Obesity Heterogeneity by Neighborhood Context in a Largely Latinx Sample

Neighborhood socioeconomic context where Latinx children live may influence body weight status. Los Angeles County and Orange County of Southern California both are on the list of the top ten counties with the largest Latinx population in the USA. This heterogeneity allowed us to estimate differential impacts of neighborhood environment on children's body mass index z-scores by race/ethnicity using novel methods and a rich data source. We geocoded pediatric electronic medical record data from a predominantly Latinx sample and characterized neighborhoods into unique residential contexts using latent profile modeling techniques. We estimated multilevel linear regression models that adjust for comorbid conditions and found that a child's place of residence independently associates with higher body mass index z-scores. Interactions further reveal that Latinx children living in Middle-Class neighborhoods have higher BMI z-scores than Asian and Other Race children residing in the most disadvantaged communities. Our findings underscore the complex relationship between community racial/ethnic composition and neighborhood socioeconomic context on body weight status during childhood.

Artificial intelligence for nailfold capillaroscopy analyses - a proof of concept application in juvenile dermatomyositis

BACKGROUND

Biomarkers for idiopathic inflammatory myopathies are difficult to identify and may involve expensive laboratory tests. We assess the potential for artificial intelligence (AI) to differentiate children with juvenile dermatomyositis (JDM) from healthy controls using nailfold capillaroscopy (NFC) images. We also assessed the potential of NFC images to reflect the range of disease activity with JDM.

METHODS

A total of 1,120 NFC images from 111 children with active JDM, diagnosed between 1990 and 2020, and 321 NFC images from 31 healthy controls were retrieved from the CureJM JDM Registry. We built a lightweight and explainable deep neural network model called NFC-Net. Images were downscaled by interpolation techniques to reduce the computational cost.

RESULTS

NFC-Net achieved high performance in differentiating patients with JDM from controls, with an area under the ROC curve (AUROC) of 0.93 (0.84, 0.99) and accuracy of 0.91 (0.82, 0.92). With sensitivity (0.85) and specificity (0.90) resulted in model precision of 0.95. The AUROC and accuracy for predicting clinical disease activity from inactivity were 0.75 (0.61, 0.81) and 0.74 (0.65, 0.79).

CONCLUSION

The good performance of the NFC-Net demonstrates that NFC images are sufficient for detecting often unrecognized JDM disease activity, providing a reliable indicator of disease status.

IMPACT

Proposed NFC-Net can accurately predict children with JDM from healthy controls using nailfold capillaroscopy (NFC) images. Additionally, it predicts the scores to JDM disease activity versus no activity. Equipped with gradients, NFC-Net is explainable and gives visual information beside the reported accuracies. NFC-Net is computationally efficient since it is applied to substantially downscaled NFC images. Furthermore, the model can be wrapped within an edge-based device like a mobile application that is accessible to both clinicians and patients.

Congenital anomalies and predisposition to severe COVID-19 among pediatric patients in the United States

BACKGROUND AND OBJECTIVE

Congenital heart defects are known to be associated with increased odds of severe COVID-19. Congenital anomalies affecting other body systems may also be associated with poor outcomes. This study is an exhaustive assessment of congenital anomalies and odds of severe COVID-19 in pediatric patients.

METHODS

Data were retrieved from the COVID-19 dataset of Cerner® Real-World Data for encounters from March 2020 to February 2022. Prior to matching, the data consisted of 664,523 patients less than 18 years old and 927,805 corresponding encounters with COVID-19 from 117 health systems across the United States. One-to-one propensity score matching was performed, and a cumulative link mixed-effects model with random intercepts for health system and patients was built to assess corresponding associations.

RESULTS

All congenital anomalies were associated with worse COVID-19 outcomes, with the strongest association observed for cardiovascular anomalies (odds ratio [OR], 3.84; 95% CI, 3.63-4.06) and the weakest association observed for anomalies affecting the eye/ear/face/neck (OR, 1.16; 95% CI, 1.03-1.31).

CONCLUSIONS AND RELEVANCE

Congenital anomalies are associated with greater odds of experiencing severe symptoms of COVID-19. In addition to congenital heart defects, all other birth defects may increase the odds for more severe COVID-19.

IMPACT

All congenital anomalies are associated with increased odds of severe COVID-19. This study is the largest and among the first to investigate birth defects across all body systems. The multicenter large data and analysis demonstrate the increased odds of severe COVID19 in pediatric patients with congenital anomalies affecting any body system. These data demonstrate that all children with birth defects are at increased odds of more severe COVID-19, not only those with heart defects. This should be taken into consideration when optimizing prevention and intervention resources within a hospital.

Chronic Neurological Disorders and Predisposition to Severe COVID-19 in Pediatric Patients in the United States

BACKGROUND

We investigated the association between chronic pediatric neurological conditions and the severity of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2).

METHODS

This matched retrospective case-control study includes patients (n = 71,656) with chronic complex neurological disorders under 18 years of age, with laboratory-confirmed diagnosis of COVID-19 or a diagnostic code indicating infection or exposure to SARS-CoV-2, from 103 health systems in the United States. The primary outcome was the severity of coronavirus disease 2019 (COVID-19), which was classified as severe (invasive oxygen therapy or death), moderate (noninvasive oxygen therapy), or mild/asymptomatic (no oxygen therapy). A cumulative link mixed effects model was used for this study.

RESULTS

In this study, a cumulative link mixed effects model (random intercepts for health systems and patients) showed that the following classes of chronic neurological disorders were associated with higher odds of severe COVID-19: muscular dystrophies and myopathies (OR = 3.22; 95% confidence interval [CI]: 2.73 to 3.84), chronic central nervous system disorders (OR = 2.82; 95% CI: 2.67 to 2.97), cerebral palsy (OR = 1.97; 95% CI: 1.85 to 2.10), congenital neurological disorders (OR = 1.86; 95% CI: 1.75 to 1.96), epilepsy (OR = 1.35; 95% CI: 1.26 to 1.44), and intellectual developmental disorders (OR = 1.09; 95% CI: 1.003 to 1.19). Movement disorders were associated with lower odds of severe COVID-19 (OR = 0.90; 95% CI: 0.81 to 0.99).

CONCLUSIONS

Pediatric patients with chronic neurological disorders are at higher odds of severe COVID-19. Movement disorders were associated with lower odds of severe COVID-19.

A causal inference study: The impact of the combined administration of Donepezil and Memantine on decreasing hospital and emergency department visits of Alzheimer's disease patients

Alzheimer's disease is the most common type of dementia that currently affects over 6.5 million people in the U.S. Currently there is no cure and the existing drug therapies attempt to delay the mental decline and improve cognitive abilities. Two of the most commonly prescribed such drugs are Donepezil and Memantine. We formally tested and confirmed the presence of a beneficial drug-drug interaction of Donepezil and Memantine using a causal inference analysis. We applied doubly robust estimators to one of the largest and high-quality medical databases to estimate the effect of two commonly prescribed Alzheimer's disease (AD) medications, Donepezil and Memantine, on the average number of hospital or emergency department visits per year among patients diagnosed with AD. Our results show that, compared to the absence of medication scenario, the Memantine monotherapy, and the Donepezil monotherapy, the combined use of Donepezil and Memantine treatment significantly reduces the average number of hospital or emergency department visits per year by 0.078 (13.8%), 0.144 (25.5%), and 0.132 days (23.4%), respectively. The assessed decline in the average number of hospital or emergency department visits per year is consequently associated with a substantial reduction in medical costs. As of 2022, according to the Alzheimer's Disease Association, there were over 6.5 million individuals aged 65 and older living with AD in the US alone. If patients who are currently on no drug treatment or using either Donepezil or Memantine alone were switched to the combined used of Donepezil and Memantine therapy, the average number of hospital or emergency department visits could decrease by over 613 thousand visits per year. This, in turn, would lead to a remarkable reduction in medical expenses associated with hospitalization of AD patients in the US, totaling over 940 million dollars per year.

Neighborhood disadvantage and pediatric inpatient opioid prescription patterns

BACKGROUND

To explore the role of children's residential environment on opioid prescribing patterns in a predominantly Latinx sample.

METHODS

We connected geocoded data from electronic medical records in a diverse sample of pediatric patients to neighborhood environments constructed using latent profile modeling techniques. We then estimated a series of multilevel models to determine whether opioid prescribing patterns vary by residential context.

RESULTS

A stepwise pattern exists between neighborhood disadvantage and pediatric opioid prescription patterns, such that higher levels of disadvantage associate with a greater likelihood of opioid prescription, independent of the patient's individual profile.

CONCLUSION

In a largely Latinx sample of children, the neighborhood in which a child lives influences whether or not they will receive opioids. Considering the differences in patient residential environment may reduce variation in opioid dispensing rates among pediatric patients.

A pilot study to improve provider adherence to NAEPP guidelines

The prevalence and morbidity of Asthma in the United States has increased since the 1991 National Asthma Education and Prevention Program (NAEPP) and updated Expert Panel Report -3 (EPR-3) guidelines in 2007 were published. To improve provider adherence to the NAEPP EPR-3 guidelines Children's Hospital of Orange County (CHOC) in California integrated the HealtheIntentSM Pediatric Asthma Registry (PAR) into the electronic medical record (EMR) in 2015.

METHODS

A serial cross-sectional design was used to compare provider management of CHOC MediCal asthma patients before 2014 (N = 6606) and after 2018 (N = 6945) integration of the Registry with NAEPP guidelines into the EMR. Four provider adherence measures (Asthma Control Test [ACT], Asthma Action Plan [AAP], inhaled corticosteroids [ICS] and spacers) were evaluated using General Linear Mixed Models and Chi square.

FINDINGS

In 2018, patients were more likely to receive an ACT, (OR = 14.95, 95% CI 12.67, 17.65, p < .001), AAP (OR = 12.70, 95% CI 11.10, 14.54, p < .001), ICS (OR = 1.85, 95% CI 8.52, 14.54, p < .001) and spacer (OR = 1.45, 95% CI 1.31, 1.6, p < .001) compared to those in 2014.

DISCUSSION

The pilot study showed integration of the Pediatric Asthma Registry into the EMR, as a computer decision support tool that was an effective intervention to increase provider adherence to NAEPP guidelines. Ongoing monitoring and education are needed to promote and sustain provider behavioral change. Additional research to include multi-sites and decreased time between evaluation years is recommended.

APPLICATION TO PRACTICE

Can be used for excellent health policy decision making as a direct impact on patient care and outcomes, by improving provider adherence to the NAEPP guidelines.

Caregiver Satisfaction with Emergency Department Care for Pediatric Patients with Neurodevelopmental Disorders

OBJECTIVE

Children with neurodevelopmental disorders (NDDs) often encounter increased adversity when navigating the health care system. In this study, we explored the pediatric emergency department (PED) experience for patients with NDDs and their caregivers compared with that of patients without NDDs.

METHODS

Data for this study were obtained from National Research Corporation patient experience survey questionnaires and electronic medical record (EMR) data for patients presenting to a PED between May 2018 and September 2019. ED satisfaction was determined by the top-box approach; ED ratings of 9/10 or 10/10 were considered to reflect high ED satisfaction. Demographics, Emergency Severity Index, ED length of stay, time from arrival to triage, time to provider assessment, and diagnoses were extracted from the EMR. Patients with NDDs were identified based on International Classification of Diseases, Tenth Revision codes; patients with intellectual disabilities, pervasive and specific developmental disorders, or attention-deficit/hyperactivity disorders were included in the NDD cohort. One-to-one propensity score matching between patients with and without NDDs was performed, and a multivariable logistic regression model was built on the matched cohort.

RESULTS

Patients with NDDs represented over 7% of survey respondents. Matching was successful for 1162 patients with NDDs (99.5%), resulting in a matched cohort sample size of 2324. Caregivers of patients with NDDs had 25% lower odds of reporting high ED satisfaction (95% confidence interval [CI], 0.62-0.91, p = 0.004).

CONCLUSION

Caregivers of patients with NDDs make up a significant proportion of survey respondents and are more likely to rate the ED poorly than caregivers of patients without NDDs. This suggests an opportunity for targeted interventions in this population to improve patient care and experience.

Birthweight and gestational age as modifiers of rehospitalization after NICU admission

OBJECTIVE

To assess interaction effects between gestational age and birth weight on 30-day unplanned hospital readmission following discharge from the neonatal intensive care unit (NICU). To assess the relationship between gestational age and day of life (DOL) on neonatal intensive care unit (NICU) admission and the risk for 30-day unplanned hospital readmission (UHR).

METHODS

Retrospective study using the study site's Children's Hospitals Neonatal Database (CHND) and electronic health records were used. Population included patients discharged from a NICU between January 2017 and March 2020. Variables encompassing demographics, gestational age, birth weight, medications, maternal data, and surgical procedures were controlled for. A statistical interaction between gestational age and DOL birth weight was tested for statistical significance.

RESULTS

2,307 neonates were included, with 7.2% readmitted within 30 days of discharge. Statistical interaction between birth weight and gestational age was statistically significant, indicating that the odds of readmission amongst low-birth-weight premature patients increases with increasing gestational age, while it decreases with increasing gestational age amongst their normal or high birth weight peers.

CONCLUSIONS

The effect of gestational age on odds of hospital readmission is dependent on birth weight.

The von Willebrand Factor Antigen Reflects the Juvenile Dermatomyositis Disease Activity Score

OBJECTIVE

This study determined if an accessible, serologic indicator of vascular disease activity, the von Willebrand factor antigen (vWF:Ag), was useful to assess disease activity in children with juvenile dermatomyositis (JDM), a rare disease, but the most common of the pediatric inflammatory myopathies.

METHODS

A total of 305 children, median age 10 years, 72.5% female, 76.5% white, with definite/probable JDM at diagnosis, were enrolled in the Ann & Robert H. Lurie Cure JM Juvenile Myositis Repository, a longitudinal database. Disease Activity Score (DAS) and vWF:Ag data were obtained at each visit. These data were analyzed using generalized estimating equation (GEE) models (both linear and logistic) to determine if vWF:Ag reflects disease severity in children with JDM. A secondary analysis was performed for untreated active JDM to exclude the effect of medications on vWF:Ag.

RESULT

The vWF:Ag test was elevated in 25% of untreated JDM. We found that patients with elevated vWF:Ag had a 2.55-fold higher DAS total (CI₉₅: 1.83-3.27, p < 0.001). Patients with difficulty swallowing had 2.57 higher odds of elevated vWF:Ag (CI_95: 1.5-4.38, p < 0.001); those with more generalized skin involvement had 2.58-fold higher odds of elevated vWF:Ag (CI₉₅: 1.27-5.23, p = 0.006); and those with eyelid peripheral blood vessel dilation had 1.32-fold higher odds of elevated vWF:Ag (CI₉₅: 1.01-1.72, p = 0.036). Untreated JDM with elevated vWF:Ag had more muscle weakness and higher muscle enzymes, neopterin and erythrocyte sedimentation rate compared to JDM patients with a normal vWF:Ag.

CONCLUSION

vWF:Ag elevation is a widely accessible concomitant of active disease in 25% of JDM.

An Introduction to Artificial Intelligence in Developmental and Behavioral Pediatrics

ABSTRACT

Technological breakthroughs, together with the rapid growth of medical information and improved data connectivity, are creating dramatic shifts in the health care landscape, including the field of developmental and behavioral pediatrics. While medical information took an estimated 50 years to double in 1950, by 2020, it was projected to double every 73 days. Artificial intelligence (AI)-powered health technologies, once considered theoretical or research-exclusive concepts, are increasingly being granted regulatory approval and integrated into clinical care. In the United States, the Food and Drug Administration has cleared or approved over 160 health-related AI-based devices to date. These trends are only likely to accelerate as economic investment in AI health care outstrips investment in other sectors. The exponential increase in peer-reviewed AI-focused health care publications year over year highlights the speed of growth in this sector. As health care moves toward an era of intelligent technology powered by rich medical information, pediatricians will increasingly be asked to engage with tools and systems underpinned by AI. However, medical students and practicing clinicians receive insufficient training and lack preparedness for transitioning into a more AI-informed future. This article provides a brief primer on AI in health care. Underlying AI principles and key performance metrics are described, and the clinical potential of AI-driven technology together with potential pitfalls is explored within the developmental and behavioral pediatric health context.

Timely orchiopexy by 18 months of age: Are we meeting the standards defined by the 2014 AUA guidelines?

BACKGROUND

Cryptorchidism is one of the most common reasons for pediatric urology referral and one of the few pediatric urologic conditions in which there are established AUA guidelines that recommend orchiopexy be performed before 18 months of age. While access to timely orchiopexy has been studied previously, there is no current study with data from a national clinical database evaluating timely orchiopexy after the AUA guidelines were published. Additionally, prior studies on delayed orchiopexy may have included patients with an ascended testis, which is a distinct population from those with true undescended testicles.

OBJECTIVES

To evaluate in a national, clinical database if timely orchiopexy improved after the AUA guidelines were published in 2014. In particular, we aim to evaluate a younger group of patients, 0-5 years of age, in an effort to account for potential ascending testes.

STUDY DESIGN

Using Cerner Real-World Data™, a national, de-identified database of 153 million individuals, we analyzed pediatric patients undergoing orchiopexy in the United States from 2000 to 2021. We included males 0-18 years old and further focused on the subset 0-5 years. Primary outcome was timely orchiopexy, defined as age at orchiopexy less than 18 months. Predictor variables included race, ethnicity and insurance status. Statistical analyses were performed using logistic regression.

RESULTS

Of the total 17,012 individuals identified as undergoing orchiopexy, 9274 were ages 0-5 at the time of surgery. Comparing time periods pre and post AUA guidelines (2000-2014 versus 2015-2021), we found a significant difference in the proportion of timely orchiopexy (51% versus 56%, respectively; p < 0.0001) (Figure). In multivariable analyses, Hispanic (OR = 0.65, p < 0.0001), African American (OR = 0.74, p < 0.0001), and Native American males (OR = 0.66, p = 0.008) were less likely to have timely orchiopexy compared to non-Hispanic White males. Individuals without insurance (OR = 0.81, p = 0.03) or with public insurance (OR = 0.88, p = 0.02) were less likely to have timely orchiopexy as compared to those with private insurance.

CONCLUSIONS

Nearly a decade after publication of the AUA cryptorchidism guidelines, a large proportion of patients are still not undergoing orchiopexy by 18 months of age. This is the first study to show that timely orchiopexy has improved among patients 0-5 years, but the majority of patients are still not undergoing timely orchiopexy. Health disparities were apparent among Hispanic, African American, Native American, and uninsured males, highlighting the need for further progress in access to pediatric surgical care.

Epidemiology of Neonatal COVID-19 in the United States

OBJECTIVES

Data on coronavirus disease 2019 (COVID-19) infections in neonates are limited. We aimed to identify and describe the incidence, presentation, and clinical outcomes of neonatal COVID-19.

METHODS

Over 1 million neonatal encounters at 109 United States health systems, from March 2020 to February 2021, were extracted from the Cerner Real World Database. COVID-19 diagnosis was assessed using severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) laboratory tests and diagnosis codes. Incidence of COVID-19 per 100 000 encounters was estimated.

RESULTS

COVID-19 was diagnosed in 918 (0.1%) neonates (91.1 per 100 000 encounters [95% confidence interval 85.3-97.2]). Of these, 71 (7.7%) had severe infection (7 per 100 000 [95% confidence interval 5.5-8.9]). Median time to diagnosis was 14.5 days from birth (interquartile range 3.1-24.2). Common signs of infection were tachypnea and fever. Those with severe infection were more likely to receive respiratory support (50.7% vs 5.2%, P < .001). Severely ill neonates received analgesia (38%), antibiotics (33.8%), anticoagulants (32.4%), corticosteroids (26.8%), remdesivir (2.8%), and COVID-19 convalescent plasma (1.4%). A total of 93.6% neonates were discharged home after care, 1.1% were transferred to another hospital, and discharge disposition was unknown for 5.2%. One neonate (0.1%) with presentation suggestive of multisystem inflammatory syndrome in children died after 11 days of hospitalization.

CONCLUSIONS

Most neonates infected with SARS-CoV-2 were asymptomatic or developed mild illness without need for respiratory support. Some had severe illness requiring treatment of COVID-19 with remdesivir and COVID-19 convalescent plasma. SARS-CoV-2 infection in neonates, though rare, may result in severe disease.

A comparative study on deep learning models for text classification of unstructured medical notes with various levels of class imbalance

BACKGROUND

Discharge medical notes written by physicians contain important information about the health condition of patients. Many deep learning algorithms have been successfully applied to extract important information from unstructured medical notes data that can entail subsequent actionable results in the medical domain. This study aims to explore the model performance of various deep learning algorithms in text classification tasks on medical notes with respect to different disease class imbalance scenarios.

METHODS

In this study, we employed seven artificial intelligence models, a CNN (Convolutional Neural Network), a Transformer encoder, a pretrained BERT (Bidirectional Encoder Representations from Transformers), and four typical sequence neural networks models, namely, RNN (Recurrent Neural Network), GRU (Gated Recurrent Unit), LSTM (Long Short-Term Memory), and Bi-LSTM (Bi-directional Long Short-Term Memory) to classify the presence or absence of 16 disease conditions from patients' discharge summary notes. We analyzed this question as a composition of 16 binary separate classification problems. The model performance of the seven models on each of the 16 datasets with various levels of imbalance between classes were compared in terms of AUC-ROC (Area Under the Curve of the Receiver Operating Characteristic), AUC-PR (Area Under the Curve of Precision and Recall), F1 Score, and Balanced Accuracy as well as the training time. The model performances were also compared in combination with different word embedding approaches (GloVe, BioWordVec, and no pre-trained word embeddings).

RESULTS

The analyses of these 16 binary classification problems showed that the Transformer encoder model performs the best in nearly all scenarios. In addition, when the disease prevalence is close to or greater than 50%, the Convolutional Neural Network model achieved a comparable performance to the Transformer encoder, and its training time was 17.6% shorter than the second fastest model, 91.3% shorter than the Transformer encoder, and 94.7% shorter than the pre-trained BERT-Base model. The BioWordVec embeddings slightly improved the performance of the Bi-LSTM model in most disease prevalence scenarios, while the CNN model performed better without pre-trained word embeddings. In addition, the training time was significantly reduced with the GloVe embeddings for all models.

CONCLUSIONS

For classification tasks on medical notes, Transformer encoders are the best choice if the computation resource is not an issue. Otherwise, when the classes are relatively balanced, CNNs are a leading candidate because of their competitive performance and computational efficiency.

Cerner real-world data (CRWD) - A de-identified multicenter electronic health records database

Cerner Real-World Data^TM (CRWD) is a de-identified big data source of multicenter electronic health records. Cerner Corporation secured appropriate data use agreements and permissions from more than 100 health systems in the United States contributing to the database as of March 2022. A subset of the database was extracted to include data from only patients with SARS-CoV-2 infections and is referred to as the Cerner COVID-19 Dataset. The December 2021 version of CRWD consists of 100 million patients and 1.5 billion encounters across all care settings. There are 2.3 billion, 2.9 billion, 486 million, and 11.5 billion records in the condition, medication, procedure, and lab (laboratory test) tables respectively. The 2021 Q3 COVID-19 Dataset consists of 130.1 million encounters from 3.8 million patients. The size and longitudinal nature of CRWD can be leveraged for advanced analytics and artificial intelligence in medical research across all specialties and is a rich source of novel discoveries on a wide range of conditions including but not limited to COVID-19.

Association of Congenital and Acquired Cardiovascular Conditions With COVID-19 Severity Among Pediatric Patients in the US

IMPORTANCE

Identifying the associations between severe COVID-19 and individual cardiovascular conditions in pediatric patients may inform treatment.

OBJECTIVE

To assess the association between previous or preexisting cardiovascular conditions and severity of COVID-19 in pediatric patients.

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study used data from a large, multicenter, electronic health records database in the US. The cohort included patients aged 2 months to 17 years with a laboratory-confirmed diagnosis of COVID-19 or a diagnosis code indicating infection or exposure to SARS-CoV-2 at 85 health systems between March 1, 2020, and January 31, 2021.

EXPOSURES

Diagnoses for 26 cardiovascular conditions between January 1, 2015, and December 31, 2019 (before infection with SARS-CoV-2).

MAIN OUTCOMES AND MEASURES

The main outcome was severe COVID-19, defined as need for supplemental oxygen or in-hospital death. Mixed-effects, random intercept logistic regression modeling assessed the significance and magnitude of associations between 26 cardiovascular conditions and COVID-19 severity. Multiple comparison adjustment was performed using the Benjamini-Hochberg false discovery rate procedure.

RESULTS

The study comprised 171 416 pediatric patients; the median age was 8 years (IQR, 2-14 years), and 50.28% were male. Of these patients, 17 065 (9.96%) had severe COVID-19. The random intercept model showed that the following cardiovascular conditions were associated with severe COVID-19: cardiac arrest (odds ratio [OR], 9.92; 95% CI, 6.93-14.20), cardiogenic shock (OR, 3.07; 95% CI, 1.90-4.96), heart surgery (OR, 3.04; 95% CI, 2.26-4.08), cardiopulmonary disease (OR, 1.91; 95% CI, 1.56-2.34), heart failure (OR, 1.82; 95% CI, 1.46-2.26), hypotension (OR, 1.57; 95% CI, 1.38-1.79), nontraumatic cerebral hemorrhage (OR, 1.54; 95% CI, 1.24-1.91), pericarditis (OR, 1.50; 95% CI, 1.17-1.94), simple biventricular defects (OR, 1.45; 95% CI, 1.29-1.62), venous embolism and thrombosis (OR, 1.39; 95% CI, 1.11-1.73), other hypertensive disorders (OR, 1.34; 95% CI, 1.09-1.63), complex biventricular defects (OR, 1.33; 95% CI, 1.14-1.54), and essential primary hypertension (OR, 1.22; 95% CI, 1.08-1.38). Furthermore, 194 of 258 patients (75.19%) with a history of cardiac arrest were younger than 12 years.

CONCLUSIONS AND RELEVANCE

The findings suggest that some previous or preexisting cardiovascular conditions are associated with increased severity of COVID-19 among pediatric patients in the US and that morbidity may be increased among individuals children younger than 12 years with previous cardiac arrest.

Southern California Pediatric and Adolescent Cancer Survivorship (SC-PACS): Establishing a Multi-Institutional Childhood, Adolescent, and Young Adult Cancer Survivorship Consortium in Southern California

Introduction

Given their risk for late effects and early mortality, childhood/adolescent cancer survivors (CACSs) should receive longitudinal monitoring and care. The Southern California Pediatric and Adolescent Cancer Survivorship (SC-PACS) consortium was established in February 2017 to combine resources and expertise across seven participating survivorship programs. Its over-arching objective is to address the unique needs of its demographically diverse CACS population through collaborative survivorship research and care initiatives. The first SC-PACS study was an assessment of survivorship needs and evaluation of current services as reported by CACSs and their parents/primary care givers (PPCGs) receiving survivorship care at consortium sites.

Methods

As an initial investigation, a cross-sectional survey for CACSs and their parents/primary care givers was conducted. The goal was to enroll 10 CACSs and 10 PPCGs from each of the seven institutions (total of 140 participants). The eligibility criteria for CACSs were age ≥13 years at the time of enrollment, >2 years from the end of treatment, sufficient cognitive function to complete the survey, and English or Spanish language proficiency. For CACSs <13 years old, their PPCGs completed the survey. This was a convenience sample using frequencies and proportions to describe participant characteristics and survey responses, which were entered into a Research Electronic Data Capture (REDCap) database.

Results

Across the consortium, of the recruitment target of 140 participants (CACSs, n=70; PPCGs, n=70), 127 (90.7%) participants were enrolled. Of the 127 participants enrolled, 65 (51.2%) were CACSs and 62 (48.8%) were PPCGs. The majority of participants were female (51.2%), were Hispanic (62.2%), spoke English as the primary language at home (57.5%), and were diagnosed between one to four years of age (45.7%). Information considered most important by both CACSs and PPCGs was related to cancer diagnosis (90.8%) and future risks as a result of cancer treatment received (98.0%). Overall, 78% of CACSs and PPCGs found the survivorship information (treatment summary) useful, and 83% felt that they received the right amount of information about their cancer.

Conclusion

Our aim was to obtain baseline data that would characterize our CACS population, inform consortium priorities, and establish a collaborative research platform. The ultimate goal of the consortium is to develop a comprehensive survivorship care approach that addresses the most important needs of cancer survivors in our catchment area and promotes best practice interventions. Future plans are to expand the needs assessment survey to obtain a wider representation of the survivor population at SC-PACS institutions, helping create strategies to improve cancer-specific education, delivery of treatment summary, and access to community resources for this demographically and socioeconomically diverse population.

Multivariable Analysis of Patient Satisfaction in the Pediatric Emergency Department

BACKGROUND

Published data on predictive factors associated with parent satisfaction from care in a pediatric emergency department (ED) visit are limited to be descriptive and obtained from small data sets. Accordingly, the purpose of this study was to determine both modifiable and nonmodifiable demographic and operational factors that influence parental satisfaction using a large and ethnically diverse site data set.

METHODS

Data consist of responses to the National Research Council (NRC) survey questionnaires and electronic medical records of 15,895 pediatric patients seen in a pediatric ED between the ages of 0 and 17 years discharged from May 2018 to September 2019. Bivariate, χ2, and multivariable logistic regression analyses were carried out using the NRC item on rating the ED between 0 and 10 as the primary outcome. Responses were coded using a top-box approach, a response of "9" or "10" represented satisfaction with the facility, and every other response was indicated as undesirable. Demographic data and NRC questionnaire were used as potential predictors.

RESULTS

Multivariable regression analysis found the following variables as independent predictors for positive parental rating of the ED: Hispanic race/ethnicity (odds ratio [OR], 1.285), primary language Spanish (OR, 2.399), and patients who had government-sponsored insurance (OR, 1.470). Those survey items with the largest effect size were timeliness of care (OR, 0.188) and managing discomfort (OR, 0.412).

CONCLUSIONS

Parental rating of an ED is associated with nonmodifiable variables such as ethnicity and modifiable variables such as timeliness of care and managing discomfort.

Standardized Health data and Research Exchange (SHaRE): promoting a learning health system

Aggregate de-identified data from electronic health records (EHRs) provide a valuable resource for research. The Standardized Health data and Research Exchange (SHaRE) is a diverse group of US healthcare organizations contributing to the Cerner Health Facts (HF) and Cerner Real-World Data (CRWD) initiatives. The 51 facilities at the 7 founding organizations have provided data about more than 4.8 million patients with 63 million encounters to HF and 7.4 million patients and 119 million encounters to CRWD. SHaRE organizations unmask their organization IDs and provide 3-digit zip code (zip3) data to support epidemiology and disparity research. SHaRE enables communication between members, facilitating data validation and collaboration as we demonstrate by comparing imputed EHR module usage to actual usage. Unlike other data sharing initiatives, no additional technology installation is required. SHaRE establishes a foundation for members to engage in discussions that bridge data science research and patient care, promoting the learning health system.

Targeted Clinical Interventions for Reducing Pediatric Readmissions

BACKGROUND

In this interventional study, we addressed the selection and application of clinical interventions on pediatric patients identified as at risk by a predictive model for readmissions.

METHODS

A predictive model for readmissions was implemented, and a team of providers expanded corresponding clinical interventions for at-risk patients at a freestanding children's hospital. Interventions encompassed social determinants of health, outpatient care, medication reconciliation, inpatient and discharge planning, and postdischarge calls and/or follow-up. Statistical process control charts were used to compare readmission rates for the 3-year period preceding adoption of the model and clinical interventions with those for the 2-year period after adoption of the model and clinical interventions. Potential financial savings were estimated by using national estimates of the cost of pediatric inpatient readmissions.

RESULTS

The 30-day all-cause readmission rates during the periods before and after predictive modeling (and corresponding 95% confidence intervals [CI]) were 12.5% (95% CI: 12.2%-12.8%) and 11.1% (95% CI: 10.8%-11.5%), respectively. More modest but similar improvements were observed for 7-day readmissions. Statistical process control charts indicated nonrandom reductions in readmissions after predictive model adoption. The national estimate of the cost of pediatric readmissions indicates an associated health care savings due to reduced 30-day readmission during the 2-year predictive modeling period at $2 673 264 (95% CI: $2 612 431-$2 735 364).

CONCLUSIONS

A combination of predictive modeling and targeted clinical interventions to improve the management of pediatric patients at high risk for readmission was successful in reducing the rate of readmission and reducing overall health care costs. The continued prioritization of patients with potentially modifiable outcomes is key to improving patient outcomes.

A Multivariable Model of Parent Satisfaction, Pain, and Opioid Administration in a Pediatric Emergency Department

Introduction

Children and adolescents are not impervious to the unprecedented epidemic of opioid misuse in the United States. In 2016 more than 88,000 adolescents between the ages of 12–17 reported misusing opioid medication, and evidence suggests that there has been a rise in opioid-related mortality for pediatric patients. A major source of prescribed opioids for the treatment of pain is the emergency department (ED). The current study sought to assess the complex relationship between opioid administration, pain severity, and parent satisfaction with children’s care in a pediatric ED.

Methods

We examined data from a tertiary pediatric care facility. A health survey questionnaire was administered after ED discharge to capture the outcome of parental likelihood of providing a positive facility rating. We abstracted patient demographic, clinical, and top diagnostic information using electronic health records. Data were merged and multivariable models were constructed.

Results

We collected data from 15,895 pediatric patients between the ages of 0–17 years (mean = 6.69; standard deviation = 5.19) and their parents. Approximately 786 (4.94%) patients were administered an opioid; 8212 (51.70%) were administered a non-opioid analgesic; and 3966 (24.95%) expressed clinically significant pain (pain score >/= 4). Results of a multivariable regression analysis from these pediatric patients revealed a three-way interaction of age, pain severity, and opioid administration (odds ratio 1.022, 95% confidence interval, 1.006, 1.038, P = 0.007). Our findings suggest that opioid administration negatively impacted parent satisfaction of older adolescent patients in milder pain who were administered an opioid analgesic, but positively influenced the satisfaction scores of parents of younger children who were administered opioids. When pain levels were severe, the relationship between age and patient experience was not statistically significant.

Conclusion

This investigation highlights the complexity of the relationship between opioid administration, pain severity, and satisfaction, and suggests that the impact of opioid administration on parent satisfaction is a function of the age of the child.

Long-term effects of malnutrition on severity of COVID-19

The COVID-19 pandemic is a public health crisis that has the potential to exacerbate worldwide malnutrition. This study examines whether patients with a history of malnutrition are predisposed to severe COVID-19. To do so, data on 103,099 COVID-19 inpatient encounters from 56 hospitals in the United States between March 2020 and June 2020 were retrieved from the Cerner COVID-19 Dataset. Patients with a history of malnutrition between 2015 and 2019 were identified, and a random intercept logistic regression models for pediatric and adult patients were built controlling for patient demographics, socioeconomic status, admission vital signs, and related comorbidities. Statistical interactions between malnutrition and patient age were significant in both the pediatric [log-odds and 95% confidence interval: 0.094 (0.012, 0.175)] and adult [- 0.014 (- 0.021, - 0.006] models. These interactions, together with the main effect terms of malnutrition and age, imply higher odds for severe COVID-19 for children between 6 and 17 years with history of malnutrition. Even higher odds of severe COVID-19 exist for adults (with history of malnutrition) between 18 and 79 years. These results indicate that the long-term effect of malnutrition predisposes patients to severe COVID-19 in an age-dependent way.

Safety and Antibody Kinetics of COVID-19 Convalescent Plasma for the Treatment of Moderate to Severe Cases of SARS-CoV-2 Infection in Pediatric Patients

BACKGROUND

Therapies against severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) and its life-threatening respiratory infection coronavirus disease 2019 (COVID-19) have been evaluated, including COVID-19 convalescent plasma (CCP). Multiple large reports of CCP treatment in adults exist. Pediatric data on CCP safety and efficacy are limited.

METHODS

Single-center prospective, open-label trial looking at safety, antibody kinetics and outcomes of CCP (10 mL/kg, max 1 unit) treatment for COVID-19 in hospitalized pediatric patients with moderate to severe disease or at high-risk for serious illness.

RESULTS

Thirteen patients were enrolled. No infusion-related adverse events occurred. No hematologic or metabolic adverse events were noted during hospitalization or at 3-weeks. Ten patients had clinical improvement by day 7 (WHO eight-category ordinal severity scale for COVID-19). Following CCP, anti-SARS-CoV-2 anti-nucleocapsid IgG increased significantly at 24 hours and high levels were sustained at 7- and 21-days. Transient IgM response was noted. Twelve patients (92.3%) were discharged home, 9 (75%) by day 7 post-CCP. One remained on invasive ventilatory support 42 days after CCP and was eventually discharged to an intermediate care facility. The single patient death was retrospectively confirmed to have had brain death before CCP.

CONCLUSION

CCP was well tolerated in pediatric patients, resulted in rapid antibody increase, and did not appear to interfere with immune responses measured at 21 days. More pediatric data are necessary to establish the efficacy of CCP, but our data suggest benefit in moderate to severe COVID-19 when used early. Other immunologic or antiviral interventions may be added as supported by emerging data.

Criticality index conducted in pediatric emergency department triage

OBJECTIVE

To develop and analyze the performance of a machine learning model capable of predicting the disposition of patients presenting to a pediatric emergency department (ED) based on triage assessment and historical information mined from electronic health records.

METHODS

We retrospectively reviewed data from 585,142 ED visits at a pediatric quaternary care institution between 2013 and 2020. An extreme gradient boosting machine learning model was trained on a randomly selected training data set (50%) to stratify patients into 3 classes: (1) high criticality (patients requiring intensive care unit [ICU] care within 4 h of hospital admission, patients who died within 4 h of admission, and patients who died in the ED); (2) moderate criticality (patients requiring hospitalization without the need for ICU care); and (3) low criticality (patients discharged home). Variables considered during model development included triage vital signs, aspects of triage nursing assessment, demographics, and historical information (diagnoses, medication use, and healthcare utilization). Historical factors were limited to the 6 months preceding the index ED visit. The model was tested on a previously withheld test data set (40%), and its performance analyzed.

RESULTS

The distribution of criticality among high, moderate, and low was 1.5%, 7.1%, and 91.4%, respectively. The one-versus-all area under the receiver operating characteristic (AUROC) curve for high and moderate criticality was 0.982 (95% CI 0.980, 0.983) and 0.968 (0.967, 0.969). The multi-class macro average AUROC and area under the receiver operating characteristic curve were 0.976 and 0.754. The features most integral to model performance included history of intravenous medications, capillary refill, emergency severity index level, history of hospitalization, use of a supplemental oxygen device, age, and history of admission to the ICU.

CONCLUSION

Pediatric ED disposition can be accurately predicted using information available at triage, providing an opportunity to improve quality of care and patient outcomes.

Development and validation of an early warning tool for sepsis and decompensation in children during emergency department triage

This study was designed to develop and validate an early warning system for sepsis based on a predictive model of critical decompensation. Data from the electronic medical records for 537,837 visits to a pediatric Emergency Department (ED) from March 2013 to December 2019 were collected. A multiclass stochastic gradient boosting model was built to identify early warning signs associated with death, severe sepsis, non-severe sepsis, and bacteremia. Model features included triage vital signs, previous diagnoses, medications, and healthcare utilizations within 6 months of the index ED visit. There were 483 patients who had severe sepsis and/or died, 1102 had non-severe sepsis, 1103 had positive bacteremia tests, and the remaining had none of the events. The most important predictors were age, heart rate, length of stay of previous hospitalizations, temperature, systolic blood pressure, and prior sepsis. The one-versus-all area under the receiver operator characteristic curve (AUROC) were 0.979 (0.967, 0.991), 0.990 (0.985, 0.995), 0.976 (0.972, 0.981), and 0.968 (0.962, 0.974) for death, severe sepsis, non-severe sepsis, and bacteremia without sepsis respectively. The multi-class macro average AUROC and area under the precision recall curve were 0.977 and 0.316 respectively. The study findings were used to develop an automated early warning decision tool for sepsis. Implementation of this model in pediatric EDs will allow sepsis-related critical decompensation to be predicted accurately after a few seconds of triage.

A High-Precision Machine Learning Algorithm to Classify Left and Right Outflow Tract Ventricular Tachycardia

Introduction

Multiple algorithms based on 12-lead ECG measurements have been proposed to identify the right ventricular outflow tract (RVOT) and left ventricular outflow tract (LVOT) locations from which ventricular tachycardia (VT) and frequent premature ventricular complex (PVC) originate. However, a clinical-grade machine learning algorithm that automatically analyzes characteristics of 12-lead ECGs and predicts RVOT or LVOT origins of VT and PVC is not currently available. The effective ablation sites of RVOT and LVOT, confirmed by a successful ablation procedure, provide evidence to create RVOT and LVOT labels for the machine learning model.

Methods

We randomly sampled training, validation, and testing data sets from 420 patients who underwent successful catheter ablation (CA) to treat VT or PVC, containing 340 (81%), 38 (9%), and 42 (10%) patients, respectively. We iteratively trained a machine learning algorithm supplied with 1,600,800 features extracted via our proprietary algorithm from 12-lead ECGs of the patients in the training cohort. The area under the curve (AUC) of the receiver operating characteristic curve was calculated from the internal validation data set to choose an optimal discretization cutoff threshold.

Results

The proposed approach attained the following performance: accuracy (ACC) of 97.62 (87.44–99.99), weighted F1-score of 98.46 (90–100), AUC of 98.99 (96.89–100), sensitivity (SE) of 96.97 (82.54–99.89), and specificity (SP) of 100 (62.97–100).

Conclusions

The proposed multistage diagnostic scheme attained clinical-grade precision of prediction for LVOT and RVOT locations of VT origin with fewer applicability restrictions than prior studies.

A High-Precision Machine Learning Algorithm to Classify Left and Right Outflow Tract Ventricular Tachycardia

Introduction

Multiple algorithms based on 12-lead ECG measurements have been proposed to identify the right ventricular outflow tract (RVOT) and left ventricular outflow tract (LVOT) locations from which ventricular tachycardia (VT) and frequent premature ventricular complex (PVC) originate. However, a clinical-grade machine learning algorithm that automatically analyzes characteristics of 12-lead ECGs and predicts RVOT or LVOT origins of VT and PVC is not currently available. The effective ablation sites of RVOT and LVOT, confirmed by a successful ablation procedure, provide evidence to create RVOT and LVOT labels for the machine learning model.

Methods

We randomly sampled training, validation, and testing data sets from 420 patients who underwent successful catheter ablation (CA) to treat VT or PVC, containing 340 (81%), 38 (9%), and 42 (10%) patients, respectively. We iteratively trained a machine learning algorithm supplied with 1,600,800 features extracted via our proprietary algorithm from 12-lead ECGs of the patients in the training cohort. The area under the curve (AUC) of the receiver operating characteristic curve was calculated from the internal validation data set to choose an optimal discretization cutoff threshold.

Results

The proposed approach attained the following performance: accuracy (ACC) of 97.62 (87.44-99.99), weighted F1-score of 98.46 (90-100), AUC of 98.99 (96.89-100), sensitivity (SE) of 96.97 (82.54-99.89), and specificity (SP) of 100 (62.97-100).

Conclusions

The proposed multistage diagnostic scheme attained clinical-grade precision of prediction for LVOT and RVOT locations of VT origin with fewer applicability restrictions than prior studies.

Multicenter study of risk factors of unplanned 30-day readmissions in pediatric oncology

Background

Pediatric oncology patients have high rates of hospital readmission but there is a dearth of research into risk factors for unplanned 30‐day readmissions among this high‐risk population.

Aim

In this study, we built a statistical model to provide insight into risk factors of unplanned readmissions in this pediatric oncology.

Methods

We retrieved 32 667 encounters from 10 418 pediatric patients with a neoplastic condition from 16 hospitals in the Cerner Health Facts Database and built a mixed‐effects model with patients nested within hospitals for inference on 75% of the data and reserved the remaining as an independent test dataset.

Results

The mixed‐effects model indicated that patients with acute lymphoid leukemia (in relapse), neuroblastoma, rhabdomyosarcoma, or bone/cartilage cancer have increased odds of readmission. The number of cancer medications taken by the patient and the administration of chemotherapy were associated with increased odds of readmission for all cancer types. Wilms Tumor had a significant interaction with administration of chemotherapy, indicating that the risk due to chemotherapy is exacerbated in patients with Wilms Tumor. A second two‐way interaction between recent history of chemotherapy treatment and infections was associated with increased odds of readmission. The area under the receiver operator characteristic curve (and corresponding 95% confidence interval) of the mixed‐effects model was 0.714 (0.702, 0.725) on the independent test dataset.

Conclusion

Readmission risk in oncology is modified by the specific type of cancer, current and past administration of chemotherapy, and increased health care utilization. Oncology‐specific models can provide decision support where model built on other or mixed population has failed.

A super learner ensemble of 14 statistical learning models for predicting COVID-19 severity among patients with cardiovascular conditions

BACKGROUND

Cardiovascular and other circulatory system diseases have been implicated in the severity of COVID-19 in adults. This study provides a super learner ensemble of models for predicting COVID-19 severity among these patients.

METHOD

The COVID-19 Dataset of the Cerner Real-World Data was used for this study. Data on adult patients (18 years or older) with cardiovascular diseases between 2017 and 2019 were retrieved and a total of 13 of these conditions were identified. Among these patients, 33,042 admitted with positive diagnoses for COVID-19 between March 2020 and June 2020 (from 59 hospitals) were identified and selected for this study. A total of 14 statistical and machine learning models were developed and combined into a more powerful super learning model for predicting COVID-19 severity on admission to the hospital.

RESULT

LASSO regression, a full extreme gradient boosting model with tree depth of 2, and a full logistic regression model were the most predictive with cross-validated AUROCs of 0.7964, 0.7961, and 0.7958 respectively. The resulting super learner ensemble model had a cross validated AUROC of 0.8006 (range: 0.7814, 0.8163). The unbiased AUROC of the super learner model on an independent test set was 0.8057 (95% CI: 0.7954, 0.8159).

CONCLUSION

Highly predictive models can be built to predict COVID-19 severity of patients with cardiovascular and other circulatory conditions. Super learning ensembles will improve individual and classical ensemble models significantly.

Agile clinical research: A data science approach to scrumban in clinical medicine

The COVID-19 pandemic has required greater minute-to-minute urgency of patient treatment in Intensive Care Units (ICUs), rendering the use of Randomized Controlled Trials (RCTs) too slow to be effective for treatment discovery. There is a need for agility in clinical research, and the use of data science to develop predictive models for patient treatment is a potential solution. However, rapidly developing predictive models in healthcare is challenging given the complexity of healthcare problems and the lack of regular interaction between data scientists and physicians. Data scientists can spend significant time working in isolation to build predictive models that may not be useful in clinical environments. We propose the use of an agile data science framework based on the Scrumban framework used in software development. Scrumban is an iterative framework, where in each iteration larger problems are broken down into simple do-able tasks for data scientists and physicians. The two sides collaborate closely in formulating clinical questions and developing and deploying predictive models into clinical settings. Physicians can provide feedback or new hypotheses given the performance of the model, and refinement of the model or clinical questions can take place in the next iteration. The rapid development of predictive models can now be achieved with increasing numbers of publicly available healthcare datasets and easily accessible cloud-based data science tools. What is truly needed are data scientist and physician partnerships ensuring close collaboration between the two sides in using these tools to develop clinically useful predictive models to meet the demands of the COVID-19 healthcare landscape.

Race, Ethnicity, and Insurance: the Association with Opioid Use in a Pediatric Hospital Setting

BACKGROUND

This study examined the association between race/ethnicity and health insurance payer type with pediatric opioid and non-opioid ordering in an inpatient hospital setting.

METHODS

Cross-sectional inpatient encounter data from June 2013 to June 2018 was retrieved from a pediatric children's hospital in Southern California (N = 55,944), and statistical analyses were performed to determine associations with opioid ordering.

RESULTS

There was a significant main effect of race/ethnicity on opioid and non-opioid orders. Physicians ordered significantly fewer opioid medications, but a greater number of non-opioid medications, for non-Hispanic African American children than non-Hispanic Asian, Hispanic/Latinx, and non-Hispanic White pediatric patients. There was also a main effect of health insurance payer type on non-opioid orders. Patients with government-sponsored plans (e.g., Medi-Cal, Medicare) received fewer non-opioid prescriptions compared with patients with both HMO and PPO coverage. Additionally, there was a significant race/ethnicity by insurance interaction on opioid orders. Non-Hispanic White patients with "other" insurance coverage received the greatest number of opioid orders. In non-Hispanic African American patients, children with PPO coverage received fewer opioids than those with government-sponsored and HMO insurance. For non-Hispanic Asian patients, children with PPO were prescribed more opioids than those with government-sponsored and HMO coverage.

CONCLUSION

Findings suggest that the relationship between race/ethnicity, insurance type, and physician decisions opioid prescribing is complex and multifaceted. Given that consistency in opioid prescribing should be seen regardless of patient background characteristics, future studies should continue to assess and monitor unequitable differences in care.

A Nested Mixed Effects Multicenter Model Examining the Risk Factors for Pediatric Trauma Return Visits Within 72 Hours

BACKGROUND

Return visits within 72 h are an important metric in evaluating the performance of emergency rooms. This has not been well studied in the pediatric trauma population. We sought to determine novel risk factors for return visits to the emergency department (ED) after trauma that may assist in identifying patients most at risk of revisit.

METHODS

We used the Cerner Health Facts Database to retrieve data from 34 EDs across the United States that care for pediatric trauma patients aged <15 y. The data consist of 610,845 patients and 816,571 ED encounters. We retrieved variables encompassing demographics, payor, current and past health care resource utilization, trauma diagnoses, other diagnoses/comorbidities, medications, and surgical procedures. We built a nested mixed effects logistic regression model to provide statistical inference on the return visits.

RESULTS

Traumas resulting from burns and corrosion, injuries to the shoulder and arms, injuries to the hip and legs, and trauma to the head and neck are all associated with increased odds of returning to the ED. Patients suffering from poisoning relating to drugs and other biological substances and patients with trauma to multiple body regions have reduced odds of returning to the ED. Longer ED length of stay and prior health care utilization (ED or inpatient) are associated with increased odds of a return visit. The sex of the patient and payor had a statistically significant effect on the risk of a return visit to the ED within 72 h of discharge.

CONCLUSIONS

Certain traumas expose patients to an increased risk for return visits to the ED and, as a result, provide opportunity for improved quality of care. Targeted interventions that include education, observation holds, or a decision to hospitalize instead of discharge home may help improve patient outcomes and decrease the rate of ED returns.

LEVEL OF EVIDENCE

III (Prognostic and Epidemiology).

A multicenter mixed-effects model for inference and prediction of 72-h return visits to the emergency department for adult patients with trauma-related diagnoses

Objective

Emergency department (ED) return visits within 72 h may be a sign of poor quality of care and entail unnecessary use of healthcare resources. In this study, we compare the performance of two leading statistical and machine learning classification algorithms, and we use the best performing approach to identify novel risk factors of ED return visits.

Methods

We analyzed 3.2 million ED encounters with at least one diagnosis under “injury, poisoning and certain other consequences of external causes” and “external causes of morbidity.” These encounters included patients 18 years or older from across 128 emergency room facilities in the USA. For each encounter, we calculated the 72-h ED return status and retrieved 57 features from demographics, diagnoses, procedures, and medications administered during the process of administration of medical care. We implemented a mixed-effects model to assess the effects of the covariates while accounting for the hierarchical structure of the data. Additionally, we investigated the predictive accuracy of the extreme gradient boosting tree ensemble approach and compared the performance of the two methods.

Results

The mixed-effects model indicates that certain blunt force and non-blunt trauma inflates the risk of a return visit. Notably, patients with trauma to the head and patients with burns and corrosions have elevated risks. This is in addition to 11 other classes of both blunt force and non-blunt force traumas. In addition, prior healthcare resource utilization, patients who have had one or more prior return visits within the last 6 months, prior ED visits, and the number of hospitalizations within the 6 months are associated with increased risk of returning to the ED after discharge. On the one hand, the area under the receiver characteristic curve (AUROC) of the mixed-effects model was 0.710 (0.707, 0.712). On the other hand, the gradient boosting tree ensemble had a lower AUROC of 0.698 CI (0.696, 0.700) on the independent test model.

Conclusions

The proposed mixed-effects model achieved the highest known AUC and resulted in the identification of novel risk factors. The model outperformed one of the leading machine learning ensemble classifiers, the extreme gradient boosting tree in terms of model performance. The risk factors we identified can assist emergency departments to decrease the number of unplanned return visits within 72 h.

HealtheDataLab - a cloud computing solution for data science and advanced analytics in healthcare with application to predicting multi-center pediatric readmissions

BACKGROUND

There is a shortage of medical informatics and data science platforms using cloud computing on electronic medical record (EMR) data, and with computing capacity for analyzing big data. We implemented, described, and applied a cloud computing solution utilizing the fast health interoperability resources (FHIR) standardization and state-of-the-art parallel distributed computing platform for advanced analytics.

METHODS

We utilized the architecture of the modern predictive analytics platform called Cerner® HealtheDataLab and described the suite of cloud computing services and Apache Projects that it relies on. We validated the platform by replicating and improving on a previous single pediatric institution study/model on readmission and developing a multi-center model of all-cause readmission for pediatric-age patients using the Cerner® Health Facts Deidentified Database (now updated and referred to as the Cerner Real World Data). We retrieved a subset of 1.4 million pediatric encounters consisting of 48 hospitals' data on pediatric encounters in the database based on a priori inclusion criteria. We built and analyzed corresponding random forest and multilayer perceptron (MLP) neural network models using HealtheDataLab.

RESULTS

Using the HealtheDataLab platform, we developed a random forest model and multi-layer perceptron model with AUC of 0.8446 (0.8444, 0.8447) and 0.8451 (0.8449, 0.8453) respectively. We showed the distribution in model performance across hospitals and identified a set of novel variables under previous resource utilization and generic medications that may be used to improve existing readmission models.

CONCLUSION

Our results suggest that high performance, elastic cloud computing infrastructures such as the platform presented here can be used for the development of highly predictive models on EMR data in a secure and robust environment. This in turn can lead to new clinical insights/discoveries.

Predictors of long ECMO runs for congenital diaphragmatic hernia

BACKGROUND

Although longer ECMO run times for patients with congenital diaphragmatic hernia (CDH) have been associated with worse outcomes, a large study has not been conducted to examine the risk factors for long ECMO runs.

METHODS

The Extracorporeal Life Support Organization (ELSO) Registry from 2000 to 2015 was used to identify predictors of long ECMO runs in CDH patients. A long run was any duration of ≥14 days. Multivariable logistic regression models were used to examine the association between demographics, pre-ECMO blood gas/ventilator settings, comorbid conditions, and therapies on long ECMO runs.

RESULTS

There were 4730 CDH-infants examined. The largest association with long ECMO runs was on-ECMO repair (OR: 3.72, 95% CI: 3.013-4.602, p < 0.001) and the use of THAM (OR: 1.463, 95% CI: 1.062-2.016, p = 0.02). Each drop in pH quartile was associated with an increased risk of long ECMO run: pH ≥ 7.3 (reference), pH 7.2-7.9 (OR 1.24, 95% CI: 0.98-1.57, p = 0.07), pH 7.08-7.19 (OR 1.46, 95% CI: 1.17-1.84, p = 0.001), pH ≤ 7.07 (OR 1.64, 95% CI: 1.29-2.07, p < 0.001).

CONCLUSIONS

We found a correlation between both pre-ECMO demographics/timing of repair and the subsequent risk of long ECMO runs, providing insight for both providers and parents about the risk factors for longer runs.

TYPE OF STUDY

Treatment Study.

LEVEL OF EVIDENCE

Level III.

Keys to improving patient satisfaction in the pediatric urology clinic: A starting point

BACKGROUND

Recent developments in healthcare have placed increased focus on patient satisfaction. Among adult populations, validated patient satisfaction tools have now been incorporated into routine practice. This process has been much slower and is significantly less standardized in pediatric populations.

OBJECTIVE

The objective of this analysis was to evaluate various factors that predict patient satisfaction in pediatric urology settings.

MATERIALS AND METHODS

Data from the National Research Corporation (NRC) Health Patient Survey were collected from children receiving urological care between 2017 and 2019. Survey data were merged with demographic and visit-related information from electronic health records (EHR). Rating of provider on the NCR Health survey was chosen as the primary outcome. Two multivariable models were analyzed by logistic regression. The first model contained only demographic and clinic-based predictor variables. The second model contained the NRC survey questions.

RESULTS

This analysis includes a total of 3232 surveys. Multivariable analysis of EHR demographic and visit data found that low income insurance (Medicaid) (OR = 1.3, p = 0.035), primary language Spanish (OR = 1.5, p = 0.012), and shorter in-office wait time (OR = 0.94, p < 0.001) were predictors for higher patient satisfaction scores. Multivariable analysis of NCR Health survey data identified physician explanation, listening, respect for patient, time spent with patient and communication between physicians and nurses as highly significant predictors of satisfaction (p < 0.001).

DISCUSSION

This analysis has identified several modifiable and non-modifiable variables that predict patient satisfaction in a population of children receiving care in a pediatric urology clinic. Limitations of this study include the possibility for response bias and lack of data on other potentially important but unmeasured factors.

CONCLUSIONS

Certain patient populations are more satisfied with the outpatient urology clinic experience. Several factors related to physician-patient communication are modifiable areas to improve patient satisfaction. Further intervention studies focusing on the modifiable areas are needed to increase patient satisfaction in pediatric urology.

The development of multiorgan dysfunction in CDH-ECMO neonates is associated with the level of pre-ECMO support

PURPOSE

Congenital diaphragmatic hernia (CDH) is the most common indication for neonatal extracorporeal membrane oxygenation (ECMO), but mortality remains at 50%. Multiorgan failure can occur in 25% and has been linked to worse outcomes. We sought to examine the factors that would increase the risk of multiorgan dysfunction (MOD).

METHODS

The Extracorporeal Life Support Organization (ELSO) database was used to identify infants with CDH (2000-2015). The primary outcome was MOD, which was defined as the presence of organ failure in ≥2 organ systems. We used a multivariable logistic regression to examine the effect of demographics, pre-ECMO respiratory status, comorbidities, and therapies on MOD.

RESULTS

There were a total of 4374 CDH infants who were treated with ECMO. Overall mortality was 52.4%. The risk models demonstrated that pre-ECMO cardiac arrest (OR 1.458, CI: 1.146-1.861, p = 0.002) and hand-bagging (OR 1.461, CI: 1.094-1.963, p = 0.032) had the strongest association with MOD. In addition, other pre-ECMO indicators of disease severity (pH, HFOV, MAP, 5-min APGAR) and pre-ECMO therapies (bicarb, neuromuscular [NM] blockers) were also associated with MOD.

CONCLUSIONS

The level of pre-ECMO support has a significant association with the development of MOD, and initiation of ECMO prior to arrest seems to be critical to avoid complications.

TYPE OF STUDY

Treatment study.

LEVEL OF EVIDENCE

Level III.

Optimal Multi-Stage Arrhythmia Classification Approach

Arrhythmia constitutes a problem with the rate or rhythm of the heartbeat, and an early diagnosis is essential for the timely inception of successful treatment. We have jointly optimized the entire multi-stage arrhythmia classification scheme based on 12-lead surface ECGs that attains the accuracy performance level of professional cardiologists. The new approach is comprised of a three-step noise reduction stage, a novel feature extraction method and an optimal classification model with finely tuned hyperparameters. We carried out an exhaustive study comparing thousands of competing classification algorithms that were trained on our proprietary, large and expertly labeled dataset consisting of 12-lead ECGs from 40,258 patients with four arrhythmia classes: atrial fibrillation, general supraventricular tachycardia, sinus bradycardia and sinus rhythm including sinus irregularity rhythm. Our results show that the optimal approach consisted of Low Band Pass filter, Robust LOESS, Non Local Means smoothing, a proprietary feature extraction method based on percentiles of the empirical distribution of ratios of interval lengths and magnitudes of peaks and valleys, and Extreme Gradient Boosting Tree classifier, achieved an F1-Score of 0.988 on patients without additional cardiac conditions. The same noise reduction and feature extraction methods combined with Gradient Boosting Tree classifier achieved an F1-Score of 0.97 on patients with additional cardiac conditions. Our method achieved the highest classification accuracy (average 10-fold cross-validation F1-Score of 0.992) using an external validation data, MIT-BIH arrhythmia database. The proposed optimal multi-stage arrhythmia classification approach can dramatically benefit automatic ECG data analysis by providing cardiologist level accuracy and robust compatibility with various ECG data sources.

A Statistical-Learning Model for Unplanned 7-Day Readmission in Pediatrics

OBJECTIVES

The rate of pediatric 7-day unplanned readmissions is often seen as a measure of quality of care, with high rates indicative of the need for improvement of quality of care. In this study, we used machine learning on electronic health records to study predictors of pediatric 7-day readmissions. We ranked predictors by clinical significance, as determined by the magnitude of the least absolute shrinkage and selection operator regression coefficients.

METHODS

Data consisting of 50 241 inpatient and observation encounters at a single tertiary pediatric hospital were retrieved; 50% of these patients' data were used for building a least absolute shrinkage and selection operator regression model, whereas the other half of the data were used for evaluating model performance. The categories of variables included were demographics, social determinants of health, severity of illness and acuity, resource use, diagnoses, medications, psychosocial factors, and other variables such as primary care no show.

RESULTS

Previous hospitalizations and readmissions, medications, multiple comorbidities, longer current and previous lengths of stay, certain diagnoses, and previous emergency department use were the most significant predictors modifying a patient's risk of 7-day pediatric readmission. The model achieved an area under the curve of 0.778 (95% confidence interval 0.763-0.793).

CONCLUSIONS

Predictors such as medications, previous and current health care resource use, history of readmissions, severity of illness and acuity, and certain psychosocial factors modified the risk of unplanned 7-day readmissions. These predictors are mostly unmodifiable, indicating that intervention plans on high-risk patients may be developed through discussions with patients and parents to identify underlying modifiable causal factors of readmissions.

A Novel Model for Enhanced Prediction and Understanding of Unplanned 30-Day Pediatric Readmission

OBJECTIVES

To develop a model to assist clinicians in reducing 30-day unplanned pediatric readmissions and to enhance understanding of risk factors leading to such readmissions.

METHODS

Data consisting of 38 143 inpatient clinical encounters at a tertiary pediatric hospital were retrieved, and 50% were used for training on a multivariate logistic regression model. The pediatric Rothman Index (pRI) was 1 of the novel candidate predictors considered. Multivariate model selection was conducted by minimization of Akaike Information Criteria. The area under the receiver operator characteristic curve (AUC) and values for sensitivity, specificity, positive predictive value, relative risk, and accuracy were computed on the remaining 50% of the data.

RESULTS

The multivariate logistic regression model of readmission consists of 7 disease diagnosis groups, 4 measures of hospital resource use, 3 measures of disease severity and/or medical complexities, and 2 variables derived from the pRI. Four of the predictors are novel, including history of previous 30-day readmissions within last 6 months (P < .001), planned admissions (P < .001), the discharge pRI score (P < .001), and indicator of whether the maximum pRI occurred during the last 24 hours of hospitalization (P = .005). An AUC of 0.79 (0.77-0.80) was obtained on the independent test data set.

CONCLUSIONS

Our model provides significant performance improvements in the prediction of unplanned 30-day pediatric readmissions with AUC higher than the LACE readmission model and other general unplanned 30-day pediatric readmission models. The model is expected to provide an opportunity to capture 39% of readmissions (at a selected operating point) and may therefore assist clinicians in reducing avoidable readmissions.

Clinical and Psychosocial Factors Associated With Patient Experience in Pediatrics

OBJECTIVE

To identify clinical and psychosocial factors associated with patient experience with care.

METHODS

We analyzed patient experience surveys, corresponding clinical and psychosocial the data of 1567 encounters using survey-weighted multivariate logistic regression analysis with willingness to recommend the facility as outcome variable.

RESULTS

Parents are less likely to recommend the facility if there were custody issues with the child during their stay, if the child has history of chronic medical condition, and if the child is male with odds ratio and corresponding 95% confidence interval of 0.38 [0.21, 0.69], 0.43 [0.24, 0.80], and 0.67 [0.45, 0.99] respectively. Parents of older patients (1-year difference) and parents of low socioeconomic status are more likely to recommend the facility (1.05 [1.01, 1.09] and 2.74 [1.72, 4.37] respectively).

CONCLUSIONS

Clinical and psychosocial factors significantly affect patient experience scores together with parent perception of provider-family communication and relationship, and hospital environment conducive for children.

The Effect of Malnutrition on the Risk of Unplanned 7-Day Readmission in Pediatrics

BACKGROUND

Malnutrition is known to be associated with higher morbidity and a risk factor of readmissions in the adult population. In this study, we explore the effect of malnutrition in pediatrics because it may differ from the adult population.

METHODS

Data for all inpatient encounters at a tertiary children's hospital within a 2-year period corresponding to 19 702 visits were obtained. The data included demographics, socioeconomic status, registered dietitian diagnosis of malnutrition, and variables of the LACE readmission model. We excluded all neonates and patients older than 21 years. A multivariable logistic model was obtained by implementing best subset regression on these variables, controlling for demographics and socioeconomic status, and considering all possible 2-way statistical interactions between malnutrition and the variables for demographics and socioeconomic status.

RESULTS

We discovered a statistical interaction effect between a patient's age and malnutrition status (P value = .002) with respect to odds of unplanned 7-day readmission. It is indicated in this interaction term that patients who were malnourished had higher odds of readmission than patients who were not malnourished. Furthermore, younger patients who were malnourished were at increased odds of readmission than their older peers, whereas among patients who were not malnourished, younger patients were at reduced odds of readmission.

CONCLUSIONS

The statistical interaction effect revealed that a patient's risk of readmission is jointly modified by the patient's age and malnutrition status. This finding advances our understanding of the complex picture of the simultaneous risk factor of unplanned 7-day readmissions in pediatrics.

Impact of anesthesiologists on parental perioperative satisfaction scores

BACKGROUND

While the focus on patient experience as an important outcome has significantly increased over the past decade, there is paucity of data on predictive factors associated with parental recommendation of a surgical facility to friends and family.

METHODS

Data for this report were obtained from a Hospital Information System and Picker Health validated surveys completed by 538 parents whose children underwent outpatient surgery from July 2014 to March 2016. Bivariate, chi-squared, and multivariate logistic regression analysis were carried out using the Picker Health item "Would you recommend this outpatient surgical facility to your friends and family?" as the primary outcome. Demographic data and 53 Picker Health items were used as potential predictors.

RESULTS

Multivariate logistic regression analysis found the following variables as independent predictors for parental recommendation: quality of perioperative communication by anesthesiologists (odds ratio [95% confidence interval]=0.23 [0.09, 0.58]); provision of information on whom to call for help after discharge (0.22 [0.07, 0.64]); child's perceived baseline health (0.37 [0.15, 0.90]); and ill-informed staff about child's procedure (0.30 [0.21, 0.79]). Variables such as child's pain and child's nausea and vomiting were not predictive for referral pattern.

CONCLUSION

Parental recommendation of a surgical facility to friends and family depends on a number of variables with the quality of perioperative communication with the anesthesiologist being the most predictive item.