Machine learning predicts unpredicted deaths with high accuracy following hepatopancreatic surgery.

BACKGROUND: Machine learning to predict morbidity and mortality-especially in a population traditionally considered low risk-has not been previously examined. We sought to characterize the incidence of death among patients with a low estimated morbidity and mortality risk based on the National Surgical Quality Improvement Program (NSQIP) estimated probability (EP), as well as develop a machine learning model to identify individuals at risk for "unpredicted death" (UD) among patients undergoing hepatopancreatic (HP) procedures. METHODS: The NSQIP database was used to identify patients who underwent elective HP surgery between 2012-2017. The risk of morbidity and mortality was stratified into three tiers (low, intermediate, or high estimated) using a k-means clustering method with bin sorting. A machine learning classification tree and multivariable regression analyses were used to predict 30-day mortality with a 10-fold cross validation. C statistics were used to compare model performance. RESULTS: Among 63,507 patients who underwent an HP procedure, median patient age was 63 (IQR: 54-71) years. Patients underwent either pancreatectomy (n=38,209, 60.2%) or hepatic resection (n=25,298, 39.8%). Patients were stratified into three tiers of predicted morbidity and mortality risk based on the NSQIP EP: low (n=36,923, 58.1%), intermediate (n=23,609, 37.2%) and high risk (n=2,975, 4.7%). Among 36,923 patients with low estimated risk of morbidity and mortality, 237 patients (0.6%) experienced a UD. According to the classification tree analysis, age was the most important factor to predict UD (importance 16.9) followed by preoperative albumin level (importance: 10.8), disseminated cancer (importance: 6.5), preoperative platelet count (importance: 6.5), and sex (importance 5.9). Among patients deemed to be low risk, the c-statistic for the machine learning derived prediction model was 0.807 compared with an AUC of only 0.662 for the NSQIP EP. CONCLUSIONS: A prognostic model derived using machine learning methodology performed better than the NSQIP EP in predicting 30-day UD among low risk patients undergoing HP surgery.

Impact of Preoperative Cholangitis on Short-term Outcomes Among Patients Undergoing Liver Resection.

BACKGROUND: The impact of preoperative cholangitis (PC) on perioperative outcomes among patients undergoing liver resection remains poorly defined. We sought to characterize the prevalence of PC among patients undergoing hepatectomy and define the impact of PC on postoperative outcomes. METHODS: Patients who underwent liver resection between 2013 and 2015 were identified using the Center for Medicare Services (CMS) 100% Limited Data Set (LDS) Standard Analytic Files (SAFs). Short-term outcomes after liver resection, stratified by the presence of PC, were examined. Subgroup analyses were performed to evaluate the relationship between the timing of liver resection relative to PC. RESULTS: Among 7392 patients undergoing liver resection, 251 patients (3.4%) experienced PC. Patients with PC were more likely to be male (59.0% vs. 50.6%) and to have a benign diagnosis (34.3% vs. 19.8%) compared with patients without PC (both p<0.05). On multivariable analysis, PC was associated with increased odds of experiencing a complication (OR 1.54, 95%CI 1.17-2.03), extended LOS (OR 2.60, 95%CI 1.99-3.39), 90-day mortality (OR 2.31, 95%CI 1.64-3.26), and higher Medicare expenditures (OR 3.32, 95%CI 2.55-4.32). Among patients with PC, requirement of both endoscopic and percutaneous biliary drainage (OR 5.16, 95%CI 1.36-9.61), as well as liver resection < 2 weeks after PC (OR 2.92, 95%CI 1.13-7.57) were associated with higher odds of 90-day mortality. CONCLUSION: Approximately 1 in 30 Medicare beneficiaries undergoing liver resection had a history of PC. PC was associated with an increased risk of adverse short-term outcomes and higher healthcare expenditures among patients undergoing hepatectomy.

Development and validation of a real-time mortality risk calculator before, during and after hepatectomy: an analysis of the ACS NSQIP database.

BACKGROUND: Although most conventional risk prediction models have been based on preoperative information, intra- and post-operative events may be more relevant to mortality after surgery. We sought to develop a mortality risk calculator based on real time characteristics associated with hepatectomy. METHODS: Patients who underwent hepatectomy between 2014 and 2017 were identified in the ACS-NSQIP dataset. Three prediction models (pre-, intra-, post-operative) were developed and validated using perioperative data. RESULTS: Among 14,720 patients, 197 (1.3%) experienced 30-day mortality. The predictive ability of the real-time mortality risk calculator was very good based on only preoperative factors (AUC; training cohort: 0.813, validation cohort: 0.731). Incorporating intra-operative variables into the model increased the AUC (training: 0.838, validation: 0.777), while the post-operative model achieved an AUC of 0.922 in the training and 0.885 in the validation cohorts, respectively. While patients with low preoperative risk had only very small fluctuations in the estimated 30-day mortality risk during the intraoperative (Δ0.4%) and postoperative (Δ0.6%) phases, patients who were already deemed high risk preoperatively had additional increased mortality risk based on factors that occurred in the intraoperative (Δ5.4%) and postoperative (Δ9.3%) periods. CONCLUSION: A real-time mortality risk calculator may better help clinicians identify patients at risk of death at the different stages of the surgical episode.

Potential disease burden of patients with substance abuse undergoing major abdominal surgery: A propensity score-matched analysis.

BACKGROUND: Over 19 million Americans have a substance abuse disorder. The current study sought to characterize the relationship between substance abuse with in-hospital outcomes following major, elective abdominal surgery. METHODS: The Nationwide Inpatient Sample was used to identify patients who underwent major abdominal surgery between 2007 to 2014. Patients with preoperative substance abuse, including alcohol, opioids, and non-opioid drugs, were identified. Propensity score matching was used to examine the association of substance abuse with perioperative outcomes. RESULTS: Among 301,659 patients, 7,925 patients (2.6%) had a history of substance abuse. Pancreatectomy was the surgical procedure with the highest proportion of patients with substance abuse history (n = 844, 4.7%). Compared with patients without a substance abuse history, patients with a substance abuse history were more likely to be younger (median age, 60 years [interquartile range (IQR) 52-69] vs 63 years [IQR 52-72]), male (n = 5,438, 67.5% vs n = 132,961, 54.7%), and be in the lowest income category (n = 2,062, 26% vs n = 64,345, 21.9%) (all P < .001). On propensity score matching, substance abuse was associated with increased odds ratio of experiencing a complication (odds ratio [OR] 1.68, 95% confidence interval [CI] 1.55-1.82), non-home discharge (OR 1.95, 95% CI 1.76-2.16), extended length of stay (OR 1.88, 95% CI 1.76-2.02), and higher expenditure (OR 1.62, 95% CI 1.49-1.77). Stratified by the type of substance abuse, patients with history of alcohol (OR 1.57, 95% CI 1.44-1.71) and drug abuse (OR 1.26, 95% CI 1.14-1.39) were more likely to experience a complication, whereas only history of alcohol abuse was associated with higher odds ratio of in-hospital mortality (OR 1.38, 95% CI 1.07-1.79) (all P < .05). CONCLUSION: Up to 1 in 50 patients undergoing complex abdominal surgery had a substance abuse history. History of substance abuse was associated with an increased risk of adverse perioperative outcomes and higher healthcare expenditures.

Low hepatitis C antibody screening rates among an insured population of Tennessean Baby Boomers.

INTRODUCTION: Chronic Hepatitis C Virus (HCV) infection is common and can cause liver disease and death. Persons born from 1945 through 1965 ("Baby Boomers") have relatively high prevalence of chronic HCV infection, prompting recommendations that all Baby Boomers be screened for HCV. If chronic HCV is confirmed, evaluation for antiviral treatment should be performed. Direct-acting antivirals can cure more than 90% of people with chronic HCV. This sequence of services can be referred to as the HCV "cascade of cure" (CoC). The Tennessee (TN) Department of Health (TDH) and a health insurer with presence in TN aimed to determine the proportion of Baby Boomers who access HCV screening services and appropriately navigate the HCV CoC in TN. METHODS: TDH surveillance data and insurance claim records were queried to identify the cohort of Baby Boomers eligible for HCV testing. Billing codes and pharmacy records from 2013 through 2015 were used to determine whether HCV screening and other HCV-related services were provided. The proportion of individuals accessing HCV screening and other steps along the HCV CoC was determined. Multivariable analyses were performed to identify factors associated with HCV screening and treatment. RESULTS: Among 501,388 insured Tennessean Baby Boomers, 7% were screened for HCV. Of the 40,019 who received any HCV-related service, 86% were screened with an HCV antibody test, 20% had a confirmatory HCV PCR, 9% were evaluated for treatment, and 4% were prescribed antivirals. Hispanics were more likely to be screened and treated for HCV than non-Hispanic whites. HCV screening was more likely to occur in the Nashville-Davidson region than in other regions of TN, but there were regional variations in HCV treatment. CONCLUSIONS: Many insured Tennessean Baby Boomers do not access HCV screening services, despite national recommendations. Demographic and regional differences in uptake along the HCV CoC should inform public health interventions aimed at mitigating the effects of chronic HCV.