The Pediatric Data Science and Analytics Subgroup of the Pediatric Acute Lung Injury and Sepsis Investigators Network: Use of Supervised Machine Learning Applications in Pediatric Critical Care Medicine Research

OBJECTIVE

Perform a scoping review of supervised machine learning in pediatric critical care to identify published applications, methodologies, and implementation frequency to inform best practices for the development, validation, and reporting of predictive models in pediatric critical care.

DESIGN

Scoping review and expert opinion.

SETTING

We queried CINAHL Plus with Full Text (EBSCO), Cochrane Library (Wiley), Embase (Elsevier), Ovid Medline, and PubMed for articles published between 2000 and 2022 related to machine learning concepts and pediatric critical illness. Articles were excluded if the majority of patients were adults or neonates, if unsupervised machine learning was the primary methodology, or if information related to the development, validation, and/or implementation of the model was not reported. Article selection and data extraction were performed using dual review in the Covidence tool, with discrepancies resolved by consensus.

SUBJECTS

Articles reporting on the development, validation, or implementation of supervised machine learning models in the field of pediatric critical care medicine.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

Of 5075 identified studies, 141 articles were included. Studies were primarily (57%) performed at a single site. The majority took place in the United States (70%). Most were retrospective observational cohort studies. More than three-quarters of the articles were published between 2018 and 2022. The most common algorithms included logistic regression and random forest. Predicted events were most commonly death, transfer to ICU, and sepsis. Only 14% of articles reported external validation, and only a single model was implemented at publication. Reporting of validation methods, performance assessments, and implementation varied widely. Follow-up with authors suggests that implementation remains uncommon after model publication.

CONCLUSIONS

Publication of supervised machine learning models to address clinical challenges in pediatric critical care medicine has increased dramatically in the last 5 years. While these approaches have the potential to benefit children with critical illness, the literature demonstrates incomplete reporting, absence of external validation, and infrequent clinical implementation.

Association of Early Postoperative Regional Oxygen Saturation Measures and Development of Necrotizing Enterocolitis in Neonates Following Cardiac Surgery

Necrotizing enterocolitis (NEC) is a relatively common complication in neonates with single ventricle physiology following heart surgery. Near-infrared spectroscopy (NIRS) is used to measure regional oxygen saturations in neonates in the postoperative period. We sought to investigate the association of somatic regional oxygen saturation (srSO2) and cerebral regional oxygen saturation (crSO2) in the early postoperative period and the subsequent development of NEC. We performed a retrospective cohort study of neonates who underwent cardiac surgery with cardiopulmonary bypass from October 2017 to September 2021 at the University of Virginia Children's Hospital. Values of srSO2 and crSO2 were captured over the first 48 h following surgery. 166 neonates were included and the median age at time of surgery was 8 days. NEC was diagnosed in 18 neonates following heart surgery with a median interval from surgery to diagnosis of 7 days. Neonates with single ventricle physiology had lower average crSO2 (62% vs 78%, p < 0.001), average srSO2 (72% vs 86%, p < 0.001), average crSO2 to srSO2 ratio (0.874 vs 0.913, p < 0.001), and an increased average srSO2-crSO2 difference (10% vs 8%, p = 0.03). Adjusting for single ventricle physiology, lower average crSO2 was associated with the development of definite NEC (modified Bell's criteria stage IIa and higher) (OR = 0.86, 95% CI 0.78-0.96, p = 0.007). Lower crSO2 values in the early postoperative period in neonates following cardiac surgery was associated with an increased risk in the subsequent development of NEC.

The Impact of Nasal Intubation on Feeding Outcomes in Neonates Requiring Cardiac Surgery: A Randomized Control Trial

Neonates who require surgery for congenital heart disease (CHD) frequently have difficulty with oral feeds post-operatively and may require a feeding tube at hospital discharge. The purpose of this study was to determine the effect of oral or nasal intubation route on feeding method at hospital discharge. This was a non-blinded randomized control trial of 62 neonates who underwent surgery for CHD between 2018 and 2021. Infants in the nasal (25 patients) and oral (37 patients) groups were similar in terms of pre-operative risk factors for feeding difficulties including completed weeks of gestational age at birth (39 vs 38 weeks), birthweight (3530 vs 3100 g), pre-operative PO intake (92% vs 81%), and rate of pre-operative intubation (22% vs 28%). Surgical risk factors were also similar including Society of Thoracic Surgeons-European Association for Cardio-Thoracic Surgery category (3.9 vs 4.1), shunt placement (32% vs 41%), cardiopulmonary bypass time (181 vs 177 min), and cross-clamp time (111 vs 105 min). 96% of nasally intubated patients took full oral feeds by discharge as compared with 78% of orally intubated infants (p = 0.05). Nasally intubated infants reach full oral feeds an average of 3 days earlier than their orally intubated peers. In this cohort of patients, nasally intubated infants reach oral feeds more quickly and are less likely to require supplemental tube feeding in comparison to orally intubated peers. Intubation route is a potential modifiable risk factor for oral aversion and appears safe in neonates. The study was approved by the University of Virginia Institutional Review Board for Health Sciences Research and was retrospectively registered on clinicaltrials.gov (NCT05378685) on May 18, 2022.

Changes in Cerebral Regional Oxygen Saturation Variability in Neonates Undergoing Cardiac Surgery: A Prospective Cohort Study

Near-infrared spectroscopy is routinely used in the monitoring of cerebral regional oxygen saturation (crSO2) in neonates following congenital heart surgery. Decreased postoperative crSO2 variability in these patients is associated with worse clinical outcomes, including neurodevelopmental outcomes. We sought to explore changes in crSO2 variability between the preoperative and postoperative periods and associations with short-term clinical outcomes in neonates undergoing cardiac surgery. We performed a prospective cohort study of neonates undergoing cardiac surgery with cardiopulmonary bypass between November 2019 and May 2021. We calculated crSO2 variability using averaged 1 min of crSO2 values for a minimum of 12 h before, and the first 48 h following surgery. 37 neonates (median age at start of monitoring 4 days (interquartile range 2-5 days)) were included in our study. We observed a 30% decrease in crSO2 variability between the preoperative and postoperative monitoring periods (p < 0.001). Preoperative crSO2 variability increased by 9% (p = 0.009) for each additional postnatal day. There were no associations between the degree of decrease in crSO2 variability postoperatively and class of cardiac lesion (e.g., aortic arch obstruction, single ventricle physiology) or short-term postoperative clinical outcomes. There was a significant decrease in postoperative crSO2 variability following neonatal cardiac surgery as compared to the preoperative period, likely influenced by several factors. The impact of interventions on crSO2 variability and resultant influence on long-term outcomes, such as neurodevelopmental outcomes, requires further exploration.

Gastrostomy tube placement in congenital cardiac surgery: a multi-institutional database study

INTRODUCTION

Neonates and infants who undergo congenital cardiac surgery frequently have difficulty with feeding. The factors that predispose these patients to require a gastrostomy tube have not been well defined. We aimed to report the incidence and describe hospital outcomes and characteristics in neonates and infants undergoing congenital cardiac surgery who required gastrostomy tube placement.

MATERIALS AND METHOD

A retrospective review was performed on patients undergoing congenital cardiac surgery between October 2015 and December 2020. Patients were identified by International Classification of Diseases 10th Revision codes, utilising the performance improvement database Vizient® Clinical Data Base, and stratified by age at admission: neonates (<1 month) and infants (1-12 months). Outcomes were compared and comparative analysis performed between admissions with and without gastrostomy tube placement.

RESULTS

There were 11,793 admissions, 3519 (29.8%) neonates and 8274 (70.2%) infants. We found an increased incidence of gastrostomy tube placement in neonates as compared to infants following congenital cardiac surgery (23.1% versus 6%, p = <0.001). Outcomes in neonates and infants were similar with increased length of stay and cost in those requiring a gastrostomy tube. Gastrostomy tube placement was noted to be more likely in neonates and infants with upper airway anomalies, congenital abnormalities, hospital infections, and genetic abnormalities.

DISCUSSION

Age at hospitalisation for congenital cardiac surgery is a definable risk factor for gastrostomy tube requirement. Additional factors contribute to gastrostomy tube placement and should be used when counselling families regarding the potential requirement of a gastrostomy tube.

Implementation of a Regional Oxygen Saturation Thought Algorithm and Association with Clinical Outcomes in Pediatric Patients Following Cardiac Surgery

Near infrared spectroscopy is routinely used in the noninvasive monitoring of cerebral and somatic regional oxygen saturation (rSO₂) in pediatric patients following surgery for congenital heart disease. We sought to evaluate the association of a bedside rSO₂ thought algorithm with clinical outcomes in a cohort of pediatric patients following cardiac surgery. This was a single-center retrospective cohort study of patients admitted following cardiac surgery over a 42-month period. The intervention was the implementation of an rSO₂ thought algorithm, the primary goal of which was to supply bedside providers with a thought aide to help identify, and guide response to, changes in rSO₂ in post-operative cardiac surgical patients. Surgical cases were stratified into two 18-month periods of observation, pre- and post-intervention allowing for a 6-month washout period during implementation of the thought algorithm. Clinical outcomes were compared between pre- and post-intervention periods. There were 434 surgical cases during the period of study. We observed a 27% relative risk reduction in our standardized mortality rate (0.61 to 0.48, p = 0.01) between the pre- and post-intervention periods. We did not observe differences in other post-operative clinical outcomes such as ventilator free days or post-operative ICU length of stay. Providing frontline clinical staff with education and tools, such as a bedside rSO₂ thought algorithm, may aide in the earlier detection of imbalance between oxygen delivery and consumption and may contribute to improved patient outcomes.

Signatures of illness in children requiring unplanned intubation in the pediatric intensive care unit: A retrospective cohort machine-learning study

Acute respiratory failure requiring the initiation of invasive mechanical ventilation remains commonplace in the pediatric intensive care unit (PICU). Early recognition of patients at risk for respiratory failure may provide clinicians with the opportunity to intervene and potentially improve outcomes. Through the development of a random forest model to identify patients at risk for requiring unplanned intubation, we tested the hypothesis that subtle signatures of illness are present in physiological and biochemical time series of PICU patients in the early stages of respiratory decompensation. We included 116 unplanned intubation events as recorded in the National Emergency Airway Registry for Children in 92 PICU admissions over a 29-month period at our institution. We observed that children have a physiologic signature of illness preceding unplanned intubation in the PICU. Generally, it comprises younger age, and abnormalities in electrolyte, hematologic and vital sign parameters. Additionally, given the heterogeneity of the PICU patient population, we found differences in the presentation among the major patient groups - medical, cardiac surgical, and non-cardiac surgical. At four hours prior to the event, our random forest model demonstrated an area under the receiver operating characteristic curve of 0.766 (0.738 for medical, 0.755 for cardiac surgical, and 0.797 for non-cardiac surgical patients). The multivariable statistical models that captured the physiological and biochemical dynamics leading up to the event of urgent unplanned intubation in a PICU can be repurposed for bedside risk prediction.

Individual illness dynamics: An analysis of children with sepsis admitted to the pediatric intensive care unit

Illness dynamics and patterns of recovery may be essential features in understanding the critical illness course. We propose a method to characterize individual illness dynamics in patients who experienced sepsis in the pediatric intensive care unit. We defined illness states based on illness severity scores generated from a multi-variable prediction model. For each patient, we calculated transition probabilities to characterize movement among illness states. We calculated the Shannon entropy of the transition probabilities. Using the entropy parameter, we determined phenotypes of illness dynamics based on hierarchical clustering. We also examined the association between individual entropy scores and a composite variable of negative outcomes. Entropy-based clustering identified four illness dynamic phenotypes in a cohort of 164 intensive care unit admissions where at least one sepsis event occurred. Compared to the low-risk phenotype, the high-risk phenotype was defined by the highest entropy values and had the most ill patients as defined by a composite variable of negative outcomes. Entropy was significantly associated with the negative outcome composite variable in a regression analysis. Information-theoretical approaches to characterize illness trajectories offer a novel way of assessing the complexity of a course of illness. Characterizing illness dynamics with entropy offers additional information in conjunction with static assessments of illness severity. Additional attention is needed to test and incorporate novel measures representing the dynamics of illness.

Pediatric Organ Dysfunction Information Update Mandate (PODIUM) Contemporary Organ Dysfunction Criteria: Executive Summary

Prior criteria for organ dysfunction in critically ill children were based mainly on expert opinion. We convened the Pediatric Organ Dysfunction Information Update Mandate (PODIUM) expert panel to summarize data characterizing single and multiple organ dysfunction and to derive contemporary criteria for pediatric organ dysfunction. The panel was composed of 88 members representing 47 institutions and 7 countries. We conducted systematic reviews of the literature to derive evidence-based criteria for single organ dysfunction for neurologic, cardiovascular, respiratory, gastrointestinal, acute liver, renal, hematologic, coagulation, endocrine, endothelial, and immune system dysfunction. We searched PubMed and Embase from January 1992 to January 2020. Study identification was accomplished using a combination of medical subject headings terms and keywords related to concepts of pediatric organ dysfunction. Electronic searches were performed by medical librarians. Studies were eligible for inclusion if the authors reported original data collected in critically ill children; evaluated performance characteristics of scoring tools or clinical assessments for organ dysfunction; and assessed a patient-centered, clinically meaningful outcome. Data were abstracted from each included study into an electronic data extraction form. Risk of bias was assessed using the Quality in Prognosis Studies tool. Consensus was achieved for a final set of 43 criteria for pediatric organ dysfunction through iterative voting and discussion. Although the PODIUM criteria for organ dysfunction were limited by available evidence and will require validation, they provide a contemporary foundation for researchers to identify and study single and multiple organ dysfunction in critically ill children.

Patterns of Organ Dysfunction in Critically Ill Children Based on PODIUM Criteria

OBJECTIVES

The goal of this study was to determine the incidence, prognostic performance, and generalizability of the Pediatric Organ Dysfunction Information Update Mandate (PODIUM) organ dysfunction criteria using electronic health record (EHR) data. Additionally, we sought to compare the performance of the PODIUM criteria with the organ dysfunction criteria proposed by the 2005 International Pediatric Sepsis Consensus Conference (IPSCC).

METHODS

Retrospective observational cohort study of critically ill children at 2 medical centers in the United States between 2010 and 2018. We assessed prevalence of organ dysfunction based on the PODIUM and IPSCC criteria for each 24-hour period from admission to 28 days. We studied the prognostic performance of the criteria to discriminate in-hospital mortality.

RESULTS

Overall, 22 427 PICU admissions met inclusion criteria, and in-hospital mortality was 2.3%. The cumulative incidence of each PODIUM organ dysfunction ranged from 15% to 30%, with an in-hospital mortality of 6% to 10% for most organ systems. The number of concurrent PODIUM organ dysfunctions demonstrated good-to-excellent discrimination for in-hospital mortality (area under the curve 0.87-0.93 for day 1 through 28) and compared favorably to the IPSCC criteria (area under the curve 0.84-0.92, P < .001 to P = .06).

CONCLUSIONS

We present the first evaluation of the PODIUM organ dysfunction criteria in 2 EHR databases. The use of the PODIUM organ dysfunction criteria appears promising for epidemiologic and clinical research studies using EHR data. More studies are needed to evaluate the PODIUM criteria that are not routinely collected in structured format in EHR databases.

iciHHV-6 in a Patient With Multisystem Inflammatory Syndrome in Children

Multisystem inflammatory syndrome in children (MIS-C) is a serious, sometimes life-threatening late complication of coronavirus disease 2019 (COVID-19) with multiorgan involvement and evidence of immune activation. The pathogenesis of MIS-C is not known, nor is the pathogenesis of the severe organ damage that is the hallmark of MIS-C. Human herpesvirus 6 (HHV-6), the virus responsible for roseola, is a ubiquitous herpesvirus that causes close to universal infection by the age of 3 years. HHV-6 remains latent for life and can be activated during inflammatory states, by other viruses, and by host cell apoptosis. HHV-6 has been associated with end-organ diseases, including hepatitis, carditis, and encephalitis. In addition, ∼1% of people have inherited chromosomally integrated human herpesvirus 6 (iciHHV-6), which is HHV-6 that has been integrated into chromosomal telomeric regions and is transmitted through the germ line. iciHHV-6 can be reactivated and has been associated with altered immune responses. We report here a case of MIS-C in which an initial high HHV-6 DNA polymerase chain reaction viral load assay prompted testing for iciHHV-6, which yielded a positive result. Additional research may be warranted to determine if iciHHV-6 is commonly observed in patients with MIS-C and, if so, whether it may play a part in MIS-C pathogenesis.

Blood pressure ranges via non-invasive and invasive monitoring techniques in premature neonates using high resolution physiologic data

BACKGROUND

There are limited evidence-based published blood pressure ranges for premature neonates. The aim of the study was to determine blood pressure ranges in a large cohort of premature neonates based on gestational and post-menstrual age.

METHODS

Retrospective observational study of premature neonates admitted to the neonatal intensive care unit at our institution between January 2009 and October 2015. We stratified data by gestational and post-menstrual age groups as well as by method of blood pressure measurement (non-invasive vs. invasive).

RESULTS

Over two billion blood pressure values in 1708 neonates were analyzed to generate heat maps and establish percentile-based reference ranges. The median gestational age of the cohort was 31 weeks (IQR 28-33 weeks). We found moderate correlation (r = 0.57) between simultaneously obtained non-invasive and invasive blood pressure measurements.

CONCLUSIONS

Our results can serve as a reference during the bedside assessment of the critically-ill neonate.

Taxonomy of Pathogen Codetection in Pediatric Case Fatalities with Adenoviral Respiratory Infection

Objective The aim of this study was to determine the prevalence and taxonomy of pathogen codetection in pediatric case fatalities associated with adenoviral respiratory infection.

Methods This retrospective case series included 107 pediatric case fatalities associated with adenoviral respiratory infection.

Results We observed a high rate of pathogen codetection with broad diversity from both respiratory and nonrespiratory sources. We noted differences in codetection characteristics based on immune status; most notably that immunocompromised cases were more likely to have bacteremia and adenoviremia.

Conclusions In pediatric case fatalities associated with adenoviral respiratory infection, we observed a high degree of pathogen codetection.

Hospital Survival After Surgical Repair of Truncus Arteriosus with Interrupted Aortic Arch: Results from a Multi-institutional Database

Truncus arteriosus (TA) is a major congenital cardiac malformation that requires surgical repair in the first few weeks of life. Interrupted aortic arch (IAA) is an associated malformation that significantly impacts the complexity of the TA operation. The aim of this study was to (1) define the comorbid conditions associated with TA and (2) determine the hospital survival and morbidity of patients with TA with and without an IAA. Data was collected from the Vizient Clinical Database/Resource Manager, formerly University HealthSystem Consortium, which encompasses more than 160 academic medical centers in the United States. The database was queried for patients admitted from 2002 to 2016 who were ≤ 4 months of age at initial admission, diagnosed with TA, and underwent complete surgical repair during that hospitalization. Of the 645 patients with TA who underwent surgery, 98 (15%) had TA with an interrupted aortic arch (TA-IAA). Both TA and TA-IAA were associated with a high prevalence of comorbidities, including DiGeorge syndrome, prematurity, and other congenital malformations. There was no difference in mortality between TA and TA-IAA (13.7-18.4%, p value = 0.227). No comorbid conditions were associated with an increased mortality in either group. However, patients with TA-IAA had a longer post-operative length of stay (LOS) compared to those without IAA (30 versus 40.3 days, p value = 0.001) and this effect was additive with each additional comorbid condition. In conclusion, the addition of IAA to TA is associated with an increased post-operative LOS, but does not increase in-hospital mortality.

Accuracy and Monitoring of Pediatric Early Warning Score (PEWS) Scores Prior to Emergent Pediatric Intensive Care Unit (ICU) Transfer: Retrospective Analysis

BACKGROUND

Current approaches to early detection of clinical deterioration in children have relied on intermittent track-and-trigger warning scores such as the Pediatric Early Warning Score (PEWS) that rely on periodic assessment and vital sign entry. There are limited data on the utility of these scores prior to events of decompensation leading to pediatric intensive care unit (PICU) transfer.

OBJECTIVE

The purpose of our study was to determine the accuracy of recorded PEWS scores, assess clinical reasons for transfer, and describe the monitoring practices prior to PICU transfer involving acute decompensation.

METHODS

We conducted a retrospective cohort study of patients ≤21 years of age transferred emergently from the acute care pediatric floor to the PICU due to clinical deterioration over an 8-year period. Clinical charts were abstracted to (1) determine the clinical reason for transfer, (2) quantify the frequency of physiological monitoring prior to transfer, and (3) assess the timing and accuracy of the PEWS scores 24 hours prior to transfer.

RESULTS

During the 8-year period, 72 children and adolescents had an emergent PICU transfer due to clinical deterioration, most often due to acute respiratory distress. Only 35% (25/72) of the sample was on continuous telemetry or pulse oximetry monitoring prior to the transfer event, and 47% (34/72) had at least one incorrectly documented PEWS score in the 24 hours prior to the event, with a score underreporting the actual severity of illness.

CONCLUSIONS

This analysis provides support for the routine assessment of clinical deterioration and advocates for more research focused on the use and utility of continuous cardiorespiratory monitoring for patients at risk for emergent transfer.

Dynamic Transitions of Pediatric Sepsis: A Markov Chain Analysis

Pediatric sepsis is a heterogeneous disease with varying physiological dynamics associated with recovery, disability, and mortality. Using risk scores generated from a sepsis prediction model to define illness states, we used Markov chain modeling to describe disease dynamics over time by describing how children transition among illness states. We analyzed 18,666 illness state transitions over 157 pediatric intensive care unit admissions in the 3 days following blood cultures for suspected sepsis. We used Shannon entropy to quantify the differences in transition matrices stratified by clinical characteristics. The population-based transition matrix based on the sepsis illness severity scores in the days following a sepsis diagnosis can describe a sepsis illness trajectory. Using the entropy based on Markov chain transition matrices, we found a different structure of dynamic transitions based on ventilator use but not age group. Stochastic modeling of transitions in sepsis illness severity scores can be useful in describing the variation in transitions made by patient and clinical characteristics.

Adenoviral Respiratory Infection-Associated Mortality in Children: A Retrospective Case Series

Viral respiratory infections are a leading cause of illness and hospitalization in young children worldwide. Case fatality rates in pediatric patients with adenoviral lower respiratory tract infection requiring intensive care unit (ICU) admission have been reported between 7 and 22%. We investigated the demographics and clinical characteristics in pediatric mortalities associated with adenoviral respiratory infection at 12 academic children's hospitals in the United States. There were 107 mortality cases included in our study, 73% of which had a chronic medical condition. The most common chronic medical condition was immunocompromised state in 37 cases (35%). The incidences of pediatric acute respiratory distress syndrome (78%) and multiple organ dysfunction syndrome (94%) were profound. Immunocompetent cases were more likely to receive mechanical ventilation within the first hour of ICU admission (60 vs. 14%, p < 0.001) and extracorporeal membrane oxygenation (27 vs. 5%, p = 0.009), and less likely to receive continuous renal replacement therapy (20 vs. 49%, p = 0.002) or have renal dysfunction (54 vs. 78%, p = 0.014) as compared with immunocompromised cases. Immunocompromised cases were more likely to have bacteremia (57 vs. 16%, p < 0.001) and adenoviremia (51 vs. 17%, p < 0.001) and be treated with antiviral medications (81 vs. 26%, p < 0.001). We observed a high burden of nonrespiratory organ system dysfunction in a cohort of pediatric case fatalities with adenoviral respiratory infection. The majority of cases had a chronic medical condition associated with an increased risk of complications from viral respiratory illness, most notably immunocompromised state. Important treatment differences were noted between immunocompromised and immunocompetent cases.

Predictive analytics in the pediatric intensive care unit for early identification of sepsis: capturing the context of age

BACKGROUND

Early recognition of patients at risk for sepsis is paramount to improve clinical outcomes. We hypothesized that subtle signatures of illness are present in physiological and biochemical time series of pediatric-intensive care unit (PICU) patients in the early stages of sepsis.

METHODS

We developed multivariate models in a retrospective observational cohort to predict the clinical diagnosis of sepsis in children. We focused on age as a predictor and asked whether random forest models, with their potential for multiple cut points, had better performance than logistic regression.

RESULTS

One thousand seven hundred and eleven admissions for 1425 patients admitted to a mixed cardiac and medical/surgical PICU were included. We identified, through individual chart review, 187 sepsis diagnoses that were not within 14 days of a prior sepsis diagnosis. Multivariate models predicted sepsis in the next 24 h: cross-validated C-statistic for logistic regression and random forest were 0.74 (95% confidence interval (CI): 0.71-0.77) and 0.76 (95% CI: 0.73-0.79), respectively.

CONCLUSIONS

Statistical models based on physiological and biochemical data already available in the PICU identify high-risk patients up to 24 h prior to the clinical diagnosis of sepsis. The random forest model was superior to logistic regression in capturing the context of age.

Disparate diagnostic accuracy of lung function tests as predictors of poor asthma control in children

Rationale: In practice, asthma control is assessed according to symptom burden and office spirometry. However, spirometry poorly tests peripheral lung function, which may be abnormal in asthma. Impluse oscillometry (IOS) and multiple-breath washout (MBW) are novel methods which measure reactance (X5) and ventilation heterogeneity (VH) in the peripheral lung, but how well these tests reflect asthma control is poorly understood. Objective: To compare the diagnostic accuracy of tests of large airways caliber (FEV₁, FEV₁/FVC, R20), peripheral zone properties (X5, VH), and airways inflammation (FeNO) as predictors of poor control in asthmatic children (44 poorly controlled/10 controlled). Methods: 54 children enriched in severe asthma completed a symptom-based control scale (ACT/cACT) and lung function tests after overnight bronchodilator withhold. The accuracy of each variable to predict poor control was ranked by area under the receiver operating characteristic (ROC) curve, sensitivity and specificity. Results: Among measures of large airways caliber, the FEV₁% had the highest ROC curve area, with low sensitivity but perfect specificity. Among measures of peripheral lung function, X5 and VH in the conducting zone had fair curve areas with higher sensitivity but lower specificity compared to the FEV₁%. VH in the acinar zone and FeNO both had poor accuracy. Conclusion: Tests of large airway and peripheral zone lung function performed disparately as predictors of poor control in a sample of children enriched in severe asthma. Further studies in a larger sample with more diverse phenotypic features are necessary to validate this preliminary conclusion.

Heart rate ranges in premature neonates using high resolution physiologic data

OBJECTIVE

There are limited evidence-based published heart rate ranges for premature neonates. We determined heart rate ranges in premature neonates based on gestational and post-menstrual age.

STUDY DESIGN

Retrospective observational study of premature neonates admitted to the neonatal intensive care unit at the University of Virginia between January 2009 and October 2015. We included gestational ages between 23 0/7 weeks and 34 6/7 weeks. We stratified data by gestational and post-menstrual age groups.

RESULTS

Over two billion heart rate values in 1703 neonates were included in our study. We established percentile-based reference ranges based on gestational and post-menstrual age. Our results demonstrate a slight increase in the initial weeks after birth, followed by a gradual decline with age. The baseline heart rate is lower with advancing gestational age.

CONCLUSIONS

Knowing heart rate reference ranges in the premature neonatal population can be helpful in the bedside assessment of the neonate.

Global respiratory syncytial virus-associated mortality in young children (RSV GOLD): a retrospective case series

Background

Respiratory syncytial virus (RSV) infection is an important cause of pneumonia mortality in young children. However, clinical data for fatal RSV infection are scarce. We aimed to identify clinical and socioeconomic characteristics of children aged younger than 5 years with RSV-related mortality using individual patient data.

Methods

In this retrospective case series, we developed an online questionnaire to obtain individual patient data for clinical and socioeconomic characteristics of children aged younger than 5 years who died with community-acquired RSV infection between Jan 1, 1995, and Oct 31, 2015, through leading research groups for child pneumonia identified through a comprehensive literature search and existing research networks. For the literature search, we searched PubMed for articles published up to Feb 3, 2015, using the key terms “RSV”, “respiratory syncytial virus”, or “respiratory syncytial viral” combined with “mortality”, “fatality”, “death”, “died”, “deaths”, or “CFR” for articles published in English. We invited researchers and clinicians identified to participate between Nov 1, 2014, and Oct 31, 2015. We calculated descriptive statistics for all variables.

Findings

We studied 358 children with RSV-related in-hospital death from 23 countries across the world, with data contributed from 31 research groups. 117 (33%) children were from low-income or lower middle-income countries, 77 (22%) were from upper middle-income countries, and 164 (46%) were from high-income countries. 190 (53%) were male. Data for comorbidities were missing for some children in low-income and middle-income countries. Available data showed that comorbidities were present in at least 33 (28%) children from low-income or lower middle-income countries, 36 (47%) from upper middle-income countries, and 114 (70%) from high-income countries. Median age for RSV-related deaths was 5·0 months (IQR 2·3–11·0) in low-income or lower middle-income countries, 4·0 years (2·0–10·0) in upper middle-income countries, and 7·0 years (3·6–16·8) in high-income countries.

Interpretation

This study is the first large case series of children who died with community-acquired RSV infection. A substantial proportion of children with RSV-related death had comorbidities. Our results show that perinatal immunisation strategies for children aged younger than 6 months could have a substantial impact on RSV-related child mortality in low-income and middle-income countries.

Funding

Bill & Melinda Gates Foundation.

Surgical Site Infection After Pediatric Cardiothoracic Surgery

BACKGROUND

Surgical site infection (SSI) occurs in 0.25% to 6% of children after cardiothoracic surgery (CTS). There are no published data regarding the financial impact of SSI after pediatric CTS. We sought to determine the attributable hospital cost and length of stay associated with SSI in children after CTS.

METHODS

We performed a retrospective, matched cohort study in a 26-bed cardiac intensive care unit (CICU) from January 2010 through December 2013. Cases with SSI were identified retrospectively and individually matched to controls 2:1 by age, gender, Risk Adjustment for Congenital Heart Surgery score, Society of Thoracic Surgeons-European Association for Cardiothoracic Surgery category, and primary cardiac diagnosis and procedure.

RESULTS

Of the 981 cases performed during the study period, 12 with SSI were identified. There were no differences in demographics, clinical characteristics, or intraoperative data. Median total hospital costs were higher in participants with SSI as compared to controls (US$219,573 vs US$82,623, P < .01). Children with SSI had longer median CICU length of stay (9 vs 3 days, P < .01), hospital length of stay (18 vs 8.5 days, P < .01), and duration of mechanical ventilation (2 vs 1 day, P < .01) and vasoactive administration (4.5 vs 1 day, P < .01).

CONCLUSIONS

Children with SSI after CTS have an associated increase in hospital costs of US$136,950/case and hospital length of stay of 9.5 days/case. The economic burden posed by SSI stress the importance of infection control surveillance, exhaustive preventative measures, and identification of modifiable risk factors.

Improving quality in measuring time to initiation of CPR during in-hospital resuscitation

OBJECTIVE

Time from the onset of "low or no flow" indicators of cardiac failure to initiation of cardiopulmonary resuscitation is an important quality metric thought to improve the likelihood of survival and preservation of end organ function. We hypothesized that delays in initiation of chest compressions were under recognized during in-hospital resuscitation and aimed to develop a system which identifies the actual time of deterioration during cardiac events.

METHODS

Retrospective review on prospectively identified resuscitation records and monitor data were compared. Return of spontaneous circulation, survival, and changes in functional status of patients pre- and post-events with chest compressions were collected as outcome measures.

RESULTS

Between October 2012 and April 2015, 59 events which met eligibility criteria occurred in either our pediatric cardiac or general pediatric intensive care units. The median time from event onset to initiation of chest compressions was 47s(s) (interquartile range (IQR) 28-80s) as assessed using monitor data, while the resuscitation record reported a median time of 0s (IQR 0-60s), reflecting the time from recognition to initiation of chest compressions. According to the resuscitation record, 81% vs. 63% of events achieved the quality standard of less than one minute depending on which review method was used (p=0.04).

CONCLUSIONS

There is a significant difference between time of deterioration to initiation of chest compressions and the time of recognition to initiation of chest compressions. Resuscitation records should be modified to include more information about the actual timing of patient deterioration.

Respiratory variation in peak aortic velocity accurately predicts fluid responsiveness in children undergoing neurosurgery under general anesthesia

The determination of fluid responsiveness in the critically ill child is of vital importance, more so as fluid overload becomes increasingly associated with worse outcomes. Dynamic markers of volume responsiveness have shown some promise in the pediatric population, but more research is needed before they can be adopted for widespread use. Our aim was to investigate effectiveness of respiratory variation in peak aortic velocity and pulse pressure variation to predict fluid responsiveness, and determine their optimal cutoff values. We performed a prospective, observational study at a single tertiary care pediatric center. Twenty-one children with normal cardiorespiratory status undergoing general anesthesia for neurosurgery were enrolled. Respiratory variation in peak aortic velocity (ΔVpeak ao) was measured both before and after volume expansion using a bedside ultrasound device. Pulse pressure variation (PPV) value was obtained from the bedside monitor. All patients received a 10 ml/kg fluid bolus as volume expansion, and were qualified as responders if stroke volume increased >15% as a result. Utility of ΔVpeak ao and PPV and to predict responsiveness to volume expansion was investigated. A baseline ΔVpeak ao value of greater than or equal to 12.3% best predicted a positive response to volume expansion, with a sensitivity of 77%, specificity of 89% and area under receiver operating characteristic curve of 0.90. PPV failed to demonstrate utility in this patient population. Respiratory variation in peak aortic velocity is a promising marker for optimization of perioperative fluid therapy in the pediatric population and can be accurately measured using bedside ultrasonography. More research is needed to evaluate the lack of effectiveness of pulse pressure variation for this purpose.

Unscheduled Pediatric Intensive Care Unit Admissions in Patients With Cardiovascular Disease: Clinical Features and Outcomes

BACKGROUND

An unscheduled readmission to the intensive care unit (ICU) is associated with increased morbidity and mortality in children. There is a paucity of data examining the impact of unscheduled admissions on outcomes in children with specific disease processes such as cardiovascular disease. We investigated the impact of scheduled versus unscheduled ICU admission on clinical outcomes and differences in patient characteristics in children with cardiovascular disease.

METHODS

This was a retrospective analysis of contemporaneously collected clinical data using the Virtual PICU Systems database. All consecutive admissions at 102 participating pediatric ICUs in patients with cardiovascular disease were collected from October 2010 to September 2012.

RESULTS

There were 48,653 admissions included in the analysis (44% scheduled and 56% unscheduled). The median patient age was 31 months. Unscheduled admissions were associated with longer ICU length of stay and increased mortality (both P < .001). Adjusting for age, weight, and primary ICU admission diagnosis (cardiovascular vs noncardiovascular), patients with unscheduled admissions had an increased odds of mortality (odds ratio = 4.8, P < .001).

CONCLUSIONS

Unscheduled ICU admissions were associated with worse clinical outcomes including increased mortality. Efforts targeted at reducing unscheduled admissions in at-risk patients are warranted.

Risk Model of Bacterial Coinfection in Children with Severe Viral Bronchiolitis

Background Among children with respiratory failure from viral lower respiratory tract infection (LRTI), up to 39% will develop pulmonary bacterial coinfection, yet nearly all will receive antibiotics. We sought to identify patients with viral LRTI requiring mechanical ventilation at low risk of bacterial coinfection through the use of a risk prediction model. Methods We performed a retrospective cohort study identifying all patients admitted to the intensive care unit with laboratory-confirmed viral LRTI requiring invasive mechanical ventilation over a 2-year period and partitioned data in experimental and validation datasets. A multivariate probit regression model was constructed including variables associated with bacterial coinfection in the experimental dataset. Model was validated and recalibrated using the validation dataset. Model discrimination was assessed using receiver operating characteristic curve analysis. Results There were 126 patients included in the analysis. Variables associated with bacterial coinfection included tracheostomy in situ, Gram-stained smear white blood cells, and bacteria. The final recalibrated model discriminating between no coinfection and coinfection had an area under the curve of 0.8696. Conclusion Our prediction model identifies patients with viral LRTI requiring mechanical ventilation at very low risk of bacterial coinfection and has the potential to decrease antibiotic utilization without negatively impacting clinical outcome.

Postoperative Abdominal NIRS Values Predict Low Cardiac Output Syndrome in Neonates

Background:

The development of low cardiac output syndrome (LCOS) after cardiopulmonary bypass (CPB) occurs in up to 25% of neonates and is associated with increased morbidity. Invasive cardiac output monitors such as pulmonary artery catheters have limited availability and are costly. Near-infrared spectroscopy (NIRS) is a noninvasive tool for monitoring regional oxygenation in neonates in the cardiac intensive care unit (CICU). We hypothesize that anterior abdominal NIRS may aid in the early identification of LCOS after cardiac surgery.

Methods:

Prospective observational study from October 2013 to October 2014 of all neonates with congenital heart disease admitted to the CICU following CPB. Abdominal NIRS values were continuously recorded upon CICU admission and for the subsequent 24-hour period. The primary outcome was the development of LCOS. Low cardiac output syndrome was defined as the presence of metabolic lactic acidosis (pH < 7.3 and lactate > 4) or addition of a new vasoactive agent or a vasoactive inotropic score > 15. Autoregressive time series models were constructed for each patient based on the continuously recorded NIRS values, and patients were stratified by development of LCOS.

Results:

Twenty-seven neonates met inclusion criteria, of whom 11 developed LCOS. Neonates who developed LCOS had lower constant NIRS values (49% vs 66%, P < .001). Constant NIRS values less than 58% best predicted development of LCOS with a sensitivity of 100% and specificity of 69%.

Conclusion:

Lower constant anterior abdominal NIRS values in the early postoperative period may allow early identification of neonates at risk for LCOS.

Unplanned ICU Transfers from Inpatient Units: Examining the Prevalence and Preventability of Adverse Events Associated with ICU Transfer in Pediatrics

Background Adverse events have been associated with unplanned intensive care unit (ICU) transfers in adults. Objective To examine trends in unplanned ICU transfers in pediatrics resulting from adverse events. Design, Setting, Patients Retrospective observational study of pediatric and cardiac ICU transfers from acute care units during a 2-year period in a tertiary care children's hospital. Methods Transfers were identified via electronic health record query and investigated for adverse events. Predefined adverse events included ICU transfers within 12 hours of admission to an acute care unit, readmissions to an ICU within 24 hours, and cardiopulmonary arrest on an acute care unit. Other adverse events examined were not predefined. Adverse events were evaluated for preventability and categorized by type, diagnosis, time of day and weekday versus weekend occurrence, and level of associated patient harm. Results There were 1,008 ICU transfers during the study period; 67% were unplanned. Of the unplanned transfers, 32% were attributed to adverse events, 35% of which were preventable. Unplanned transfers associated with a high rate of preventable adverse events included readmission to an ICU within 24 hours (58%, p = 0.002) and ICU transfer within 12 hours of acute care admission (34%). Conclusions We observed a high rate of preventable adverse events associated with unplanned pediatric ICU transfers, many of which were due to inappropriate triage. Readmission to an ICU within 24 hours of transfer to an acute care unit was significantly associated with preventability.

The Use of Ophthalmic Ultrasonography to Identify Retinal Injuries Associated With Abusive Head Trauma

Abusive head trauma includes any nonaccidental injury inflicted to a child's head and body. It is often characterized by, but not limited to, the repetitive acceleration-deceleration forces with or without blunt head impact. It has a mortality rate of 30%, and 80% of survivors experience permanent neurologic damage. In this case series, we hypothesize that bedside ultrasonography can be useful in the identification of retinal injuries that are consistent with abusive head trauma. Ocular manifestations of abusive head trauma are identified by dilated ophthalmic examination showing retinal hemorrhages that are too numerous to count, multilayered, and extending to the periphery. Traumatic retinoschisis, splitting of the retinal layers with or without blood accumulating in the intervening space, is exclusive for abusive head trauma in infants without a history of significant cerebral crush injury. Direct visualization of intraocular structures is difficult when the eyelids are swollen shut or when dilatation must be delayed. We present a series of 11 patients with brain injuries who underwent ophthalmic point-of-care ultrasonography that revealed traumatic retinoschisis on average 60 hours earlier than direct ophthalmic visualization. Dilated ophthalmic examinations and autopsy reports confirmed retinoschisis and other forms of retinal hemorrhages that were too numerous to count, multilayered, and extending to the periphery in all 11 patients. One patient did not have a dilated ophthalmic examination; however, traumatic retinoschisis and retinal hemorrhages were confirmed on autopsy. Ocular point-of-care ultrasonography is a promising tool to investigate abusive head trauma through the identification of traumatic retinoschisis and retinal hemorrhages when pupillary dilatation and direct ophthalmic examination is delayed.

Monitoring the harm associated with use of anticoagulants in pediatric populations through trigger-based automated adverse-event detection

BACKGROUND

The safety profile of anticoagulants, which are being used with increasing frequency in pediatric populations, is not well studied. Automatic triggers built into electronic health record systems (EHR) have been shown to be an effective way to monitor for and identify medication errors. Anticoagulant-associated adverse events were examined through the use of an anticoagulant trigger panel.

METHODS

In a retrospective, five-year (September 2007-September 2012) observational study, four automated triggers were used to detect anticoagulant-related adverse events: activated partial thromboplastin time (aPTT) > 100 seconds in patients on an unfractionated heparin (UFH) infusion, International Normalized Ratio (INR) > 4, anti-factor Xa (anti-FXa) >1.5U/mL for patients on enoxaparin, and the documented use of protamine.

RESULTS

For the 1,664 triggers evaluated, 12 were associated with the aPTT trigger, only 1 of which was preventable. Receiver operator characteristic curve analysis indicated that increasing the aPTT trigger > 140 seconds would optimize sensitivity and specificity. The INR trigger identified four outpatients with adverse events. No adverse events were associated with the anti-FXa trigger. The protamine trigger identified 12 adverse events and was associated with more severe events. Minimal overlap was found with protamine and aPTT triggers.

CONCLUSION

Laboratory- and medication-based triggers can be effective monitoring tools for anticoagulants. For patients receiving a UFH infusion, an aPTT cutoff value of > 140 seconds is more precise. We also found that protamine use as a trigger adds value to a trigger-based anticoagulant monitoring system. Continued improvement in the logic algorithms associated with the EHR-based trigger tool will allow expanded use of this tool in a clinical manner.

Off-label drug use in a single-center pediatric cardiac intensive care unit

BACKGROUND

The frequency of off-label drug use and its association with morbidity and mortality in the cardiac intensive care unit (CICU) has not been previously studied.

METHODS

Patients less than 18 years of age admitted to the CICU from June to August 2008 were retrospectively identified. Patient demographics were collected for 30 days or until CICU discharge. Off-label drug use was defined as the prescription of a medication that lacked a labeled indication based on patient's age as reported in the Micromedex drug database and electronic Physician's Desk Reference.

RESULTS

Eighty-two patients were admitted to the CICU during the study period. In all, 40 (46%) patients were male; the median age was 10.6 months. Common diagnoses were left-to-right shunt lesions (20.7%) and single-ventricle lesions (20.7%), with an overall mortality of 2.4%. Of all drugs prescribed, 36% were off-label. In all, 94% of the patients received ≥1 drug off-label. The median number of drugs prescribed off-label was four. Patients receiving more than four off-label medications were younger, had longer CICU lengths of stay (median 9.5 vs 2 days, P < .001), and increased ventilator days (median two vs one day, P < .001).

CONCLUSIONS

Off-label drug use in the CICU is common. Frequency of use is likely higher in patients with a higher severity of illness. Further safety, efficacy, and pharmaceutical trials are warranted to optimize the use of these drugs to improve outcomes.

Hospital readmission in children with complex chronic conditions discharged from subacute care

OBJECTIVES

Children with complex chronic conditions (CCC) are responsible for a disproportionate number of hospital readmissions. This study sought to determine 30-day hospital readmission rates in children with CCC discharged from a rehabilitation and transitional care hospital and to identify factors associated with increased risk of readmission.

METHODS

We conducted a retrospective cohort study identifying children with CCC discharged over an 18-month period from a subacute care facility staffed by hospitalists from a freestanding children's hospital. The primary outcome measure was readmission to the referring acute care hospital within 30 days of the subacute discharge.

RESULTS

Of the 272 discharged patients meeting inclusion criteria as children with at least 1 CCC, 19% had at least 1 readmission within 30 days of discharge. On univariate analysis, readmission was associated with the number of home medications (P = .001), underlying chronic respiratory illness (P < .001), home apnea or pulse oximetry monitor use (P = .02), tracheostomy and/or ventilator dependence (P = .003), length of stay (P = .04), and number of follow-up appointments (P = .02). On multivariate analysis, the number of discharge medications was associated with increased odds of readmission (odds ratio: 1.11 [95% confidence interval: 1.03-1.20]; P = .01). Receiver operating curve analysis identified a cutoff of 8 medications as most associated with readmission; in patients discharged with ≥8 medications, the hospital readmission rate was 29%.

CONCLUSIONS

This is the first known study that investigated hospital readmission rates in children with CCC discharged from a subacute facility and specifically identified the number of discharge medications as a significant risk factor for readmission.

Existing data analysis in pediatric critical care research

Our objectives were to review and categorize the existing data sources that are important to pediatric critical care medicine (PCCM) investigators and the types of questions that have been or could be studied with each data source. We conducted a narrative review of the medical literature, categorized the data sources available to PCCM investigators, and created an online data source registry. We found that many data sources are available for research in PCCM. To date, PCCM investigators have most often relied on pediatric critical care registries and treatment- or disease-specific registries. The available data sources vary widely in the level of clinical detail and the types of questions they can reliably answer. Linkage of data sources can expand the types of questions that a data source can be used to study. Careful matching of the scientific question to the best available data source or linked data sources is necessary. In addition, rigorous application of the best available analysis techniques and reporting consistent with observational research standards will maximize the quality of research using existing data in PCCM.

Demographics, trends, and outcomes in pediatric acute myocarditis in the United States, 2006 to 2011

BACKGROUND

There is a lack of clear diagnostic and management guidelines for acute myocarditis in the pediatric population. We used a multi-institutional database to characterize demographics, practice variability, and outcomes in this population.

METHODS AND RESULTS

Patients with acute myocarditis (n=514) were identified from April 2006 to March 2011 using the Pediatric Health Information System database, and regional variations in management and outcomes were analyzed. Ninety-seven patients (18.9%) received extracorporeal membrane oxygenation, 22 (4.3%) received ventricular assist device, 21 (4.1%) received heart transplantation, and 37 (7.2%) died. Of the 104 patients who received extracorporeal membrane oxygenation or ventricular assist device, 17 (16.3%) had heart transplantation, 25 (24%) died, and 62 (59.6%) showed recovery of myocardial function. There was a decrease in the use of endomyocardial biopsy (P=0.03) and an increase in the use of magnetic resonance imaging (P<0.01) over the study period. Although the use of medications and procedures varied between different regions, the occurrence of death or heart transplantation showed no significant regional associations. The use of extracorporeal membrane oxygenation (odds ratio, 5.8; 95% confidence interval, 2.9-11.4; P<0.01), ventricular assist device (odds ratio, 8.2; 95% confidence interval, 2.7-24.9; P<0.01), and vasoactive medications (odds ratio, 5.7; 95% confidence interval, 1.2-26.1; P=0.03) was independently associated with death/transplantation.

CONCLUSIONS

There is significant temporal and regional variation in the diagnostic modalities and management used for pediatric myocarditis, which continues to have high morbidity and mortality. Extracorporeal membrane oxygenation, ventricular assist device, and vasoactive medications are independently associated with increased mortality/transplantation.

Severe hypercalcemia without hypercalciuria in a previously healthy infant

Williams-Beuren syndrome (WBS) is a multisystem disorder that has a broad range of clinical findings including characteristic facial appearance, supravalvular aortic stenosis, dental and developmental abnormalities, and endocrinologic disorders including but not limited to the development of hypercalcemia. We present the case of a 10-month-old girl, with a history of intrauterine growth restriction, who presented with symptoms of weight loss and poor feeding. She was found to have severe elevation of her serum calcium to 20 mg/dL. She was subsequently diagnosed with WBS by fluorescent in situ hybridization analysis. The exact etiology of hypercalcemia in patients with WBS is unknown, but there are several hypotheses. Treatment of hypercalcemia in WBS is achieved with intravenous (IV) fluids, loop diuretics, and a low calcium diet; bisphosphonate therapy is required if adequate decreases in the serum calcium level are not achieved with traditional therapy.

Impact of perioperative RSV or influenza infection on length of stay and risk of unplanned ICU admission in children: a case-control study

Background

Children with viral respiratory infections who undergo general anesthesia are at increased risk of respiratory complications. We investigated the impact of RSV and influenza infection on perioperative outcomes in children undergoing general anesthesia.

Methods

We performed a retrospective case-control study. All patients under the age of 18 years who underwent general anesthesia at our institution with confirmed RSV or influenza infection diagnosed within 24 hours following induction between October 2002 and September 2008 were identified. Controls were randomly selected and were matched by surgical procedure, age, and time of year in a ratio of three controls per case. The primary outcome was postoperative length of stay (LOS).

Results

Twenty-four patients with laboratory-confirmed RSV or influenza who underwent general anesthesia prior to diagnosis of viral infection were identified and matched to 72 controls. Thirteen cases had RSV and 11 had influenza. The median postoperative LOS was three days (intra-quartile range 1 to 8 days) for cases and two days (intra-quartile range 1 to 5 days) for controls. Patients with influenza had a longer postoperative LOS (p < 0.001) and patients with RSV or influenza were at increased risk of unplanned admission to the PICU (p = 0.04) than matched controls.

Conclusions

Our results suggest that children with evidence of influenza infection undergoing general anesthesia, even in the absence of symptoms previously thought to be associated with a high risk of complications, may have a longer postoperative hospital LOS when compared to matched controls. RSV and influenza infection was associated with an increased risk of unplanned PICU admission.

Time series model to predict burden of viral respiratory illness on a pediatric intensive care unit

BACKGROUND

In the United States, viral respiratory infections are a leading cause of illness and hospitalization in young children. Caring for children with severe viral respiratory illness can have a substantial impact on resource utilization in the pediatric intensive care unit (PICU). The objective was to build a robust model that captures the periodicity of severe pediatric viral respiratory illness and forecasts the incidence of viral respiratory illness in the PICU.

METHODS

This was a retrospective time series analysis in a PICU at a quaternary care children's hospital. Patients were less than 18 years of age with laboratory-confirmed respiratory syncytial virus, influenza, parainfluenza, or adenovirus infection on or during admission from October 2002 to September 2008. Time series modeling techniques were to used to model viral incidence, using maximum likelihood estimation to identify model parameters.

RESULTS

A total of 289 patients were included in the analysis. An autoregressive model of order 10 that included an exogenous variable of community viral incidence from the previous month was able to explain and predict viral incidence in the PICU. A limitation of the study was that it included a single institution.

CONCLUSIONS

The identified model, derived from historical data from both a PICU and the local community, produced accurate 1-month and 3-month forecasts of severe viral respiratory illness presentation to the PICU. These results suggest that time series models may be useful tools in forecasting the burden of severe viral respiratory illness at the institutional level, helping institutions make decisions to optimize the distribution of resources.