Evaluating variation in perioperative red blood cell transfusion for patients undergoing elective gastrointestinal cancer surgery.

BACKGROUND

Patients undergoing gastrointestinal cancer surgery often receive packed red blood cell transfusions. Understanding practice variation is critical to support efforts working toward responsible transfusion use. We measured the extent and importance of variation in perioperative packed red blood cell transfusion use across physicians and hospitals among gastrointestinal cancer surgery patients.

METHODS

We identified patients who underwent elective gastrointestinal cancer resection between 2007 and 2019 using linked administrative health data sets in Ontario, Canada. We used funnel plots to describe variation in transfusion use, adjusted for patient case mix. Hierarchical regression models quantified patient-level, between-physician, and between-hospital variation in transfusion use with R measures, variance partition coefficients, and median odds ratios.

RESULTS

Of 59,964 included patients (median age 69 years; 43.2% female; 75.8% colorectal resections), 18.0% received perioperative packed red blood cell transfusions. Funnel plots showed variation in transfusion use among physicians and hospitals. Patient characteristics, such as age, comorbidity, and procedure type, combined to explain 12.8% of the variation. After adjusting for case mix, systematic between-physician and between-hospital differences were responsible for 2.8% and 2.1% of the variation, respectively. This translated to an approximately 30% difference in the odds of transfusion for 2 similar patients treated by distinct physicians (median odds ratio: 1.35, 95% confidence interval 1.30-1.40) and hospitals (median odds ratio: 1.30, 95% confidence interval 1.23-1.42). We observed comparable effects across procedure-type subgroups.

CONCLUSION

Transfusion provision is highly driven by patient factors. Yet the impact of the treating physician and hospital on variation relative to other factors is important and reflects opportunities to target modifiable processes of care to standardize perioperative packed red blood cell transfusion practice.