Inflammation After Surgical Trauma: Bleeding or Clotting?

Abstract

Bellomy et al 1 present an article in this issue of Anesthesia & Analgesia that examines “attributable mortality” in patients following major noncardiac surgery, seeking to identify deaths due to bleeding versus deaths due to venous thromboembolism (VTE).1 The size of the study is impressive: more than 6.5 million surgical procedures over 11 years in the National Surgical Quality Improvement Project (NSQIP) registry. Seventy confounding variables from NSQIP, including procedure type sorted into 244 categories, were included in a per-year multivariable model intended to define the contribution of “need for transfusion” or “venous thromboembolism” to the outcome of interest: death within 30 days of the surgical procedure. The result of this well-concocted statistical soup was calculation of attributable mortality for each of the 2 conditions, expressed as excess deaths per 100,000 cases. Around 5% of patients received at least 1 transfusion; around 1% had a diagnosed VTE. Transfusion was associated with mortality in every year of study but VTE only in the latter years. Transfusion accounted for 6–52 deaths per 100,000 cases over the years of study, while attributable mortality due to VTE was much less common, ranging from 2 to 5 per 100,000. These results held up in sensitivity analysis examining intraoperative versus postoperative complications and in a subset of patients undergoing vascular surgical procedures. As a high-level overview, this study is clear enough, but the results become fuzzier when examined more closely. Bleeding was defined as the need for 1 or more units of red blood cell (RBC) transfusion within 72 hours of surgery, which is a discrete field in NSQIP. (This field was redefined from 4 or more RBC units in 2010; this affected the number of cases counted but not the attributable mortality.) The authors did not have the ability to separate single unit versus massive transfusion or to know if the transfusion was indicated based on modern evidence. Similarly, VTE was defined as the presence of a specific diagnostic code for deep venous thrombosis or pulmonary embolus, with no knowledge of whether it occurred as the result of symptomatic disease or random screening, or whether it was directly fatal or clinically irrelevant. It is also worth noting that recommendations for VTE screening changed in the middle of this study period, with publication in 2012 of the widely read College of Chest Physicians guidance.2 Guidelines for both VTE screening and postoperative transfusion have evolved substantially throughout the past 2 decades, which may have affected both outcome variables.3 Like many longitudinal registry-based studies, this one should be regarded as a broad glimpse of reality: surgical patients die more often in association with a transfusion than in association with a VTE. Attempting to read too much into pathophysiological cause and effect exceeds the specificity of the NSQIP data set. The authors devoted some discussion to the question of whether these results should temper the use of thromboembolic prophylaxis in the perioperative period. Wisely, they caution against this conclusion. The study was not intended, nor the data able, to answer the question. It seems certain that VTE prophylaxis saves more lives than it takes due to bleeding Inflammation After Surgical Trauma: Bleeding or Clotting?