The impact of serum albumin and serum protein levels on POSSUM score of patients with proximal femur fractures

Background: POSSUM was developed to predict risk-adjusted mortality and morbidity rates for surgical procedures. We evaluated the impact of serum albumin and serum protein levels on POSSUM scores. Methods: Medical files of 2269 patients operated for proximal femur fractures were reviewed. Preoperative serum albumin levels were available for 387 patients (mean 35.1 g/l, range 22-49) and serum protein levels for 279 patients (mean 61.6 g/l, range 40-86). Results: Serum albumin and protein levels were inversely associated with mortality in multivariate models (albumin, OR = 0.89, p = 0.009; protein, OR = 0.92, p = 0.009) and in composite outcome models as well (albumin, OR = 0.955, p = 0.219, protein, OR = 0.94, p = 0.014). The area under the curve (AUC) for POSSUM prediction of mortality (n = 1770) was 0.632 (95% CI: 0.580-0.684, p < 0.001). The AUC for a model including serum protein levels was 0.742 (95% CI: 0.649-0.834, p < 0.001). Hospitalisation time was longer for patients with lower serum proteins levels (p = 0.045), with an inverse correlation (Pearson correlation ≥0.164, p = 0.011). Conclusions: Lower preoperative serum albumin and serum protein levels were associated with increased risk for mortality, increased hospitalisation time and poorer outcomes in patients operated for proximal femoral fractures. Including those values to POSSUM scores would increase their predictive power.