Back to Annual Meeting Program

Postoperative mortality rates after surgery in patients with liver cirrhosis are high. Risk factors for mortality may help planning therapy in those high risk patients. We, therefore, evaluated/updated potential risk factors (including Child- and MELD-scores) for perioperative mortality after more than 200 operations in patients with cirrhosis performed during the last decade.
Methods: Since 2001 231 various general surgical procedures (80% intraabdominal, 20% abdominal wall) were performed in patients with liver cirrhosis (38% emergent). Cirrhosis was classified according to Child (41% A; 38% B, 21% C) and MELD-score (median 11). Procedures were subclassified as major (laparotomy with resection) or minor (abdominal wall, ‘minor’ laparotomy, laparoscopy). Univariate and multivariate (binary logistic regression) analysis was undertaken to identify risk factors for mortality. Multivariate analysis was performed in different models to exclude collinearity due to overlapping parameters (Child, MELD, laboratory values).
Results: Overall postoperative mortality was 17%. In univariate analysis the CHILD classification (mortality: 6% Child A; 11% Child B, 45% Child C; p<0.001), higher/increasing MELD score (p<0.001), higher/increasing ASA score (p<0.001), emergency procedures (35% vs 5% elective; p<0.001), major procedures (p<0.02), need for transfusions (36% vs 4% in patients without transfusions; p<0.001) and various preoperative laboratory values (anemia, thrombocytopenia, hyponatremia; all p<0.05) were associated with increased mortality. In multivariate risk factor analyses blood transfusions (p<0.001; RR 7), ASA score (p<0.01), Child class (p<0.02) and a thrombocytopenia (p<0.02) were independent predictors for mortality. The MELD score, emergent procedures and extent of surgery showed a trend but did not significantly predict mortality in the multivariate model.
Conclusions: Patients requiring blood transfusions have a very high risk for mortality. Preoperative liver function and co-morbidity also predict early mortality after surgery. In our series the CHILD score was a better predictor for postoperative mortality than the MELD-score.