Evaluation of Automated Finger Compression for Capillary Refill Time Measurement in Pediatrics.

OBJECTIVES: Early shock reversal is crucial to improve patient outcomes. Capillary refill time (CRT) is clinically important to identify and monitor shock in children but has issues with inconsistency. To minimize inconsistency, we evaluated a CRT monitoring system using an automated compression device. Our objective was to determine proper compression pressure in children. METHODS: Clinician force for CRT was collected during manual CRT measurement as a reference for automated compression in a previous study (12.9 N, 95% confidence interval, 12.5-13.4; n = 454). An automated compression device with a soft inflation bladder was fitted with a force sensor. We evaluated the effectiveness of the automated pressure to eliminate pulsatile blood flow from the distal phalange. Median and variance of CRT analysis at each pressure was compared. RESULTS: A comparison of pressures at 300 to 500 mm Hg on a simulated finger yielded a force of 5 to 10 N, and these pressures were subsequently used for automated compression for CRT. Automated compression was tested in 44 subjects (median age, 33 months; interquartile range [IQR], 14-56 months). At interim analysis of 17 subjects, there was significant difference in the waveform with residual pulsatile blood flow (9/50: 18% at 300 mm Hg, 5/50:10% at 400 mm Hg, 0/51: 0% at 500 mm Hg, P = 0.008). With subsequent enrollment of 27 subjects at 400 and 500 mm Hg, none had residual pulsatile blood flow. There was no difference in the CRT: median 1.8 (IQR, 1.06-2.875) in 400 mm Hg vs median 1.87 (IQR, 1.25-2.8325) in 500 mm Hg, P = 0.81. The variance of CRT was significantly larger in 400 mm Hg: 2.99 in 400 mm Hg vs. 1.35 in 500 mm Hg, P = 0.02, Levene's test. Intraclass correlation coefficient for automated CRT was 0.56 at 400 mm Hg and 0.78 at 500 mm Hg. CONCLUSIONS: Using clinician CRT measurement data, we determined either 400 or 500 mm Hg is an appropriate pressure for automated CRT, although 500 mm Hg demonstrates superior consistency.

Novel Uses of Traditional Algorithms for Septic Arthritis.

BACKGROUND: Pediatric musculoskeletal infection (MSKI) is a bacterial infection of the bone, joint, and/or muscle that can be difficult to diagnose. The Kocher and Caird algorithms were developed to distinguish septic arthritis (SA) from transient synovitis (TS) in the hip. These algorithms have been applied to all patients presenting with painful, swollen monoarticular joints regardless of suspicion for SA. The aim of this analysis was to assess the test performance of Kocher and Caird to distinguish MSKI among all pediatric patients presenting with monoarticular joint pain. A secondary aim was to validate the original algorithms. METHODS: We conducted a secondary analysis of a prospective cohort study evaluating the test performance of procalcitonin for suspected SA in a pediatric emergency department. Patients aged 0 to 16 years old who presented with a painful or swollen monoarticular joint were considered for enrollment. We compared the test performance of the traditional algorithms in an expanded population of MSKI versus alternate joint pain using sensitivity, specificity, and area under the curve (AUC). As a sensitivity analysis, missing data for predictors like temperature, erythrocyte sedimentation rate, C-reactive protein, and inability to bear weight were multiply imputed using the Stata program, mi impute, for changed equations. RESULTS: The Caird algorithm had better test performance compared to the Kocher in all populations. Both algorithms were most discriminative in comparing SA to TS in all joints (AUC: 0.84 Caird and 0.75 Kocher). However, the Caird criteria performed almost as well discriminating MSKI from other causes of monoarticular joint pain in all joints (AUC: 0.79; 95% confidence interval: 0.72, 0.85) and nonhip joints (AUC: 0.80; 95% confidence interval: 0.71, 0.88). CONCLUSION: Based on the findings of this study, it is clinically reasonable to apply the Caird algorithm to pediatric patients presenting with monoarticular joint pain of medium and large joints such as the knee, ankle, shoulder, elbow, and wrist, in addition to the hip. LEVEL OF EVIDENCE: Level III-retrospective study of novel applications of SA algorithms.

Comparison of Bedside and Video-Based Capillary Refill Time Assessment in Children.

OBJECTIVES: Capillary refill time (CRT) to assess peripheral perfusion in children with suspected shock may be subject to poor reproducibility. Our objectives were to compare video-based and bedside CRT assessment using a standardized protocol and evaluate interrater and intrarater consistency of video-based CRT (VB-CRT) assessment. We hypothesized that measurement errors associated with raters would be low for both standardized bedside CRT and VB-CRT as well as VB-CRT across raters. METHODS: Ninety-nine children (aged 1-12 y) had 5 consecutive bedside CRT assessments by an experienced critical care clinician following a standardized protocol. Each CRT assessment was video recorded on a black background. Thirty video clips (10 with bedside CRT < 1 s, 10 with CRT 1-2 s, and 10 with CRT > 2 s) were randomly selected and presented to 10 clinicians twice in randomized order. They were instructed to push a button when they visualized release of compression and completion of a capillary refill. The correlation and absolute difference between bedside and VB-CRT were assessed. Consistency across raters and within each rater was analyzed using the intraclass correlation coefficient (ICC). A Generalizability study was performed to evaluate sources of variation. RESULTS: We found moderate agreement between bedside and VB-CRT observations (r = 0.65; P < 0.001). The VB-CRT values were shorter by 0.17 s (95% confidence interval, 0.09-0.25; P < 0.001) on average compared with bedside CRT. There was moderate agreement in VB-CRT across raters (ICC = 0.61). Consistency of repeated VB-CRT within each rater was moderate (ICC = 0.71). Generalizability study revealed the source of largest variance was from individual patient video clips (57%), followed by interaction of the VB-CRT reviewer and patient video clip (10.7%). CONCLUSIONS: Bedside and VB-CRT observations showed moderate consistency. Using video-based assessment, moderate consistency was also observed across raters and within each rater. Further investigation to standardize and automate CRT measurement is warranted.

Discriminative Accuracy of Procalcitonin and Traditional Biomarkers in Pediatric Acute Musculoskeletal Infection.

OBJECTIVES: Septic arthritis (SA) is responsible for 20% of pediatric musculoskeletal infections (MSKI) and can have significant consequences. Early detection of SA is critical, and procalcitonin (PCT) has emerged as a promising biomarker. This study assessed the test performance of PCT and traditional biomarkers for suspected SA. METHODS: We conducted a prospective study at two pediatric emergency departments (ED). Data collected measured serum levels of C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), white blood cell (WBC) count, and PCT. Box and whisker plots were generated to compare the of the biomarkers by positive MSKI or a non-MSKI diagnosis. The diagnostic performance of biomarkers was examined using the area under the receiver operating characteristic curve (AUC), and optimal cut -points were identified using the Liu method. RESULTS: Procalcitonin performed reasonably well for detection of MSKI (AUC, 0.72; confidence interval [95% CI], 0.59-0.84). However, CRP and ESR performed better (AUC, 0.88 and 0.78, respectively). White blood cell count was not predictive of MSKI. Patients with a PCT value >0.1 ng/mL, ESR values >19.5 mm/h, and a temperature higher than 99.0°F were more than twice as likely to have acute MSKI. A high CRP level was most predictive of acute MSKI, and patients with levels >2.38 mg/dL were 3.5 times more likely to have acute MSKI. CONCLUSIONS: Procalcitonin is a potential biomarker for the clinical differential of MSKI in the pediatric ED. Additional research is warranted to establish the optimal diagnostic level for PCT, to increase sample size, and to examine any impact on cost.

Full Finger Reperfusion Time Measured by Pulse Oximeter Waveform Analysis in Children.

OBJECTIVES: Capillary refill time is a noninvasive method to assess tissue perfusion to determine shock status. Capillary refill time is defined as the time required to regain skin color after blanching pressure is applied. Although common methods to measure capillary refill time depend on clinicians' visual assessment, a new approach using a pulse oximeter waveform analysis exists, referred to as full finger reperfusion time. We aim to evaluate reproducibility and validity of the novel full finger reperfusion time measurement using clinicians' visual capillary refill time assessment as a reference standard. DESIGN: Prospective observational study. SETTING: PICUs and operating suites at a large academic children's hospital. PATIENTS: Ninety-nine children 1-12 years old with various skin color tones. INTERVENTIONS: Each child had 10 measurements, including five full finger reperfusion time and five clinician capillary refill time, alternating second and third digits. MEASUREMENTS AND MAIN RESULTS: Eighteen children had prolonged capillary refill time (> 2 s) and four children with capillary refill time greater than 3 seconds. Four-hundred eighty-five data pairs were analyzed. Intraclass correlation coefficient of full finger reperfusion time within each patient was 0.76 (95% CI, 0.68-0.83), demonstrating good reproducibility. Correlation coefficient between full finger reperfusion time and clinician capillary refill time was moderate: r = 0.37 (p < 0.0001; 95% CI, 0.29-0.44) for the pairs and r = 0.52 (p < 0.0001; 95% CI, 0.36-0.65) for patient average. Bland-Altman plot showed a consistent difference between full finger reperfusion time and clinician capillary refill time (full finger reperfusion time 1.14 s longer). Weak association was found between force and full finger reperfusion time (ß = -0.033 ± 0.016; 95% CI, -0.065 to -0.0016; p = 0.04), finger thickness (ß = -0.20 ± 0.089; 95% CI, -0.37 to -0.19; p = 0.03), except for color tone (p = 0.31). Finger temperature was associated with full finger reperfusion time (ß = -0.18 ± 0.041; 95% CI, -0.26 to -0.0999; p < 0.0001). CONCLUSIONS: Full finger reperfusion time demonstrated good reproducibility. Full finger reperfusion time showed moderate correlation with clinician capillary refill time. Full finger reperfusion time was 1.14 seconds longer than capillary refill time. Future studies should focus on the clinical value of full finger reperfusion time as a monitoring device for hemodynamics in critically ill children.

Regionalization of Isolated Pediatric Femur Fracture Treatment: Recent Trends Observed Using the Kids' Inpatient Database.

BACKGROUND: Isolated pediatric femur fractures have historically been treated at local hospitals. Pediatric referral patterns have changed in recent years, diverting patients to high volume centers. The purpose of this investigation was to assess the treatment location of isolated pediatric femur fractures and concomitant trends in length of stay and cost of treatment. METHODS: A cross-sectional analysis of surgical admissions for femoral shaft fracture was performed using the 2000 to 2012 Kids' Inpatient Database. The primary outcome was hospital location and teaching status. Secondary outcomes included the length of stay and mean hospital charges. Polytrauma patients were excluded. Data were weighted within each study year to produce national estimates. RESULTS: A total of 35,205 pediatric femoral fracture cases met the inclusion criteria. There was a significant shift in the treatment location over time. In 2000, 60.1% of fractures were treated at urban, teaching hospitals increasing to 81.8% in 2012 (P<0.001). Mean length of stay for all hospitals decreased from 2.59 to 1.91 days (P<0.001). Inflation-adjusted total charges increased during the study from $9499 in 2000 to $25,499 in 2012 per episode of treatment (P<0.001). Total charges per hospitalization were ∼$8000 greater at urban, teaching hospitals in 2012. CONCLUSIONS: Treatment of isolated pediatric femoral fractures is regionalizing to urban, teaching hospitals. Length of stay has decreased across all institutions. However, the cost of treatment is significantly greater at urban institutions relative to rural hospitals. This trend does not consider patient outcomes but the observed pattern appears to have financial implications. LEVEL OF EVIDENCE: Level III-case series, database study.

Use of a Clinical Guideline and Orderset to Reduce Hospital Admissions for Croup.

BACKGROUND: Studies have found infrequent interventions after croup admission. Our objectives were to achieve 25% reduction in (1) admission rate and (2) neck radiograph utilization among patients presenting to the emergency department. METHODS: At our tertiary children's hospital, we implemented clustered interventions including education, guideline, and orderset integration. We included patients 3 months to 8 years old with an emergency department, observation, or inpatient encounter for croup. We excluded patients with direct or ICU admissions, complex chronic conditions, or concurrent asthma, pneumonia, or bronchiolitis. We reviewed a random sample of 60% of encounters from baseline (October 1, 2017 to September 30, 2019) and implementation (October 1, 2019 to September 30, 2020) periods. We conducted a posthoc analysis from October 1, 2017 to December 1, 2021 to assess sustainment during coronavirus disease 2019. Interrupted time series analysis was used to evaluate changes in outcome, process, and balancing measures. RESULTS: There were 2906 (2123 baseline and 783 implementation) encounters included. Extrapolating preintervention trend estimates, the baseline admission rate of 8.7% decreased to 5.5% postintervention (relative decrease 37% [95% confidence interval: 8 to 66]) and sustained over 26 months after implementation. Admission rate in patients receiving 2 or fewer racemic epinephrine was significantly lower in implementation (1.7%) compared with baseline (6.3%), relative decrease of 72% (95% confidence interval: 68 to 88). There were no significant changes in neck radiographs, length of stay, or revisits. CONCLUSIONS: Croup quality improvement interventions were associated with a significant decrease in hospital admissions with no increase in revisits.

Pediatric Simulation of Intrinsic PEEP and Patient-Ventilator Trigger Asynchrony During Mechanical Ventilation.

BACKGROUND: Intrinsic PEEP during mechanical ventilation occurs when there is insufficient time for expiration to functional residual capacity before the next inspiration, resulting in air trapping. Increased expiratory resistance (RE), too rapid of a patient or ventilator breathing rate, or a longer inspiratory to expiratory time ratio (TI/TE) can all be causes of intrinsic PEEP. Intrinsic PEEP can result in increased work of breathing and patient-ventilator asynchrony (PVA) during patient-triggered breaths. We hypothesized that the difference between intrinsic PEEP and ventilator PEEP acts as an inspiratory load resulting in trigger asynchrony that needs to be overcome by increased respiratory muscle pressure (Pmus). METHODS: Using a Servo lung model (ASL 5000) and LTV 1200 ventilator in pressure control mode, we developed a passive model demonstrating how elevated RE increases intrinsic PEEP above ventilator PEEP. We also developed an active model investigating the effects of RE and intrinsic PEEP on trigger asynchrony (expressed as percentage of patient-initiated breaths that failed to trigger). We then studied if trigger asynchrony could be reduced by increased Pmus. RESULTS: Intrinsic PEEP increased significantly with increasing RE (r = 0.97, P = .006). Multivariate logistic regression analysis showed that both RE and negative Pmus levels affect trigger asynchrony (P < .001). CONCLUSIONS: A passive lung model describes the development of increasing intrinsic PEEP with increasing RE at a given ventilator breathing rate. An active lung model shows how this can lead to trigger asynchrony since the Pmus needed to trigger a breath is greater with increased RE, as the inspiratory muscles must overcome intrinsic PEEP. This model will lend itself to the study of intrinsic PEEP engendered by a higher ventilator breathing rate, as well as higher TI/TE, and will be useful in ventilator simulation scenarios of PVA. The model also suggests that increasing ventilator PEEP to match intrinsic PEEP can improve trigger asynchrony through a reduction in RE.

Demographic Risk Factors for Malnutrition in Patients With Cleft Lip and Palate.

OBJECTIVES/HYPOTHESIS: Patients with cleft lip and/or palate (CLP) are at increased risk of malnutrition. Acute and chronic malnutrition have been associated with elevated risk of postsurgical wound complications, adding morbidity and cost to patients and their families. To study the association between demographic factors, including insurance type, race, and median neighborhood income (MNI), and malnutrition in patients with CLP. STUDY DESIGN: Retrospective cohort study. METHODS: Retrospective review was performed in patients undergoing their first cleft-related surgery at a large tertiary pediatric hospital from 2006 to 2018. Demographic data, weight and height at surgery, type of insurance, race, and primary residential address were collected. Geocoded information on MNI was generated using patient address. World Health Organization Z-scores for weight-for-age (WFA) and height-for-age (HFA) were used as proxies for acute and chronic malnutrition, respectively. Linear regression models were generated to analyze the relationship of insurance type, race, and MNI on WFA and HFA Z-scores. RESULTS: About 313 patients met inclusion criteria. Increasing MNI predicted increasing WFA Z-score (0.05 increase in WFA per $1,000 increase, P = .047) as well as HFA Z-score (0.09 increase in HFA per $1,000 increase, P = .011). The effect of MNI was not independently modified by race for either WFA (P = .841) nor HFA (P = .404). Race and insurance type did not predict WFA or HFA. CONCLUSIONS: Lower MNI is a significant independent risk factor for acute and chronic malnutrition in children with CLP. Combined with previous investigation linking malnutrition to surgical outcomes in this population, this offers a target area for intervention to improve patient outcomes. LEVEL OF EVIDENCE: 3 Laryngoscope, 132:1482-1486, 2022.

Factors Associated With Bronchiolitis Guideline Nonadherence at US Children's Hospitals.

BACKGROUND: The objective with this study was to explore factors associated with nonadherence to national bronchiolitis guidelines at 52 children's hospitals. METHODS: We included patients 1 month to 2 years old with emergency department (ED) or admission encounters between January 2016 and December 2018 and bronchiolitis diagnoses in the Pediatric Health Information System database. We excluded patients with any intensive care, stay >7 days, encounters in the preceding 30 days, chronic medical conditions, croup, pneumonia, or asthma. Guideline nonadherence was defined as receiving any of 5 tests or treatments: bronchodilators, chest radiographs, systemic steroids, antibiotics, and viral testing. Nonadherence outcomes were modeled by using mixed effects logistic regression with random effects for providers and hospitals. Adjusted odds ratio (aOR) >1 indicates greater likelihood of nonadherence. RESULTS: A total of 198 028 encounters were included (141 442 ED and 56 586 admission), and nonadherence was 46.1% (ED: 40.2%, admissions: 61.0%). Nonadherence increased with patient age, with both ED and hospital providers being more likely to order tests and treatments for children 12 to 24 months compared with infants 1 ot 2 months (ED: aOR, 3.39; 95% confidence interval [CI], 3.20-3.60; admissions: aOR, 2.97; CI, 2.79-3.17]). Admitted non-Hispanic Black patients were more likely than non-Hispanic white patients to receive guideline nonadherent care (aOR, 1.16; CI, 1.10-1.23), a difference driven by higher use of steroids (aOR, 1.29; CI, 1.17-1.41) and bronchodilators (aOR, 1.39; CI, 1.31-1.48). Hospital effects were prominent for viral testing in ED and admission encounters (intraclass correlation coefficient of 0.35 and 0.32, respectively). CONCLUSIONS: Multiple factors are associated with national bronchiolitis guideline nonadherence.

Improving Care and Outcomes for Pediatric Musculoskeletal Infections.

BACKGROUND: Pediatric musculoskeletal infection (MSKI) is a common cause of hospitalization with associated morbidity. To improve the care of pediatric MSKI, our objectives were to achieve 3 specific aims within 24 months of our quality improvement (QI) interventions: (1) 50% reduction in peripherally inserted central catheter (PICC) use, (2) 25% reduction in sedations per patient, and (3) 50% reduction in empirical vancomycin administration. METHODS: We implemented 4 prospective QI interventions at our tertiary children's hospital: (1) provider education, (2) centralization of admission location, (3) coordination of radiology-orthopedic communication, and (4) implementation of an MSKI infection algorithm and order set. We included patients 6 months to 18 years of age with acute osteomyelitis, septic arthritis, or pyomyositis and excluded patients with complex chronic conditions or ICU admission. We used statistical process control charts to analyze outcomes over 2 general periods: baseline (January 2015-October 17, 2016) and implementation (October 18, 2016-April 2019). RESULTS: In total, 224 patients were included. The mean age was 6.1 years, and there were no substantive demographic or clinical differences between baseline and implementation groups. There was an 81% relative reduction in PICC use (centerline shift 54%-11%; 95% confidence interval 70-92) and 33% relative reduction in sedations per patient (centerline shift 1.8-1.2; 95% confidence interval 21-46). Empirical vancomycin use did not change (centerline 20%). CONCLUSIONS: Our multidisciplinary MSKI QI interventions were associated with a significant decrease in the use of PICCs and sedations per patient but not empirical vancomycin administration.

Performance Evaluation of Nasal Prong Interface for CPAP Delivery on a Critical Care Ventilator: A Bench Experiment.

BACKGROUND: The RAM cannula (Neotech, Valencia, CA) has become a commonly used interface for CPAP in neonatal intensive care. Performance characteristics of this interface used with a critical care ventilator are not well described. METHODS: This was a bench study utilizing a lung simulator configured as an actively breathing infant (weights of 800 g, 1.5 kg, and 3 kg) with moderate lung disease and a critical care ventilator in CPAP mode with leak compensation on. Three sizes of the RAM cannulae (preemie, newborn, and infant) were compared to 3 BabyFlow nasal prongs (Dräger Medical, Lübeck, Germany) (medium, large, and extra-large). Fabricated nasal models produced a 70% occlusive fit for the RAM cannula and an occlusive fit with the Dräger prongs. Delivered flow and pressure levels were recorded at 9 CPAP levels between 5 and 20 cm H2O. RESULTS: The Dräger prongs produced a mean airway pressure ([Formula: see text]) within 0.20 cm H2O (range -0.10 to 0.35) of the set CPAP across all evaluated prong sizes and CPAP levels. In contrast, the RAM cannula produced [Formula: see text] values that averaged 8.5 cm H2O (range -15 to -3.5) below the set CPAP levels. The deficit in delivered versus target CPAP level for the RAM cannula increased with greater set CPAP. Set CPAP of 5 cm H2O delivered [Formula: see text] values that ranged from 0.6 to 1.5 cm H2O (difference of 3.5-4.4 cm H2O). Set CPAP of 20 cm H2O delivered [Formula: see text] values that ranged from 5.0 to 8.4 cm H2O (difference of 11.7-15 cm H2O). Inspiratory flow required to achieve set CPAP levels did not differ between interfaces, suggesting high resistance in the RAM cannula device masks the delivered CPAP levels. CONCLUSIONS: Use of the RAM cannula with a 30% leak on a critical care ventilator delivered [Formula: see text] values lower than set CPAP. This may be clinically meaningful and should be considered when choosing a nasal interface.

Low Use of Vitamin A in Children Hospitalized for Measles in the United States.

Vitamin A reduces measles morbidity/mortality and and is recommended for management. We studied 142 patients hospitalized at US Children's hospitals for measles between January 1, 2004, and March 31, 2019, and found only 47 (33%) received vitamin A. Patients with complex chronic conditions were less likely to be treated. This study highlights a concerning gap between recommendations and practice for hospital management of measles.

Objective Measures of Vest Therapy Adherence Among Pediatric Subjects With Cystic Fibrosis.

BACKGROUND: Airway clearance techniques (ACTs), including high-frequency chest wall oscillation (vest therapy), are important for maintaining lung function for patients with cystic fibrosis, but daily completion of ACTs is time-consuming and cumbersome. Patient adherence is a persistent challenge, and adherence data are largely patient reported, which may reduce accuracy. To provide definitive adherence data, this study utilized a Bluetooth-enabled vest therapy system to remotely collect objective adherence data from a cohort of pediatric subjects. METHODS: Vest usage data were collected over a 12-month period and compared to each subject vest prescription. Because vest prescriptions are multifaceted, we developed metrics to examine adherence to the various facets (eg, daily treatment, treatments per day, treatment duration, frequency settings, and frequency and pressure settings combined) as well as an overall adherence metric. RESULTS: Among the 73 enrolled subjects, 62 (50% male, age range 2-19 y) completed the study. Among adolescent subjects age 13-19 y, average adherence to daily vest therapy (44.0%) was significantly lower than that among subjects 0-6 y old (77.8%) and subjects 7-12 y old (89.5%). As more prescribed therapy components were combined into a single adherence metric, all age groups declined in mean adherence rates, with overall adherence a decreasing function of age. CONCLUSIONS: Employing new technology to remotely collect vest usage data allows for a granular examination of vest therapy adherence. While maintaining high levels of treatment adherence becomes increasingly difficult as children age, we also found substantial reductions in adherence rates among all age groups when more complex aspects of therapy prescriptions, such as frequency and pressure settings, were examined. These data illustrate areas providers and care teams can focus on to improve patient adherence to vest prescriptions.

Effects of proactive and rescue enteral tube feedings on weight change in children undergoing treatment for high-grade CNS tumors.

BACKGROUND: Children with high-grade CNS cancers frequently experience malnutrition during treatment. We assessed the effects of proactive enteral tube (ET) placement/enteral tube feedings (ETF) on weight in infants/children with high-grade CNS tumors treated with aggressive chemotherapy. METHODS: We conducted a retrospective study of patients age 0 to 19 years treated for new high-grade CNS tumors between 2002 and 2017 at a tertiary pediatric hospital system. Patients underwent placement of proactive ET (≤ 31 days postdiagnosis; n = 45), rescue ET (> 31 days, due to weight loss; n = 9), or no ET (n = 18). Most received surgically placed ET (98%), with percutaneous endoscopic gastrojejunostomy or gastrojejunostomy tubes favored to allow jejunal feeding. The majority of patients with ET used ETF (91%). Using mixed-effects regression models, we examined differences in mean weights between ET/ETF groups across the first year of treatment. We also evaluated observed weight changes. RESULTS: All infants (n = 22, median age, 1.5 years) had proactive ET placed and 21 of 22 used proactive ETF. Infants showed an initial increase in mean percentage weight change that eventually leveled off, for an estimated increase of 10.4% over the year. For the pediatric cohort (n = 50, median, 8.1 years), those receiving proactive ETF experienced weight increases (+9.9%), those with rescue ETF experienced an initial decline and eventually rebounded for no net change (0.0%), and those with no ETF demonstrated an initial decline that persisted (-11.9%; P interaction < .001). Analysis of observed weights revealed nearly identical patterns. CONCLUSIONS: Proactive ETF was effective at maintaining weight and/or facilitating weight gain over the first year of treatment and was acceptable to patients/families.

Supervised Machine Learning Applied to Automate Flash and Prolonged Capillary Refill Detection by Pulse Oximetry.

OBJECTIVE: Develop an automated approach to detect flash (<1.0 s) or prolonged (>2.0 s) capillary refill time (CRT) that correlates with clinician judgment by applying several supervised machine learning (ML) techniques to pulse oximeter plethysmography data. MATERIALS AND METHODS: Data was collected in the Pediatric Intensive Care Unit (ICU), Cardiac ICU, Progressive Care Unit, and Operating Suites in a large academic children's hospital. Ninety-nine children and 30 adults were enrolled in testing and validation cohorts, respectively. Patients had 5 paired CRT measurements by a modified pulse oximeter device and a clinician, generating 485 waveform pairs for model training. Supervised ML models using gradient boosting (XGBoost), logistic regression (LR), and support vector machines (SVMs) were developed to detect flash (<1 s) or prolonged CRT (≥2 s) using clinician CRT assessment as the reference standard. Models were compared using Area Under the Receiver Operating Curve (AUC) and precision-recall curve (positive predictive value vs. sensitivity) analysis. The best performing model was externally validated with 90 measurement pairs from adult patients. Feature importance analysis was performed to identify key waveform characteristics. RESULTS: For flash CRT, XGBoost had a greater mean AUC (0.79, 95% CI 0.75-0.83) than logistic regression (0.77, 0.71-0.82) and SVM (0.72, 0.67-0.76) models. For prolonged CRT, XGBoost had a greater mean AUC (0.77, 0.72-0.82) than logistic regression (0.73, 0.68-0.78) and SVM (0.75, 0.70-0.79) models. Pairwise testing showed statistically significant improved performance comparing XGBoost and SVM; all other pairwise model comparisons did not reach statistical significance. XGBoost showed good external validation with AUC of 0.88. Feature importance analysis of XGBoost identified distinct key waveform characteristics for flash and prolonged CRT, respectively. CONCLUSION: Novel application of supervised ML to pulse oximeter waveforms yielded multiple effective models to identify flash and prolonged CRT, using clinician judgment as the reference standard. TWEET: Supervised machine learning applied to pulse oximeter waveform features predicts flash or prolonged capillary refill.

Demographics, Complications and Resource Utilization for Patients Hospitalized for Measles Across Us Children's Hospitals.

Our objective was to retrospectively describe measles hospitalizations in 52 US children's hospitals. We identified 136 patients hospitalized for measles in 2004-2018; 17% (23/136) had complex chronic conditions, 2 of whom died or were in hospice. Among noncomplex patients only 39% received vitamin A, median length of stay was 3 days and median adjusted estimated costs were $5896.

Identifying Patients With Kawasaki Disease Safe for Early Discharge: Development of a Risk Prediction Model at a US Children's Hospital.

OBJECTIVES: To develop a model to predict risk of intravenous immunoglobulin (IVIg) nonresponse in patients with Kawasaki disease (KD) to assist in early discharge decision-making. METHODS: Retrospective cohort study of 430 patients 0 to 18 years old discharged from a US children's hospital January 1, 2010, through July 31, 2017 with a diagnosis of KD. IVIg nonresponse was defined as at least 1 of the following: temperature ≥38.0°C between 36 hours and 7 days after initial IVIg dose, receipt of a second IVIg dose after a temperature ≥38.0°C at least 20 hours after initial IVIg dose, or readmission within 7 days with administration of a second IVIg dose. Backward stepwise logistic regression was used to select a predictive model. RESULTS: IVIg nonresponse occurred in 19% (81 of 430) of patients. We identified a multivariate model (which included white blood cell count, hemoglobin level, platelet count, aspartate aminotransferase level, sodium level, albumin level, temperature within 6 hours of first IVIg dose, and incomplete KD) with good predictive ability (optimism-adjusted concordance index: 0.700) for IVIg nonresponse. Stratifying into 2 groups by a predictive probability cutoff of 0.10, we identified 26% of patients at low risk for IVIg nonresponse, with a sensitivity and specificity of 90% and 30%, respectively, and a negative predictive value of 93%. CONCLUSIONS: We developed a model with good predictive value for identifying risk of IVIg nonresponse in patients with KD at a US children's hospital. Patients at lower risk may be considered for early discharge by using shared decision-making. Our model may be used to inform implementation of electronic health record tools and future risk prediction research.

Vaccine-preventable, hospitalizations among American Indian/Alaska Native children using the 2012 Kid's Inpatient Database.

Our aim was to assess the odds of hospitalization for a vaccine-preventable, infectious disease (VP-ID) in American Indian/Alaska Native (AI/AN) children compared to other racial and ethnic groups using the 2012 Kid's Inpatient Database (KID) The KID is a nationally representative sample, which allows for evaluation of VP-ID in a non-federal, non-Indian Health Service setting. In a cross-sectional analysis, we evaluated the association of race/ethnicity and a composite outcome of hospitalization due to vaccine-preventable infection using multivariate logistic regression. AI/AN children were more likely (OR=1.81, 95% CI=1.34, 2.45) to be admitted to the hospital in 2012 for a VP-ID compared to Non-Hispanic white children after adjusting for age, sex, chronic disease status, metropolitan location, and median household income. This disparity highlights the necessity for a more comprehensive understanding of immunization and infectious disease exposure among American Indian children, especially those not covered or evaluated by Indian Health Service.