Evaluating the appropriateness of red blood cell transfusions: The limitations of retrospective medical record reviews INTERNATIONAL JOURNAL FOR QUALITY IN HEALTH CARE Audet, A. M., GOODNOUGH, L. T., Parvin, C. A. 1996; 8 (1): 41-49


Several studies have looked at the appropriateness of red blood cell transfusions, using retrospective chart reviews to assess compliance with guidelines. The goal of this study was to determine the quality of medical chart documentation, and assess the validity and the feasibility of using retrospective chart review data as part of a quality improvement (QI) program, to evaluate the appropriateness of peri-operative transfusions.The charts of 188 patients admitted for elective orthopedic surgery were reviewed. Both intra-operative and post-operative transfusion events (n = 353) were analyzed.Only 68% of post-operative transfusion events on the day of surgery and 35% of transfusion events on days after surgery were accompanied by documentation of blood loss and/or change in vital signs. Symptoms were recorded in only 10% of post-operative transfusion events. The rationale for transfusion was recorded in only 16% of post-operative transfusion events on the day of surgery, in 27% on post-operative days and in 95% of intra-operative transfusions. The documentation of rationale was not different for transfusion events involving autologous blood (31%) or allogeneic blood (32%). This study provided information on transfusion practices. Single unit transfusions occurred in only 47 and 34% of patients receiving autologous and allogeneic blood, respectively. Only 19% of patients transfused with more than one allogeneic blood units, were clinically reassessed between blood units, compared to 34% of patients receiving more than one autologous blood unit. We conclude that retrospective chart reviews are limited by inadequate documentation and may not be the optimal source of information to determine the appropriateness of a transfusion. Improved methods (e.g. prospective reviews, or other system-level data collection methods) are needed for QI programs to influence practice. Despite its limitations, the information obtained suggests that current practice could be improved.

View details for Web of Science ID A1996UA20800006

View details for PubMedID 8680816