Although delivering less oxygen is associated with acute kidney injury, little is known about its association with long-term mortality and how the association may be modified by transfusion.

Minute-by-minute pump flow and hemoglobin were used to estimate oxygen delivery indexed to body surface area (DO₂I) during cardiopulmonary bypass. In this retrospective study, the associations between mean DO₂I and cumulative DO₂I deficit <300 mL/min/m² threshold and long-term mortality were estimated using Cox proportional and nonproportional hazard models.

At a mean follow-up of 5.00 (SD, 1.86) years (range, 3.10-8.12 year), 698 of 4203 patients (17%) had died. Patients who died had lower mean DO₂I (230 [SD, 40] mL/min/m² vs 251 [SD, 47] mL/min/m²) and greater cumulative DO₂I deficit (9.94 [SD, 8.64] mL/min/m² vs 7.01 [SD, 7.07] L/min/m²). After adjusting for demographics, comorbidities, laboratory values, type and status of the operations, blood pressure, and red blood cell transfusion during bypass, we found that mean DO₂I was associated with mortality (hazard ratio [HR], 0.997; 95% CI, 0.995-0.999; P = .001). When cumulative DO₂I deficit was added to the model, it replaced mean DO₂I (HR, 1.031; 95% CI, 1.018-1.045; P < .001) of dying for each liter of O₂/m² deficit, with a small time-varying component (HR, 0.994; 95% CI, 0.990-0.999) per year of follow-up (P = .010). After adjustment, red blood cell transfusion did not remain in the models.

There is an inverse relationship between decreased DO₂I (measured as mean or cumulative deficit) and long-term mortality. Red blood cell transfusion did not significantly modify the relationship between DO₂I and mortality.