baseline and time-varying characteristics.

Results

18,843 patients (blood groups A, B and AB), with complete covariates contributed to the analysis. Factors associated with transfusion of mismatched blood included: younger patient age (p<0.0001); lower hemoglobin (p<0.0001); higher creatinine (p<0.0001); intervention during hospitalization (OR=4.6, p<0.0001); and, patient ABO group whereby blood types A and B were much less likely to receive a mismatched unit compared to type AB patients (p<0.0001). There was a statistically significant interaction between patient blood type and the effect of receiving mismatched blood (p=0.034) with type A patients incurring a 79% higher risk of death (RR=1.79, 95% CI: 1.20, 2.67; p=0.0047); other patient blood types did not suggest increased risk. Similar results were observed when suspected trauma patients (≥ 6 units within 24 hours) were excluded from the analysis (Table 1).

Table 1

Multivariate Cox regression models for overall survival. Time Origin: from day of 1st transfusion; Stratified by: cumulative number of fresh blood transfused (≤7 days), cumulative number of middle age units transfused ( >7 days but ≤ 28 days), cumulative number of older units transfused (>28 days), year of first admission and comorbidities; Adjusted for: sex, age at admission (years), any intervention (time-varying), creatinine (μmol/L, time-varying) and hemoglobin (g/dL, time-varying).

Full Cohort N=18,843Full Cohort excluding Trauma Patients (≥ 6 units in 24 hours) N=17,435
RR95% CIpRR95% CIp
ABO Mismatch vs no mismatch 1.22 0.92, 1.63 0.1732 1.14 0.84, 1.55 0.3999 
Mismatch by Patients ABO group       
A: mismatch vs no mismatch 1.79 1.20, 2.67 0.0047 1.68 1.08, 2.62 0.0216 
B: mismatch vs no mismatch 0.64 0.30, 1.34 0.2319 0.72 0.34, 1.50 0.3793 
AB: mismatch vs no mismatch 1.01 0.62, 1.65 0.9674 0.91 0.55, 1.51 0.7189 
Interaction test (2df)   0.0340   0.0806 
Full Cohort N=18,843Full Cohort excluding Trauma Patients (≥ 6 units in 24 hours) N=17,435
RR95% CIpRR95% CIp
ABO Mismatch vs no mismatch 1.22 0.92, 1.63 0.1732 1.14 0.84, 1.55 0.3999 
Mismatch by Patients ABO group       
A: mismatch vs no mismatch 1.79 1.20, 2.67 0.0047 1.68 1.08, 2.62 0.0216 
B: mismatch vs no mismatch 0.64 0.30, 1.34 0.2319 0.72 0.34, 1.50 0.3793 
AB: mismatch vs no mismatch 1.01 0.62, 1.65 0.9674 0.91 0.55, 1.51 0.7189 
Interaction test (2df)   0.0340   0.0806 
Conclusion

Controlling for known potential confounders through Cox regression yielded evidence of increased risk of in-hospital mortality among blood type A patients receiving group O red cells. This association remained after suspected trauma patients were excluded from the analyses. Further study of the association observed in this study is warranted.

Disclosures:

Cook:CIHR: Research Funding. Heddle:CIHR: Research Funding; Canadian Blood Services: Membership on an entity’s Board of Directors or advisory committees, Research Funding; Health Canada: Research Funding. Eikelboom:CIHR: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.

Sign in via your Institution