Predicting poor outcome following hepatectomy: analysis of 2313 hepatectomies in the NSQIP database

Background: For the past two decades multiple series have documented that liver resection has become safer. The purpose of this study was to determine the current status of hepatic resection in the USA by analysing the multi-institutional experience within the National Surgical Quality Improvement Program (NSQIP) dataset. Methods: Of the 363 897 cases in the 2005-2007 NSQIP Participant Use File, 2313 elective open hepatectomy cases were identified (1344 partial, 230 left, 510 right and 229 extended hepatectomies). A total of 57 perioperative risk factors and 28 postoperative complications were compared. To determine the applicability of NSQIP general risk models to hepatic surgery, the prognostic value of standard multivariate analysis was compared with the NSQIP general surgery aggregate risk indices (expected probability of morbidity [morbprob], expected probability of mortality [mortprob]). Results: The median age of patients listed in the database was 60 years; sex distributions were equivalent; 78% were White; 65% of patients had an ASA score of 3 or 4, and the most prevalent co-morbidity was hypertension (46%). A total of 41% of patients had disseminated cancer, 19% of whom had received chemotherapy within 30 days of surgery. The overall 30-day mortality rate was 2.5% (57/2313) and the 30-day major morbidity rate was 19.6% (453/2313). Multivariate analysis identified nine risk factors associated with major morbidity and two risk factors associated with mortality. In contrast, the morbprob and mortprob statistics did not predict outcomes accurately. For those patients who developed major morbidity, the median length of stay was longer (10 vs. 6 days; P = 0.001) and the mortality rate was higher (11.3% vs. 0.3%; P = 0.001). Conclusions: Analysis of the NSQIP experience with hepatectomy indicates that the current mortality and major morbidity rate benchmarks are 2.5% and 19.6%, respectively. Poor outcomes were associated with nutritional status, liver function and the extent of hepatectomy. The NSQIP general surgery morbprob and mortprob values were relatively poor predictors of post-hepatectomy observed morbidity, indicating the need for specialty-specific NSQIP modelling.