A model for clinical estimation of perioperative hemorrhage

The purpose of this study was to assess the accuracy of estimated blood loss (EBL) as a reliable predictor of actual blood loss during orthopedic procedures. Between 1999 and 2002, 198 orthopedic cases were reviewed. A retrospective review compiled preoperative and postoperative demographic and laboratory data from the surgical patients. Estimated blood loss data was collected from the perioperative and anesthesia reports. Statistical analysis of EBL vs. change in hemoglobin yielded a correlation coefficient of 0.189 and a p value of 0.008. We used multiple linear regression to obtain a model to predict change in hemoglobin based on EBL and the intravenous fluids received. The model is as follows: predicted change in hemoglobin = 1.001 x estimated blood loss (in liters) + 0.441 x intravenous fluids received (in liters) + 2.334. The study population included 198 patients, 126 males and 72 females, who met our inclusion criteria. The mean age was 68.1 years (range: SD 12.5), including 126 males (64%) and 72 females (37%). The mean amount of perioperative intravenous fluids given was 1,732 mL (SD: 773). The mean surgical time was 64.8 minutes (SD: 23.1). The mean preoperative hematocrit and hemoglobin levels were 40.9 g/dL (SD: 4.3) and 13.9 g/dL (SD: 1.6), respectively. The mean postoperative hematocrit and hemoglobin levels were 32.0 g/dL (SD: 6.0) and 10.7 g/dL (SD: 1.6), respectively. The mean difference of preoperative hemoglobin vs. postoperative hemoglobin was 3.3 g/dL (SD 2.1). In this retrospective study, clinical estimation of blood loss was closely correlated with actual change in perioperative hemoglobin. Accurately predicting the postoperative hemoglobin level may prevent many unnecessary blood transfusions and related complications.