Risk factors for blood transfusion after revision shoulder arthroplasty

Abstract

BackgroundHigher blood transfusion rates have been reported in revision shoulder arthroplasty than in primary arthroplasty. This study seeks to evaluate factors associated with blood transfusions after revision shoulder arthroplasty and to evaluate the effectiveness of tranexamic acid (TXA) to reduce blood transfusion rates in revision shoulder arthroplasty. MethodsAll shoulder arthroplasties performed at one institution were retrospectively reviewed using a research database over two time periods: before (November 2011 to November 2012) and after (April 2015 to April 2017) a protocol for the use of TXA was implemented. The inclusion criterion was any revision shoulder arthroplasty performed during the study time periods. Demographic, preoperative, intraoperative, and postoperative factors were reviewed and compared based on transfusion. Patients who received TXA were compared with those patients who did not. Two-tailed Fisher exact tests, unpaired Welch’s t-tests, and multivariate logistic regression analysis were used to compare groups. ResultsA total of 135 revision shoulder arthroplasties were performed and evaluated during the study period. The overall transfusion rate for revision shoulder arthroplasty was 14.1%. Preoperative hematocrit (Hct) <42 was associated with a higher transfusion rate (18% vs. 5%; P = .03) on univariate analysis. Patients who received a transfusion had a significantly longer length of hospital stay (4.3 vs. 2.0 nights; P < .001). After controlling for covariates in the multivariate analysis, gender, body mass index, heart disease, diabetes mellitus, a history of a bleeding disorder, American Society of Anesthesiologists score, surgery type, and humeral stem retention were not individually associated with the need for a blood transfusion. A higher estimated blood loss was associated with a higher risk of blood transfusion (odds ratio, 2.04 per 100 mL increase; P < .001). On multivariate analysis, a one-unit decrease in preoperative Hct increased the odds of a blood transfusion by 1.20 (P = .03). On multivariate analysis, the use of TXA was associated with a decreased transfusion rate, but this only approached significance (odds ratio, 0.26; confidence interval, 0.05-1.24; P = .09). ConclusionBlood transfusion after revision shoulder arthroplasty is more common in patients with lower preoperative Hct and a higher estimated blood loss on multivariate analysis with an overall transfusion rate of 14%. The use of perioperative TXA was associated with a decreased risk of transfusion, but this only approached significance.