Published online Aug 26, 2019. doi: 10.12998/wjcc.v7.i16.2176
Peer-review started: April 24, 2019
First decision: June 4, 2019
Revised: June 22, 2019
Accepted: July 3, 2019
Article in press: July 3, 2019
Published online: August 26, 2019
Surgical site infections (SSI) reportedly account for > 50% of infectious complications after hepatectomy for hepatocellular carcinoma (HCC). It has a significant impact on morbidity, mortality, prolonged hospitalization, costs, and even long-term oncology outcomes. Hence, SSI prevention has been considered a top priority for improving perioperative outcomes. Previous studies suggest that many factors can influence SSIs in patients undergoing hepatectomy. However, some of these factors remain controversial.
Models to identify the patients with an increased risk of developing SSI are limited. National Nosocomial Infection surveillance (NNIS) risk index was developed using data from a wide range of patients undergoing various surgical procedures with different disease conditions. Hence, the applicability of NNIS is limited in patients undergoing hepatectomy for HCC. To develop an effective forecasting model to screen out patients at high risk of SSI is vital for improving individual clinical decision making and the perioperative morbidity rate.
In this study, we aimed to investigate the risk factors for SSI after hepatectomy for HCC, and develop a prediction nomogram for SSI by analyzing clinical data from a consecutive series of patients undergoing hepatectomy at our institution and validate the prediction model in an external cohort.
The data of 640 patients with HCC who underwent attempted curative liver resection were retrospectively collected from two academic institutions in China. The records of all patients were reviewed. We identified the independent predictors of SSI using multivariate logistic regression analysis. Then, a nomogram was formulated based on the identified factors, using the rms package in R, version 3.2.1 (http://www.r-project.org/). The performance of prediction model was assessed using an external cohort from the second hospital.
The logistic regression identified three pre-operative variables (serum albumin level, repeat hepatectomy, and ASA score) and one intra-operative variable (duration of operation) as independent predictors of overall SSI. We developed a nomogram to predict SSI in patients after hepatectomy for HCC by integrating the four factors identified. Our nomogram showed better prediction accuracy compared to the NNIS risk index. Finally, we stratified the patients of the entire cohort into three groups with a distinct risk of SSI, based on the predicted risk distribution using the nomogram.
Our nomogram appears to indicate a higher accuracy for predicting SSI, as compared to the NNIS risk index. Our prediction model integrated the information of hepatic surgery history and liver function, which were significantly associated with SSI in our population. The NNIS risk index was developed using a wide range of patients, whereas our prediction model was established only using patients who underwent hepatectomy for HCC. The increased relevance could explain the better performance of our prediction model in this population.
This nomogram based on identified factors is able to stratify patients into three groups with distinct risks of SSI, and performs well on external validation. We primarily focused on the preoperative and intra-operative predictors because we aimed to develop a prediction model to identify suitable patients for enhanced recovery after surgery at a relatively early time point. In the future, we will assess the performance of this model among a diverse population of patients.