Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4073225 | Journal of Shoulder and Elbow Surgery | 2016 | 8 Pages |
BackgroundMalnutrition is an established risk factor for postoperative complications. The purpose of this investigation was to determine the overall prevalence of malnutrition in total shoulder arthroplasty (TSA) patients, the differences in prevalence across obesity subgroups, and the overall complication risk of malnourished patients compared with normal patients.MethodsThe American College of Surgeons National Surgical Quality Improvement Program database was queried for TSA cases from 2005 to 2013 for this retrospective cohort study. Malnutrition was defined as preoperative albumin concentration of <3.5 g/dL. Rates of postoperative complications were compared between normal and malnourished patients.ResultsWe identified 4,655 TSA cases, with preoperative albumin measurements available for 1681 patients (36.1%). Propensity score adjustment successfully reduced selection bias, with adjusted P values of >.05 for demographics, body mass index, and modified Charlson Comorbidity Index. Of the cohort with albumin measurements, 7.6% of patients were malnourished according to our criteria. Bivariate analysis showed malnourished patients had higher rates of pulmonary complications, anemia requiring transfusion, extended length of stay (LOS), and death (all P < .05). Propensity-adjusted multivariable logistic regression demonstrated that malnutrition was significantly associated (all P < .05) with postoperative transfusion (odds ratio, 2.49), extended LOS (odds ratio, 1.69), and death (odds ratio, 18.09).ConclusionThe overall prevalence of malnutrition was 7.6%. Malnourished patients were at a significantly increased risk for blood transfusion, longer hospital LOS, and death within 30 days of surgery. Multivariable analysis showed TSA patients with preoperative albumin levels of <3.5 g/dL are at much higher risk for morbidity and death after surgery than patients with albumin levels within normal reference ranges.