Enhanced Critical Congenital Cardiac Disease Screening by Combining Interpretable Machine Learning Algorithms.

Critical Congenital Heart Disease (CCHD) screening that only uses oxygen saturation (SpO2), measured by pulse oximetry, fails to detect an estimated 900 US newborns annually. The addition of other pulse oximetry features such as perfusion index (PIx), heart rate, pulse delay and photoplethysmography characteristics may improve detection of CCHD, especially those with systemic blood flow obstruction such as Coarctation of the Aorta (CoA). To comprehensively study the most relevant features associated with CCHD, we investigated interpretable machine learning (ML) algorithms by using Recursive Feature Elimination (RFE) to identify an optimal subset of features. We then incorporated the trained ML models into the current SpO2-alone screening algorithm. Our proposed enhanced CCHD screening system, which adds the ML model, improved sensitivity by approximately 10 percentage points compared to the current standard SpO2-alone method with minimal to no impact on specificity.Clinical relevance- This establishes proof of concept for a ML algorithm that combines pulse oximetry features to improve detection of CCHD with little impact on false positive rate.

Acute kidney injury in preterm infants with necrotizing enterocolitis.

Purpose: Acute kidney injury (AKI) is an independent predictor of morbidity and mortality in critically ill infants and children. AKI develops in an estimated one-third of the neonatal intensive care unit (NICU) population; however, literature on the incidence of AKI in premature infants with a diagnosis of necrotizing enterocolitis (NEC) is limited. The objectives of this study were to describe the incidence of AKI in infants with radiographically confirmed NEC, assess these infants for independent risk factors associated with development of AKI and evaluate if the presence of AKI is associated with increased mortality. Study design: We conducted a retrospective chart review of premature infants, gestational age (GA) 23-34 weeks, who developed modified Bell's level 2 or 3 NEC while admitted to two tertiary NICUs within our health system between 2010 and 2015. AKI was defined and staged according to modified Kidney Disease: Improving Global Outcomes (KDIGO) criteria. Results: 77 infants with Bell's level II (63.6%) and III (36.4%) NEC were studied. AKI occurred in 42.9% of infants (Stage 1: 18.2%; Stage 2: 13%; Stage 3: 11.7%). Bell's Stage III NEC, lower GA, maternal preeclampsia/eclampsia, gentamicin/vancomycin exposure, and empiric antibiotic use were independently associated with AKI. AKI was strongly associated with mortality (HR 20.3 95%CI 2.5-162.8, p = .005) in an adjusted Cox model. Conclusions: AKI is common in premature infants who develop NEC. More severe NEC was found to be an independent risk factor for AKI. Additionally, AKI in infants with NEC increases mortality risk significantly.

Results from the New Jersey statewide critical congenital heart defects screening program.

BACKGROUND AND OBJECTIVE: New Jersey was the first state to implement legislatively mandated newborn pulse oximetry screening (POxS) in all licensed birthing facilities to detect critical congenital heart defects (CCHDs). The objective of this report was to evaluate implementation of New Jersey's statewide POxS mandate. METHODS: A 2-pronged approach was used to collect data on infants screened in all New Jersey birthing facilities from August 31, 2011, through May 31, 2012. Aggregate screening results were submitted by each birthing facility. Data on failed screens and clinical characteristics of those newborns were reported to the New Jersey Birth Defects Registry (NJBDR). Three indicators were used to distinguish the added value of mandated POxS from standard clinical care: prenatal congenital heart defect diagnosis, cardiology consultation or echocardiogram indicated or performed before PoxS, or clinical findings at the time of POxS warranting a pulse oximetry measurement. RESULTS: Of 75,324 live births in licensed New Jersey birthing facilities, 73,320 were eligible for screening, of which 99% were screened. Forty-nine infants with failed POxS were reported to the NJBDR, 30 of whom had diagnostic evaluations solely attributable to the mandated screening. Three of the 30 infants had previously unsuspected CCHDs and 17 had other diagnoses or non-CCHD echocardiogram findings. CONCLUSIONS: In the first 9 months after implementation, New Jersey achieved a high statewide screening rate and established surveillance mechanisms to evaluate the unique contribution of POxS. The screening mandate identified 3 infants with previously unsuspected CCHDs that otherwise might have resulted in significant morbidity and mortality and also identified other significant secondary targets such as sepsis and pneumonia.

Statewide NICU central-line-associated bloodstream infection rates decline after bundles and checklists.

OBJECTIVE: In 2008, all 18 regional referral NICUs in New York state adopted central-line insertion and maintenance bundles and agreed to use checklists to monitor maintenance-bundle adherence and report checklist use. We sought to confirm whether adopting standardized bundles and using central-line maintenance checklists reduced central-line-associated bloodstream infections (CLABSI). METHODS: This was a prospective cohort study that enrolled all neonates with a central line who were hospitalized in any of 18 NICUs. Each NICU reported CLABSI and central-line utilization data and checklist use. We used χ(2) to compare CLABSI rates in the preintervention (January to December 2007) versus the postintervention (March to December 2009) periods and Poisson regression to model adjusted CLABSI rates. RESULTS: Each study period included more than 55 000 central-line days and more than 200 000 patient-days. CLABSI rates decreased 67% statewide (risk ratio: 0.33 [95% confidence interval: 0.27-0.41]; P < .0005); after adjusting for the altered central-line-associated bloodstream infection definition in 2008, by 40% (risk ratio: 0.60 [95% confidence interval: 0.48-0.75]; P < .0005). A total of 13 of 18 NICUs reported using maintenance checklists for 10% to 100% of central-line days. The checklist-use rate was associated with the CLABSI rate (coefficient: -0.57, P = .04). A total of 10 of 18 NICUs were independent CLABSI rate predictors, ranging from 1 site with greatly reduced risk (incidence rate ratio: 0.04, P < .0005) to 1 site with greatly increased risk (incidence rate ratio: 2.87, P < .0005). CONCLUSIONS: Although standardizing central-line care elements led to a significant statewide decline in NICU CLABSIs, site of care remains an independent risk factor. Using maintenance checklists reduced CLABSIs.

Effectiveness of pulse oximetry screening for congenital heart disease in asymptomatic newborns.

OBJECTIVE: To determine the sensitivity, specificity, predictive value, and accuracy of a program of pulse oximetry screening of asymptomatic newborns for critical congenital cardiovascular malformation (CCVM). METHODS: Pulse oximetry was performed on asymptomatic newborns in the well-infant nurseries of 2 hospitals. Cardiac ultrasound was performed on infants with positive screens (saturation 24 hours). Data regarding true and false positives as well as negatives were collected and analyzed. RESULTS: Oximetry was performed on 11 281 asymptomatic newborns, and 3 cases of CCVM were detected (total anomalous pulmonary venous return x2, truncus arteriosus). During the study interval, there were 9 live births of infants with CCVM from a group of 15 fetuses with CCVM detected by fetal echocardiography. Six infants with CCVM were symptomatic before screening. There was 1 false-positive screen. Two infants with negative screens were readmitted (coarctation, hypoplastic left pulmonary artery with aorto-pulmonary collaterals). Other cardiac diagnoses in the database search were nonurgent, including cases of patent foramen ovale, peripheral pulmonic stenosis, and ventricular septal defect. The prevalence of critical CCVM among all live births was 1 in 564 and among the screened population was 1 in 2256 (sensitivity: 60%; specificity: 99.95%; positive predictive value: 75%; negative predictive value: 99.98%; accuracy: 99.97%). CONCLUSIONS: This screening test is simple, noninvasive, and inexpensive and can be administered in conjunction with state-mandated screening. The false-negative screen patients had lesions not amenable to detection by oximetry. The sensitivity, specificity, and predictive value in this population are satisfactory, indicating that screening should be applied to larger populations, particularly where lower rates of fetal detection result in increased CCVM prevalence in asymptomatic newborns.