Machine learning to predict pediatric choledocholithiasis: A Western Pediatric Surgery Research Consortium retrospective study.

BACKGROUND: The purpose of this study was to accurately predict pediatric choledocholithiasis with clinical data using a computational machine learning algorithm. METHODS: A multicenter retrospective cohort study was performed on children <18 years of age who underwent cholecystectomy between 2016 to 2019 at 10 pediatric institutions. Demographic data, clinical findings, laboratory, and ultrasound results were evaluated by bivariate analyses. An Extra-Trees machine learning algorithm using k-fold cross-validation was used to determine predictive factors for choledocholithiasis. Model performance was assessed using the area under the receiver operating characteristic curve on a validation dataset. RESULTS: A cohort of 1,597 patients was included, with an average age of 13.9 ± 3.2 years. Choledocholithiasis was confirmed in 301 patients (18.8%). Obesity was the most common comorbidity in all patients. Choledocholithiasis was associated with the finding of a common bile duct stone on ultrasound, increased common bile duct diameter, and higher serum concentrations of aspartate aminotransferase, alanine transaminase, lipase, and direct and peak total bilirubin. Nine features (age, body mass index, common bile duct stone on ultrasound, common bile duct diameter, aspartate aminotransferase, alanine transaminase, lipase, direct bilirubin, and peak total bilirubin) were clinically important and included in the machine learning algorithm. Our 9-feature model deployed on new patients was found to be highly predictive for choledocholithiasis, with an area under the receiver operating characteristic score of 0.935. CONCLUSION: This multicenter study uses machine learning for pediatric choledocholithiasis. Nine clinical factors were highly predictive of choledocholithiasis, and a machine learning model trained using medical and laboratory data was able to identify children at the highest risk for choledocholithiasis.

Pediatric DUCT Score: A Highly Specific Predictive Model for Choledocholithiasis in Children.

BACKGROUND: Current adult guidelines for the management of choledocholithiasis (CDL) may not be appropriate for children. We hypothesized adult preoperative predictive factors are not reliable for predicting CDL in children. STUDY DESIGN: A multicenter retrospective cohort study was performed evaluating children (≤18 years of age) who underwent cholecystectomy for gallstone disease at 10 children's hospitals. Univariate and multivariable analyses were used to identify factors independently associated with CDL. Patients were stratified into risk groups demonstrating the presence of predictive factors for CDL. Statistical analyses were performed, and chi-square analyses were used with a significance of p < 0.05. RESULTS: A total of 979 cholecystectomy patients were analyzed. The diagnosis of CDL was confirmed in 222 patients (22.7%) by magnetic resonance cholangiopancreatography, endoscopic retrograde cholangiopancreatography, or intraoperative cholangiography. Three predictive factors were identified: (1) Dilated common bile duct ≥6 mm; (2) Ultrasound with Choledocholithiasis; and (3) Total bilirubin ≥1.8 mg/dL (pediatric DUCT criteria). Risk groups were based on the number of predictive factors: very high (3), high (2), intermediate (1), and low (0). The pediatric DUCT criteria demonstrated accuracies of >76%, specificity of >78%, and negative predictive values of >79%. Adult factors (elevated aspartate aminotransferase/alanine aminotransferase, pancreatitis, BMI, and age) did not independently predict CDL. Based on risk stratification, the high- and very-high-risk groups demonstrated higher predictive capacity for CDL. CONCLUSIONS: Our study demonstrated that the pediatric DUCT criteria, incorporating common bile duct dilation, choledocholithiasis seen on ultrasound, and total bilirubin ≥1.8 mg/dL, highly predicts the presence of choledocholithiasis in children. Other adult preoperative factors are not predictive of common bile duct stone in children.

Child physical abuse and COVID-19: Trends from nine pediatric trauma centers.

BACKGROUND: Economic, social, and psychologic stressors are associated with an increased risk for abusive injuries in children. Prolonged physical proximity between adults and children under conditions of severe external stress, such as witnessed during the COVID-19 pandemic with "shelter-in-place orders", may be associated with additional increased risk for child physical abuse. We hypothesized that child physical abuse rates and associated severity of injury would increase during the early months of the pandemic as compared to the prior benchmark period. METHODS: We conducted a nine-center retrospective review of suspected child physical abuse admissions across the Western Pediatric Surgery Research Consortium. Cases were identified for the period of April 1-June 30, 2020 (COVID-19) and compared to the identical period in 2019. We collected patient demographics, injury characteristics, and outcome data. RESULTS: There were no significant differences in child physical abuse cases between the time periods in the consortium as a whole or at individual hospitals. There were no differences between the study periods with regard to patient characteristics, injury types or severity, resource utilization, disposition, or mortality. CONCLUSIONS: Apparent rates of new injuries related to child physical abuse did not increase early in the COVID-19 pandemic. While this may suggest that pediatric physical abuse was not impacted by pandemic restrictions and stresses, it is possible that under-reporting, under-detection, or delays in presentation of abusive injuries increased during the pandemic. Long-term follow-up of subsequent rates and severity of child abuse is needed to assess for unrecognized injuries that may have occurred.

Minimizing variance in pediatric surgical care through implementation of a perioperative colon bundle: A multi-institution retrospective cohort study.

BACKGROUND: Employing an institutional initiative to minimize variance in pediatric surgical care, we implemented a set of perioperative bundled interventions for all colorectal procedures to reduce surgical site infections (SSIs). METHODS: Implementation of a standard colon bundle at two children's hospitals began in December 2014. Subjects who underwent a colorectal procedure during the study period were analyzed. Demographics, outcomes, and complications were compared with Wilcoxon Rank-Sum, Chi-square and Fisher exact tests, as appropriate. Multivariable logistic regression was performed to assess the influence of time period (independent of protocol implementation) on the rate of subsequent infection. RESULTS: One hundred and forty-five patients were identified (preprotocol=68, postprotocol= 77). Gender, diagnosis, procedure performed and wound classification were similar between groups. Superficial SSIs (21% vs. 8%, p=0.031) and readmission (16% vs. 4%, p=0.021) were significantly decreased following implementation of a colon bundle. Median hospital days, cost, reoperation, intraabdominal abscess, and anastomotic leak were unchanged before and after protocol implementation (all p > 0.05). Multivariable logistic regression found time period to be independent of SSIs (OR: 0.810, 95% CI: 0.576-1.140). CONCLUSION: Implementation of a standard pediatric perioperative colon bundle can reduce superficial SSIs. Larger prospective studies are needed to evaluate the impact of colon bundles in reducing complications, hospital stay and cost. LEVEL OF EVIDENCE: III - Retrospective cohort study.

Minimizing variance in Care of Pediatric Blunt Solid Organ Injury through Utilization of a hemodynamic-driven protocol: a multi-institution study.

BACKGROUND: An expedited recovery protocol for management of pediatric blunt solid organ injury (spleen, liver, and kidney) was instituted across two Level 1 Trauma Centers, managed by nine pediatric surgeons within three hospital systems. METHODS: Data were collected for 18months on consecutive patients after protocol implementation. Patient demographics (including grade of injury), surgeon compliance, National Surgical Quality Improvement Program (NSQIP) complications, direct hospital cost, length of stay, time in the ICU, phlebotomy, and re-admission were compared to an 18-month control period immediately preceding study initiation. RESULTS: A total of 106 patients were treated (control=55, protocol=51). Demographics were similar among groups, and compliance was 78%. Hospital stay (4.6 vs. 3.5days, p=0.04), ICU stay (1.9 vs. 1.0days, p=0.02), and total phlebotomy (7.7 vs. 5.3 draws, p=0.007) were significantly less in the protocol group. A decrease in direct hospital costs was also observed ($11,965 vs. $8795, p=0.09). Complication rates (1.8% vs. 3.9%, p=0.86, no deaths) were similar. CONCLUSIONS: An expedited, hemodynamic-driven, pediatric solid organ injury protocol is achievable across hospital systems and surgeons. Through implementation we maintained quality while impacting length of stay, ICU utilization, phlebotomy, and cost. Future protocols should work to further limit resource utilization. TYPE OF STUDY: Retrospective cohort study. LEVEL OF EVIDENCE: Level II.