Victor A. Ferraris, MD, PhD, Marion Hochstetler, MD, Jeremiah T. Martin, MBBCh, FRCSI, Angela Mahan, MD, and Sibu P. Saha, MD, MBA
Surgery. 2015 Sep;158(3):608-17. doi: 10.1016/j.surg.2015.02.027. Epub 2015 May 29.
Ferraris et al. set out to study the dichotomy of evidence that blood transfusion has been associated with adverse outcomes and can also be a life-saving measure in specific clinical situations. The investigators sought to determine if certain patient characteristics were associated with transfusion portending to a good or bad outcome. Using the NSQIP database from 2010 to 2012, Ferraris et al. assessed 470,407 patients, of which 32,953 (7%) received a blood transfusion within 72 hours of an operation. Using propensity matching, these patients were stratified into 10 groups depending on the predicted risk of mortality and predicted risk of serious morbidity. Patients with the highest predicted risk of mortality or serious complications had no incremental increased risk associated with blood transfusion, while the patients with low predicted risk of morbidity or mortality had a 10 to 13-fold increased mortality and a 8 to 12-fold increase in complications associated with transfusion. The greatest limitation of this study is that there was no quantification of the amount of blood transfused, therefore, the investigators could not comment on how increasing amounts of transfusion affected outcomes. This is an inherent limitation to the use of the NSQIP database. These findings are also reminiscent of the TRICC trial where the “younger” and “healthier” critically ill ICU patients had increased mortality in the liberal transfusion group.
Daniel J. Johnson, BS; Steven M Frank, MD
Department of Anesthesiology and Critical Care Medicine
Johns Hopkins Medical Institutions