Development and validation of a practical machine learning model to predict sepsis after liver transplantation

BACKGROUND

Postoperative sepsis is one of the main causes of mortality after liver transplantation (LT). Our study aimed to develop and validate a predictive model for postoperative sepsis within 7 d in LT recipients using machine learning (ML) technology.

METHODS

Data of 786 patients received LT from January 2015 to January 2020 was retrospectively extracted from the big data platform of Third Affiliated Hospital of Sun Yat-sen University. Seven ML models were developed to predict postoperative sepsis. The area under the receiver-operating curve (AUC), sensitivity, specificity, accuracy, and f1-score were evaluated as the model performances. The model with the best performance was validated in an independent dataset involving 118 adult LT cases from February 2020 to April 2021. The postoperative sepsis-associated outcomes were also explored in the study.

RESULTS

After excluding 109 patients according to the exclusion criteria, 677 patients underwent LT were finally included in the analysis. Among them, 216 (31.9%) were diagnosed with sepsis after LT, which were related to more perioperative complications, increased postoperative hospital stay and mortality after LT (all p < .05). Our results revealed that a larger volume of red blood cell infusion, ascitic removal, blood loss and gastric drainage, less volume of crystalloid infusion and urine, longer anesthesia time, higher level of preoperative TBIL were the top 8 important variables contributing to the prediction of post-LT sepsis. The Random Forest Classifier (RF) model showed the best overall performance to predict sepsis after LT among the seven ML models developed in the study, with an AUC of 0.731, an accuracy of 71.6%, the sensitivity of 62.1%, and specificity of 76.1% in the internal validation set, and a comparable AUC of 0.755 in the external validation set.

CONCLUSIONS

Our study enrolled eight pre- and intra-operative variables to develop an RF-based predictive model of post-LT sepsis to assist clinical decision-making procedure.

Bacterial translocation in patients undergoing major gastrointestinal surgery and its role in postoperative sepsis

Bacteria of the human intestinal microflora have a dual role. They promote digestion and are part of a defense mechanism against pathogens. These bacteria could become potential pathogens under certain circumstances. The term “bacterial translocation” describes the passage of bacteria of the gastrointestinal tract through the intestinal mucosa barrier to mesenteric lymph nodes and other organs. In some cases, the passage of bacteria and endotoxins could result in blood stream infections and in multiple organ failure. Open elective abdominal surgery more frequently results in malfunction of the intestinal barrier and subsequent bacterial translocation and blood stream infections than laparoscopic surgery. Postoperative sepsis is a common finding in patients who have undergone non-elective abdominal surgeries, including trauma patients treated with laparotomy. Postoperative sepsis is an emerging issue, as it changes the treatment plan in surgical patients and prolongs hospital stay. The association between bacterial translocation and postoperative sepsis could provide novel treatment options.

Hospital variability of postoperative sepsis and sepsis-related mortality after elective coronary artery bypass grafting surgery

PURPOSE

Hospital variability of postoperative sepsis and sepsis-related mortality after elective CABG surgery was not known in Australia.

MATERIAL AND METHODS

Population-based analysis of all elective patients who underwent CABG surgery in public and private hospitals between 2007 and 2014 using linked data from the state-wide Admitted Patient Data Collection and the NSW Registry of Births, Deaths, and Marriages.

RESULTS

We identified 18,928 (9464 pairs) matched patients who had elective CABG surgery in public hospitals (n = 9) and private hospitals (n = 13) during the study period. When compared to public hospital patients, private hospital patients had a significantly lower rate of post-CABG sepsis (13.3 vs 20.4 per 1000 admissions, P < 0.001; treatment effects: -7.1, 95%CI: -11.1 to -3.3), a lower in-hospital mortality rate (6.1 vs 9.9 per 1000 admissions, P = 0.006; treatment effects: -3.8, 95%CI: -6.5 to -1.1), and a lower rate of 30-day readmission (11.9% vs 13.9%, P < 0.001; treatment effects: -2.0%, 95%CI: -3.1% to -1.0%). In addition, for both public and private hospital groups, there were significant differences for all outcomes when comparing the worst and best performance quintile hospitals.

CONCLUSIONS

Hospital variability of postoperative sepsis, in-hospital mortality and readmission after elective CABG existed between and within public and private hospitals.

Probiotics and Synbiotics Decrease Postoperative Sepsis in Elective Gastrointestinal Surgical Patients: a Meta-Analysis

BACKGROUND

The health benefits of probiotics and synbiotics are well established in healthy adults, but their role in preventing postoperative sepsis remains controversial. This meta-analysis assesses the impact of probiotics and synbiotics on the incidence of postoperative sepsis in gastrointestinal (GI) surgical patients.

METHODS

A comprehensive literature search of all published randomized control trials (RCTs) was conducted using PubMed, Cochrane Central Registry of Controlled Trials, and Google Scholar (1966-2015). Inclusion criteria included RCTs comparing the use of any strain or dose of a specified probiotic/synbiotic with placebo or a "no treatment" control group. The incidence of postoperative sepsis (within 1 month of surgery) and postoperative mortality were analyzed.

RESULTS

Fifteen RCTs involving 1201 patients (192 receiving probiotics, 413 receiving synbiotics, and 596 receiving placebo) were analyzed. Overall, probiotic and synbiotic uses significantly reduced the risk of developing postoperative sepsis by 38 % (relative risk (RR) = 0.62, 95 % confidence interval (CI) 0.52-0.74, p < 0.001).

CONCLUSIONS

The use of probiotic/synbiotic supplementation is associated with a significant reduction in the risk of developing postoperative sepsis in patients undergoing elective GI surgery. Probiotic/synbiotic supplementation is a valuable adjunct in the care of patients undergoing GI surgery. Additional studies are required to determine the optimal dose and strain of probiotic/synbiotic.

Postoperative sepsis in cancer patients undergoing major elective digestive surgery is associated with increased long-term mortality

BACKGROUND

Major postoperative events (acute respiratory failure, sepsis, and surgical complications) are frequent early after elective gastroesophageal and pancreatic surgery. It is unclear whether these complications impact equally on long-term outcome.

METHODS

Prospective observational study including the patients admitted to the surgical intensive care unit between January 2009 and October 2011 after elective gastroesophageal and pancreatic surgery. Risk factors for 30-day major postoperative events and long-term outcome were evaluated.

RESULTS

During the study period, 259 patients were consecutively included. Among them, 166 (64%), 54 (21%), and 39 (15%) patients underwent pancreatic surgery, gastric surgery, and esophageal surgery, respectively. Using the Clavien-Dindo classification, 117 patients (45%) developed at least 1 postoperative complication, including 60 (23%) patients with acute respiratory failure, 77 (30%) with sepsis, and 89 (34%) with surgical complications. The median follow-up from the time of intensive care unit admission was 34 months (95% confidence interval, 30-37 months). The 1-year survival was 95% (95% confidence interval, 92-98). Among the perioperative variables, postoperative sepsis and an American Society of Anesthesiologists score higher than 2 were independently associated with long-term mortality. In septic patients, death (n = 16) was significantly associated with cancer recurrence (n = 10; P < .0001). Independent factors associated with postoperative sepsis were a Sequential Organ Failure Assessment score on day 1, a systemic inflammatory response syndrome on day 3, positive intraoperative microbiological samples, Simplified Acute Physiology Score II and an American Society of Anesthesiologists score higher than 2 (P < .005).

CONCLUSIONS

Postoperative sepsis was the only major postoperative event associated with long-term mortality. Postoperative sepsis may reflect a deep impairment of immune response, which is potentially associated with cancer recurrence and mortality.