Hospital variation in transfusion and infection after cardiac surgery: a cohort study

Transfusion practices in hospitalised patients are being re-evaluated, in part due to studies indicating adverse effects in patients receiving large quantities of stored blood. Concomitant with this re-examination have been reports showing variability in the use of specific blood components. This investigation was designed to assess hospital variation in blood use and outcomes in cardiac surgery patients.

We evaluated outcomes in 24,789 Medicare beneficiaries in the state of Michigan, USA who received coronary artery bypass graft surgery from 2003 to 2006. Using a cohort design, patients were followed from hospital admission to assess transfusions, in-hospital infection and mortality, as well as hospital readmission and mortality 30 days after discharge. Multilevel mixed-effects logistic regression was used to calculate the intrahospital correlation coefficient (for 40 hospitals) and compare outcomes by transfusion status.

Overall, 30% (95 CI, 20% to 42%) of the variance in transfusion practices was attributable to hospital site. Allogeneic blood use by hospital ranged from 72.5% to 100% in women and 49.7% to 100% in men. Allogeneic, but not autologous, blood transfusion increased the odds of in-hospital infection 2.0-fold (95% CI 1.6 to 2.5), in-hospital mortality 4.7-fold (95% CI 2.4 to 9.2), 30-day readmission 1.4-fold (95% CI 1.2 to 1.6), and 30-day mortality 2.9-fold (95% CI 1.4 to 6.0) in elective surgeries. Allogeneic transfusion was associated with infections of the genitourinary system, respiratory tract, bloodstream, digestive tract and skin, as well as infection with Clostridium difficile. For each 1% increase in hospital transfusion rates, there was a 0.13% increase in predicted infection rates.

Allogeneic blood transfusion was associated with an increased risk of infection at multiple sites, suggesting a system-wide immune response. Hospital variation in transfusion practices after coronary artery bypass grafting was considerable, indicating that quality efforts may be able to influence practice and improve outcomes.