Robinder G. Khemani

Serum creatinine as stratified in the RIFLE score for acute kidney injury is associated with mortality and length of stay for children in the pediatric intensive care unit.

Objective:To evaluate the ability of the RIFLE criteria to characterize acute kidney injury in critically ill children. Design:Retrospective analysis of prospectively collected clinical data. Setting:Multidisciplinary, tertiary care, 20-bed pediatric intensive care unit. Patients:All 3396 admissions between July 2003 and March 2007. Interventions:None. Measurements and Main Results:A RIFLE score was calculated for each patient based on percent change of serum creatinine from baseline (risk = serum creatinine ×1.5; injury = serum creatinine ×2; failure = serum creatinine ×3). Primary outcome measures were mortality and intensive care unit length of stay. Logistic and linear regressions were performed to control for potential confounders and determine the association between RIFLE score and mortality and length of stay, respectively.One hundred ninety-four (5.7%) patients had some degree of acute kidney injury at the time of admission, and 339 (10%) patients had acute kidney injury develop during the pediatric intensive care unit course. Almost half of all patients with acute kidney injury had their maximum RIFLE score within 24 hrs of intensive care unit admission, and approximately 75% achieved their maximum RIFLE score by the seventh intensive care unit day. After regression analysis, any acute kidney injury on admission and any development of or worsening of acute kidney injury during the pediatric intensive care unit stay were independently associated with increased mortality, with the odds of mortality increasing with each grade increase in RIFLE score (p < .01). Patients with acute kidney injury at the time of admission had a length of stay twice that of those with normal renal function, and those who had any acute kidney injury develop during the pediatric intensive care unit course had a four-fold increase in pediatric intensive care unit length of stay. Also, other than being admitted with RIFLE risk score, an independent relationship between any acute kidney injury at the time of pediatric intensive care unit admission, any acute kidney injury present during the pediatric intensive care unit course, or any worsening RIFLE scores during the pediatric intensive care unit course and increased pediatric intensive care unit length of stay were identified after controlling for the same high-risk covariates (p < .01). Conclusions:RIFLE criteria serves well to describe acute kidney injury in critically ill pediatric patients.

Pediatric Acute Respiratory Distress Syndrome: Consensus Recommendations From the Pediatric Acute Lung Injury Consensus Conference

OBJECTIVE To describe the final recommendations of the Pediatric Acute Lung Injury Consensus Conference. DESIGN Consensus conference of experts in pediatric acute lung injury. SETTING Not applicable. SUBJECTS PICU patients with evidence of acute lung injury or acute respiratory distress syndrome. INTERVENTIONS None. METHODS A panel of 27 experts met over the course of 2 years to develop a taxonomy to define pediatric acute respiratory distress syndrome and to make recommendations regarding treatment and research priorities. When published, data were lacking a modified Delphi approach emphasizing strong professional agreement was used. MEASUREMENTS AND MAIN RESULTS A panel of 27 experts met over the course of 2 years to develop a taxonomy to define pediatric acute respiratory distress syndrome and to make recommendations regarding treatment and research priorities. When published data were lacking a modified Delphi approach emphasizing strong professional agreement was used. The Pediatric Acute Lung Injury Consensus Conference experts developed and voted on a total of 151 recommendations addressing the following topics related to pediatric acute respiratory distress syndrome: 1) Definition, prevalence, and epidemiology; 2) Pathophysiology, comorbidities, and severity; 3) Ventilatory support; 4) Pulmonary-specific ancillary treatment; 5) Nonpulmonary treatment; 6) Monitoring; 7) Noninvasive support and ventilation; 8) Extracorporeal support; and 9) Morbidity and long-term outcomes. There were 132 recommendations with strong agreement and 19 recommendations with weak agreement. Once restated, the final iteration of the recommendations had none with equipoise or disagreement. CONCLUSIONS The Consensus Conference developed pediatric-specific definitions for acute respiratory distress syndrome and recommendations regarding treatment and future research priorities. These are intended to promote optimization and consistency of care for children with pediatric acute respiratory distress syndrome and identify areas of uncertainty requiring further investigation.

Comparison of the Pulse Oximetric Saturation/Fraction of Inspired Oxygen Ratio and the Pao2/Fraction of Inspired Oxygen Ratio in Children

BACKGROUND Although diagnostic criteria for acute lung injury (ALI) and ARDS are clear, invasive arterial sampling is required for computation of Pao(2)/fraction of inspired oxygen (Fio(2)) [PF] ratios. The pulse oximetric saturation (Spo(2))/Fio(2) (SF) ratio may be a reliable noninvasive alternative to the PF ratio for identifying children with lung injury. METHODS We electronically queried blood gas measurements from two tertiary care pediatric ICUs (PICUs). Included in the analysis were corresponding measurements of Spo(2), Pao(2), and Fio(2) charted within 15 min of each other when Spo(2) values were between 80% and 97%. Computed PF and SF ratios were compared to identify threshold values for SF ratios that correspond to PF criteria for ALI (< or = 300) and ARDS (< or = 200). Data from one PICU were used for derivation and validated with measurements from the second PICU. RESULTS From the 1,298 observations in the derivation data set, SF ratio could be predicted by the regression equation SF = 76 + 0.62 x PF (p < 0.0001, R(2) = 0.61). SF ratios of 263 and 201 corresponded to PF ratios of 300 and 200, respectively. The ALI SF cutoff of 263 had 93% sensitivity and 43% specificity, and the ARDS cutoff of 201 had 84% sensitivity and 78% specificity. Applying these values to the 1,845 observations in the validation data set yielded a sensitivity of 86% and specificity of 47% for ALI and a sensitivity of 68% and specificity of 84% for ARDS. CONCLUSION SF ratio is a reliable noninvasive marker for PF ratio to identify children with ALI or ARDS.

Comparison of SpO2 to PaO2 based markers of lung disease severity for children with acute lung injury.

Objective:Given pulse oximetry is increasingly substituting for arterial blood gas monitoring, noninvasive surrogate markers for lung disease severity are needed to stratify pediatric risk. We sought to validate prospectively the comparability of SpO2/Fio2 to PaO2/Fio2 and oxygen saturation index to oxygenation index in children. We also sought to derive a noninvasive lung injury score. Design:Prospective, multicentered observational study in six pediatric intensive care units. Patients:One hundred thirty-seven mechanically ventilated children with SpO2 80% to 97% and an indwelling arterial catheter. Interventions:Simultaneous blood gas, pulse oximetry, and ventilator settings were collected. Derivation and validation data sets were generated, and linear mixed modeling was used to derive predictive equations. Model performance and fit were evaluated using the validation data set. Measurements and Main Results:One thousand one hundred ninety blood gas, SpO2, and ventilator settings from 137 patients were included. Oxygen saturation index had a strong linear association with oxygenation index in both derivation and validation data sets, given by the equation oxygen saturation index = 2.76 1 0.547*oxygenation index (derivation). 1/SpO2/Fio2 had a strong linear association with 1/PaO2/Fio2 in both derivation and validation data sets given by the equation 1/SpO2/Fio2 = 0.00232 1 0.443/PaO2/Fio2 (derivation). SpO2/Fio2 criteria for acute respiratory distress syndrome and acute lung injury were 221 (95% confidence interval 215–226) and 264 (95% confidence interval 259–269). Multivariate models demonstrated that oxygenation index, serum pH, and Paco2 were associated with oxygen saturation index (p < .05); and 1/PaO2/Fio2, mean airway pressure, serum pH, and Paco2 were associated with 1/SpO2/Fio2 (p < .05). There was strong concordance between the derived noninvasive lung injury score and the original pediatric modification of lung injury score with a mean difference of 20.0361 &agr;0.264 sd. Conclusions:Lung injury severity markers, which use SpO2, are adequate surrogate markers for those that use PaO2 in children with respiratory failure for SpO2 between 80% and 97%. They should be used in clinical practice to characterize risk, to increase enrollment in clinical trials, and to determine disease prevalence. (Crit Care Med 2012; 40:–1316)

Reducing the incidence of necrotizing enterocolitis in neonates with hypoplastic left heart syndrome with the introduction of an enteral feed protocol.

Objective: Neonates with hypoplastic left heart syndrome are prone to gastrointestinal complications, including necrotizing enterocolitis, during initiation or advancement of enteral feeds. A feeding protocol was developed to standardize practice across a multidisciplinary team. The purpose of this study was to examine the impact of a standardized feeding protocol on the incidence of necrotizing enterocolitis and overall postoperative gastrointestinal morbidity. Design: Retrospective case–control study. Setting: Cardiothoracic intensive care unit of a tertiary care children’s hospital. Patients: Ninety-eight neonates with hypoplastic left heart syndrome admitted to the cardiothoracic intensive care unit after first-stage palliation. Intervention: A retrospective chart review was performed. Two groups were analyzed: the preprotocol group (n = 52) was examined from January 2000 through December 31, 2001, and the postprotocol group (n = 46) from February 2002 through December 31, 2003. Measurements and Main Results: The incidence of suspected or diagnosed necrotizing enterocolitis as defined by the modified Bell staging criteria was recorded. Data were also collected regarding postoperative day of enteral feed initiation, postoperative day full feeds attained, and postoperative hospital length of stay. Necrotizing enterocolitis was detected in 14 preprotocol (27%) and three postprotocol (6.5%) patients (p < .01). Enteral feeds were initiated later in the postprotocol group (7.5 vs. 5.5 days, p < .001), and number of days to full feeds was also later in the postprotocol group (7 vs. 4 days, p = .02). Hospital length of stay tended to be shorter in the postprotocol group (21.5 vs. 28 days, p = .25). Conclusion: Measures directed at reducing the incidence of necrotizing enterocolitis may reduce morbidity in neonates with hypoplastic left heart syndrome and reduce cost by decreasing hospital length of stay. A standardized feeding protocol instituted to address these problems likely contributed to reducing the incidence of necrotizing enterocolitis in this high-risk population.

Outcome analysis of neonates with congenital diaphragmatic hernia treated with venovenous vs venoarterial extracorporeal membrane oxygenation.

PURPOSE Venoarterial extracorporeal membrane oxygenation (ECMO) (VA) is used more commonly in neonates with congenital diaphragmatic hernia (CDH) than venovenous ECMO (VV). We hypothesized that VV may result in comparable outcomes in infants with CDH requiring ECMO. METHODS We retrospectively analyzed the Extracorporeal Life Support Organization (ELSO) database (1991-2006). Multivariate logistic regression analyses were used to compare VV- and VA-associated mortality. RESULTS Four thousand one hundred fifteen neonates required ECMO, with an overall mortality rate of 49.6%. Venoarterial ECMO was used in 82% and VV in 18% of neonates. Pre-ECMO inotrope use and complications were equivalent between VA and VV. The mortality rate for VA and VV was 50% and 46%, respectively. After adjusting for birth weight, gestational age, prenatal diagnosis, ethnicity, Apgar scores, pH less than 7.20, Paco(2) greater than 50, requiring high-frequency ventilation, and year of ECMO, there was no difference in mortality between VV vs VA. Renal complications and on-ECMO inotrope use were more common with VV, whereas neurologic complications were more common with VA. The conversion rate from VV to VA was 18%; conversion was associated with a 56% mortality rate. CONCLUSION The short-term outcomes of VV and VA are comparable. Patients with CDH who fail VV may be predisposed to a worse outcome. Nevertheless, VV offers equal benefit to patients with CDH requiring ECMO while preserving the native carotid.

Unsupervised pattern discovery in electronic health care data using probabilistic clustering models

Bedside clinicians routinely identify temporal patterns in physiologic data in the process of choosing and administering treatments intended to alter the course of critical illness for individual patients. Our primary interest is the study of unsupervised learning techniques for automatically uncovering such patterns from the physiologic time series data contained in electronic health care records. This data is sparse, high-dimensional and often both uncertain and incomplete. In this paper, we develop and study a probabilistic clustering model designed to mitigate the effects of temporal sparsity inherent in electronic health care records data. We evaluate the model qualitatively by visualizing the learned cluster parameters and quantitatively in terms of its ability to predict mortality outcomes associated with patient episodes. Our results indicate that the model can discover distinct, recognizable physiologic patterns with prognostic significance.

Characteristics of Children Intubated and Mechanically Ventilated in 16 PICUs

BACKGROUND When designing multicenter clinical trials, it is important to understand the characteristics of children who have received ventilation in PICUs. METHODS This study involved the secondary analysis of an existing data set of all children intubated and mechanically ventilated from 16 US PICUs who were initially screened for a multicenter clinical trial on pediatric acute lung injury (ALI). RESULTS A total of 12,213 children between 2 weeks and 18 years of age who were intubated and mechanically ventilated were included, representing 30% of PICU admissions (center range, 20 to 64%). Of the children who received ventilation, 22% had cyanotic congenital heart disease; 26% had respiratory failure but not bilateral pulmonary infiltrates on chest radiograph; 8% had chronic respiratory disease; 7% had upper airway obstruction; and 5% had reactive airway disease. At least 1,457 patients (15%) with respiratory failure lacked an arterial line. Of these patients, 97% had a positive end-expiratory pressure <or= 8 cm H(2)O, and 80% were supported on an Fio(2) of <or= 0.40. Moreover, 104 of 904 patients (12%) with pulse oximetric saturation (Spo(2)) and Fio(2) measurements available would have met the oxygenation criteria for ALI according to Spo(2)/Fio(2) ratio criteria. CONCLUSIONS At least 30% of children in a cross-section of US PICUs are endotracheally intubated, and 25% of those with respiratory failure do not fulfill the radiographic criteria for ALI. Although few patients without an indwelling arterial line require more than modest ventilator support, many may still meet the oxygenation criteria for ALI. These findings will facilitate sample size calculations and help to determine feasibility for future trials on pediatric mechanical ventilation.

The association between the end tidal alveolar dead space fraction and mortality in pediatric acute hypoxemic respiratory failure

Objective: To investigate the relationship of markers of oxygenation, PaO2/FIO2 ratio, SpO2/FIO2 ratio, oxygenation index, oxygen saturation index, and dead space (end tidal alveolar dead space fraction) with mortality in children with acute hypoxemic respiratory failure. Design: Retrospective. Setting: Single-center tertiary care pediatric intensive care unit. Patients: Ninety-five mechanically ventilated children with a PaO2/FIO2 ratio <300 within 24 hrs of the initiation of mechanical ventilation. Interventions: None. Main Results: The end tidal alveolar dead space fraction, PaO2/FIO2 ratio, SpO2/FIO2 ratio, oxygenation index, and oxygen saturation index were all associated with mortality (p < .02). There was a small correlation between the end tidal alveolar dead space fraction and decreasing PaO2/FIO2 (r2 = .21) and SpO2/FIO2 ratios (r2 = .22), and increasing oxygenation index (r2 = .25) and oxygen saturation index (r2 = .24). In multivariate logistic regression modeling, the end tidal alveolar dead space fraction was independently associated with mortality (p < .02). Oxygenation index, oxygen saturation index, and the end tidal alveolar dead space fraction were all acceptable discriminators of mortality with receiver operating characteristic plot area under the curves ≥0.7. Conclusions: In pediatric acute hypoxemic respiratory failure, easily obtainable pulmonary specific markers of disease severity (SpO2/FIO2 ratio, oxygen saturation index, and the end tidal alveolar dead space fraction) may be useful for the early identification of children at high risk of death. Furthermore, the end tidal alveolar dead space fraction should be considered for risk stratification of children with acute hypoxemic respiratory failure, given that it was independently associated with mortality.

The design of future pediatric mechanical ventilation trials for acute lung injury.

Pediatric practitioners face unique challenges when attempting to translate or adapt adult-derived evidence regarding ventilation practices for acute lung injury or acute respiratory distress syndrome into pediatric practice. Fortunately or unfortunately, there appears to be selective adoption of adult practices for pediatric mechanical ventilation, many of which pose considerable challenges or uncertainty when translated to pediatrics. These differences, combined with heterogeneous management strategies within pediatric critical care, can complicate clinical practice and make designing robust clinical trials in pediatric acute respiratory failure particularly difficult. These issues surround the lack of explicit ventilator protocols in pediatrics, either computer or paper based; differences in modes of conventional ventilation and perceived marked differences in the approach to high-frequency oscillatory ventilation; challenges with patient recruitment; the shortcomings of the definition of acute lung injury and acute respiratory distress syndrome; the more reliable yet still somewhat unpredictable relationship between lung injury severity and outcome; and the reliance on potentially biased surrogate outcome measures, such as ventilator-free days, for all pediatric trials. The purpose of this review is to highlight these challenges, discuss pertinent work that has begun to address them, and propose potential solutions or future investigations that may help facilitate comprehensive trials on pediatric mechanical ventilation and define clinical practice standards.