RESUMEN
Patients with advanced chronic liver disease (ACLD) or cirrhosis undergoing surgery have an increased risk of morbidity and mortality in contrast to the general population. This is a retrospective, observational study to evaluate the predictive capacity of surgical risk scores in European patients with ACLD. Cirrhosis was defined by the presence of thrombocytopenia with <150,000/uL and splenomegaly, and AST-to-Platelet Ratio Index >2, a nodular liver edge seen via ultrasound, transient elastography of >15 kPa, and/or signs of portal hypertension. We assessed variables related to 90-day mortality and the discrimination and calibration of current surgical scores (Child-Pugh, MELD-Na, MRS, NSQIP, and VOCAL-Penn). Only patients with ACLD and major surgeries included in VOCAL-Penn were considered (n = 512). The mortality rate at 90 days after surgery was 9.8%. Baseline disparities between the H. Mar and VOCAL-Penn cohorts were identified. Etiology, obesity, and platelet count were not associated with mortality. The VOCAL-Penn showed the best discrimination (C-statistic90D = 0.876) and overall predictive capacity (Brier90D = 0.054), but calibration was not excellent in our cohort. VOCAL-Penn was suboptimal in patients with diabetes (C-statistic30D = 0.770), without signs of portal hypertension (C-statistic30D = 0.555), or with abdominal wall (C-statistic30D = 0.608) or urgent (C-statistic180D = 0.692) surgeries. Our European cohort has shown a mortality rate after surgery similar to those described in American studies. However, some variables included in the VOCAL-Penn score were not associated with mortality, and VOCAL-Penn's discriminative ability decreases in patients with diabetes, without signs of portal hypertension, and with abdominal wall or urgent surgeries. These results should be validated in larger multicenter and prospective studies.