A microfluidic device for real-time on-demand intravenous oxygen delivery.

SignificanceThe treatment of hypoxemia that is refractory to the current standard of care is time-sensitive and requires skilled caregivers and use of specialized equipment (e.g., extracorporeal membrane oxygenation). Most patients experiencing refractory hypoxemia will suffer organ dysfunction, and death is common in this cohort. Here, we describe a new strategy to stabilize and support patients using a microfluidic device that administers oxygen gas directly to the bloodstream in real time and on demand using a process that we call sequential shear-induced bubble breakup. If successful, the described technology may help to avoid or decrease the incidence of ventilator-related lung injury from refractory hypoxemia.

Administration of Milrinone Following Tetralogy of Fallot Repair Increases Postoperative Volume Administration Without Improving Cardiac Output.

BACKGROUND: Phosphodiesterase inhibitors are known to relieve symptoms in the setting of heart failure, although their effects in restrictive ventricular physiology have been poorly characterized. We explored the association between the use of milrinone and volume administration during the first 72 hours following surgical repair of tetralogy of Fallot (TOF). METHODS: We reviewed all cases of primary surgical repair of TOF with pulmonary stenosis or atresia at Boston Children's Hospital between 2011 and 2020. To adjust for baseline differences between patients who did and did not receive milrinone, we matched patients with similar propensity scores in a 1:1 ratio (use of milrinone versus not). We then compared the need for volume administration during the first 72 hours postoperatively, vital signs, and measures of cardiac output between the matched cohorts. Additionally, in the group of patients receiving milrinone, linear regression modeling was used to explore the relationship between total dose of milrinone and total volume administration. RESULTS: Among 351 included patients, 134 received perioperative milrinone. A total of 212 patients (106 per group) were matched based on anatomic and surgical risks using a propensity score. After propensity matching, compared with nonmilrinone-treated patients, milrinone-treated patients were given postoperative volume more frequently (66% vs 52%; difference 14% [95% confidence interval, CI, 1%-27%]; P = .036). Milrinone-treated patients had a slower recovery of tachycardia during the first 12 hours (difference in slope 0.30 [95% CI, 0.14-0.47] beats per minute [BPM]/h; P < .001), and the intergroup difference peaked at 12 hours postoperatively (8 [95% CI, 5-12] BPM). Milrinone administration was not associated with improved cardiac output, including arteriovenous oxyhemoglobin saturation difference. In propensity-matched patients receiving milrinone, the total volume administered during the first 72 postoperative hours was significantly associated with the cumulative dose of postoperative milrinone ( r = 0.20; 95% CI, 0.01-0.38; P = .036). Based on the slope of the regression line, for every 1000 µg/kg of milrinone (equivalent to ~0.25 µg/kg/min for 72 hours) administered in the first 72 postoperative hours, an estimated 11.0 (95% CI, 0.6-21.4) mL/kg additional volume was administered. CONCLUSIONS: The use of milrinone within the first 72 hours following TOF repair is associated with more frequent administration of volume, a positive association between a higher total dose of postoperative milrinone and the amount of postoperative volume administered, a higher heart rate, and a lower blood pressure, but is not associated with improved cardiac output.

Shock Index, Coronary Perfusion Pressure, and Rate Pressure Product As Predictors of Adverse Outcome After Pediatric Cardiac Surgery.

OBJECTIVES: To determine whether shock index, coronary perfusion pressure, or rate pressure product in the first 24 hours after congenital heart surgery are independent predictors of subsequent clinically significant adverse outcomes. DESIGN: A retrospective cohort study. SETTING: A tertiary care center. PATIENTS: All patients less than 18 years old who underwent cardiac surgery at Boston Children's Hospital between January 1, 2010, and December 31, 2018. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Shock index (heart rate/systolic blood pressure), coronary perfusion pressure (diastolic blood pressure-right atrial pressure), and rate pressure product (heart rate × systolic blood pressure) were calculated every 5 seconds, and the median value for the first 24 hours of cardiac ICU admission for each was used as a predictor. The composite, primary outcome was the occurrence of any of the following adverse events in the first 7 days following cardiac ICU admission: cardiopulmonary resuscitation, extracorporeal cardiopulmonary resuscitation, mechanical circulatory support, unplanned surgery, heart transplant, or death. The association of each variable of interest with this outcome was tested in a multivariate logistic regression model. Of the 4,161 patients included, 296 (7%) met the outcome within the specified timeframe. In a multivariate regression model adjusted for age, surgical complexity, inotropic and respiratory support, and organ dysfunction, shock index greater than 1.83 was significantly associated with the primary outcome (odds ratio, 6.6; 95% CI, 4.4-10.0), and coronary perfusion pressure greater than 35 mm Hg was protective against the outcome (odds ratio, 0.5; 0.4-0.7). Rate pressure product was not found to be associated with the outcome. However, the predictive ability of the shock index and coronary perfusion pressure models were not superior to their component hemodynamic variables alone. CONCLUSIONS: Both shock index and coronary perfusion pressure may offer predictive value for adverse outcomes following cardiac surgery in children, although they are not superior to the primary hemodynamic variables.

Avoidable Serum Potassium Testing in the Cardiac ICU: Development and Testing of a Machine-Learning Model.

OBJECTIVES: To create a machine-learning model identifying potentially avoidable blood draws for serum potassium among pediatric patients following cardiac surgery. DESIGN: Retrospective cohort study. SETTING: Tertiary-care center. PATIENTS: All patients admitted to the cardiac ICU at Boston Children's Hospital between January 2010 and December 2018 with a length of stay greater than or equal to 4 days and greater than or equal to two recorded serum potassium measurements. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We collected variables related to potassium homeostasis, including serum chemistry, hourly potassium intake, diuretics, and urine output. Using established machine-learning techniques, including random forest classifiers, and hyperparameter tuning, we created models predicting whether a patient's potassium would be normal or abnormal based on the most recent potassium level, medications administered, urine output, and markers of renal function. We developed multiple models based on different age-categories and temporal proximity of the most recent potassium measurement. We assessed the predictive performance of the models using an independent test set. Of the 7,269 admissions (6,196 patients) included, serum potassium was measured on average of 1 (interquartile range, 0-1) time per day. Approximately 96% of patients received at least one dose of IV diuretic and 83% received a form of potassium supplementation. Our models predicted a normal potassium value with a median positive predictive value of 0.900. A median percentage of 2.1% measurements (mean 2.5%; interquartile range, 1.3-3.7%) was incorrectly predicted as normal when they were abnormal. A median percentage of 0.0% (interquartile range, 0.0-0.4%) critically low or high measurements was incorrectly predicted as normal. A median of 27.2% (interquartile range, 7.8-32.4%) of samples was correctly predicted to be normal and could have been potentially avoided. CONCLUSIONS: Machine-learning methods can be used to predict avoidable blood tests accurately for serum potassium in critically ill pediatric patients. A median of 27.2% of samples could have been saved, with decreased costs and risk of infection or anemia.

Value of provocative electrophysiology testing in the management of pediatric patients after congenital heart surgery.

OBJECTIVE: To determine the impact of provocative electrophysiology testing in postoperative congenital heart disease (CHD) patients on the management of supraventricular tachycardia (SVT) and clinical outcomes. METHODS: This is a retrospective study including patients <18 years of age with surgery for CHD who had postoperative SVT between 2006 and 2017. Postoperative outcomes were compared between patients with and without postoperative electrophysiology testing using the Wilcoxon rank sum test, Fisher's exact test, Kaplan-Meier method with the log-rank test, and Cox proportional hazard model. RESULTS: From 341 patients who had SVT after surgery for CHD, 65 (19%) had postoperative electrophysiology testing. There was no significant difference in baseline patient characteristics or surgical complexity between patients with and without electrophysiology testing. Patients with inducible SVT on electrophysiology testing were more likely to have recurrence of SVT prior to hospital discharge with an odds ratio 4.0 (95% confidence interval 1.3, 12.0). Patients who underwent postoperative electrophysiology testing had shorter intensive care unit (12 [6, 20] vs 16 [9, 32] days, HR 2.1 [95% CI 1.6, 2.8], P < .001) and hospital (25 [13, 38] vs 31 [18, 54] days, HR 1.8 [95% CI 1.4, 2.4], P < .001) length of stay. CONCLUSION: Postoperative electrophysiology testing was associated with improved postoperative outcomes, likely related to the ability to predict recurrence of arrhythmia and tailored antiarrhythmic management.

Sensitivity of a Next-Generation NIRS Device to Detect Low Mixed Venous Oxyhemoglobin Saturations in the Single Ventricle Population.

Regional cerebral oxygenation index (rSO2) based on near-infrared spectroscopy (NIRS) is frequently used to detect low venous oxyhemoglobin saturation (ScvO2). We compared the performance of 2 generations of NIRS devices. Clinically obtained, time-matched cerebral rSO2 and ScvO2 values were compared in infants monitored with the FORE-SIGHT (n = 73) or FORE-SIGHT ELITE (n = 47) by linear regression and Bland-Altman analyses. In both devices, cerebral rSO2 correlated poorly with measured ScvO2 (FORE-SIGHT partial correlation 0.50 [95% confidence interval {CI}, 0.40-0.58]; FORE-SIGHT ELITE partial correlation 0.47 [0.39-0.55]) and mean bias was +8 (standard deviation [SD] 13.2) for FORE-SIGHT and +14 (SD 12.5) for FORE-SIGHT ELITE. When ScvO2 was <30%, rSO2 was <40 in 8% of FORE-SIGHT ELITE readings. Future NIRS should be validated in more hypoxic cohorts.

Impact of the superior cavopulmonary anastomosis on cerebral oxygenation.

BACKGROUND: Patients with univentricular heart disease may undergo a superior cavopulmonary anastomosis, an operative intervention that raises cerebral venous pressure and impedance to cerebral venous return. The ability of infantile cerebral autoregulation to compensate for this is not well understood. MATERIALS AND METHODS: We identified all patients undergoing a superior cavopulmonary anastomosis (cases) and compared metrics of cerebral oxygenation upon admission to the ICU with patients following repair of tetralogy of Fallot or arterial switch operation (controls). The primary endpoint was cerebral venous oxyhaemoglobin saturation measured from an internal jugular venous catheter. Other predictor variables included case-control assignment, age, weight, sex, ischemic times, arterial oxyhaemoglobin saturation, mean arterial blood pressure, and superior caval pressure. RESULTS: A total of 151 cases and 350 controls were identified. The first post-operative cerebral venous oxyhaemoglobin saturation was significantly lower following superior cavopulmonary anastomosis than in controls (44 ± 12 versus 59 ± 15%, p < 0.001), as was arterial oxyhaemoglobin saturation (81 ± 9 versus 98 ± 5%, p < 0.001). Cerebral venous oxyhaemoglobin saturation correlated poorly with superior caval pressure in both groups. When estimated by linear mixed effects model, arterial oxyhaemoglobin saturation was the primary determinant of central venous oxyhaemoglobin saturation in both groups (ß = 0.79, p = 3 × 10-14); for every 1% point increase in arterial oxyhaemoglobin saturation, there was a 0.79% point increase in venous oxyhaemoglobin saturation. In this model, no other predictors were significant, including superior caval pressure and case-control assignment. CONCLUSION: Cerebral autoregulation appears to remain intact despite acute imposition of cerebral venous hypertension following superior cavopulmonary anastomosis. Following superior cavopulmonary anastomosis, cerebral venous oxyhaemoglobin saturation is primarily determined by arterial oxyhaemoglobin saturation.

Changes in tissue oxygen tension, venous saturation, and Fick-based assessments of cardiac output during hyperoxia.

BACKGROUND: Hyperoxemia (arterial oxygen tension >100 mm Hg) may occur in critically ill patients and have effects on mixed venous saturation (SvO2 ) and on Fick-based estimates of cardiac output (CO). We investigated the effect of hyperoxemia on SvO2 and on assessments of CO using the Fick equation. METHODS: Yorkshire swine (n = 14) were anesthetized, intubated, and paralyzed for instrumentation. SvO2 (co-oximetry) and tissue oxygen tension (tPO2 , implantable electrodes) in brain and myocardium were measured during systematic manipulation of arterial oxygen tension (PaO2 ) using graded hyperoxia (fraction of inspired oxygen 0.21 → 0.8). Secondarily, oxygen- and carbon dioxide-based estimates of CO (FickO2 and FickCO2 , respectively) were compared with measurements from a flow probe placed on the aortic root. RESULTS: Independent of changes in measured oxygen delivery, cerebral and myocardial tPO2 increased in proportion to PaO2 , as did SvO2 (P < 0.001 for all). Based on mixed model analysis, each 100 mm Hg increase in PaO2 resulted in a 4.8 ± 0.9% increase in SvO2 under the conditions tested. Because neither measured oxygen consumption, arterial oxyhemoglobin saturation or cardiac output varied significantly during hyperoxia, changes in SvO2 resulted in successively increasing errors in FickO2 during hyperoxia (34% during normoxia, 72% during FiO2 0.8). FickCO2 lacked the progressively worsening errors present in FickO2 , but correlated poorly with CO. CONCLUSION: SvO2 acutely changes following changes in PaO2 even absent changes in measured DO2 . This may lead to errors in FickO2 estimates of CI. Further work is necessary to understand the impact of this phenomenon in disease states.

Oxygen delivery using engineered microparticles.

A continuous supply of oxygen to tissues is vital to life and interruptions in its delivery are poorly tolerated. The treatment of low-blood oxygen tensions requires restoration of functional airways and lungs. Unfortunately, severe oxygen deprivation carries a high mortality rate and can make otherwise-survivable illnesses unsurvivable. Thus, an effective and rapid treatment for hypoxemia would be revolutionary. The i.v. injection of oxygen bubbles has recently emerged as a potential strategy to rapidly raise arterial oxygen tensions. In this report, we describe the fabrication of a polymer-based intravascular oxygen delivery agent. Polymer hollow microparticles (PHMs) are thin-walled, hollow polymer microcapsules with tunable nanoporous shells. We show that PHMs are easily charged with oxygen gas and that they release their oxygen payload only when exposed to desaturated blood. We demonstrate that oxygen release from PHMs is diffusion-controlled, that they deliver approximately five times more oxygen gas than human red blood cells (per gram), and that they are safe and effective when injected in vivo. Finally, we show that PHMs can be stored at room temperature under dry ambient conditions for at least 2 mo without any effect on particle size distribution or gas carrying capacity.

Injectable Oxygen: Interfacing Materials Chemistry with Resuscitative Science.

Intravascular oxygen delivery holds great potential to treat numerous hypoxic conditions and emergencies, including pulmonary disorders, hypoxic tumors, hemorrhagic shock, stroke, cardiac arrest and so on. Tremendous effort has been made in the past to find material solutions for the development of intravenous oxygen carriers and have ranged from blood substitutes to microbubbles with limited success. This paper highlights previous and recent progress in perfluorocarbon-emulsions and microbubbles as intravenous gas carriers, including concerns over their long-term stability, in vivo safety profiles, and oxygen transport efficacy. Their use as potential resuscitative therapeutics for treating various types of cardiac arrest is also discussed.

Tunable Polymer Microcapsules for Controlled Release of Therapeutic Gases.

Encapsulation and delivery of oxygen, carbon dioxide, and other therapeutic gases, using polymeric microcapsules (PMCs) is an emerging strategy to deliver gas as an injectable therapeutic. The gas cargo is stored within the PMC core and its release is mediated by the physiochemical properties of the capsule shell. Although use of PMCs for the rapid delivery of gases has been well described, methods which tune the material properties of PMCs for sustained release of gas are lacking. In this work, we describe a simple method for the high-yield production of gas-in-oil-filled PMCs with tunable sizes and core gas content from preformed polymers using the sequential phase separation and self-emulsification of emulsion-based templates. We demonstrate that prolonged gas release occurs from gas-in-oil PMCs loaded with oxygen and carbon dioxide gas, each of which could have significant clinical applications.

A Device for the Quantification of Oxygen Consumption and Caloric Expenditure in the Neonatal Range.

BACKGROUND: The accurate measurement of oxygen consumption (VO2) and energy expenditure (EE) may be helpful to optimize the treatment of critically ill patients. However, current techniques are limited in their ability to accurately quantify these end points in infants due to a low VO2, low tidal volume, and rapid respiratory rate. This study describes and validates a new device intended to perform in this size range. METHODS: We created a customized device that quantifies inspiratory volume using a pneumotachometer and concentrations of oxygen and carbon dioxide gas in the inspiratory and expiratory limbs. We created a customized algorithm to achieve precise time alignment of these measures, incorporating bias flow and compliance factors. The device was validated in 3 ways. First, we infused a certified gas mixture (50% oxygen/50% carbon dioxide) into an artificial lung circuit, comparing measured with simulated VO2 and carbon dioxide production (VCO2) within a matrix of varying tidal volume (4-20 mL), respiratory rate (20-80 bpm), and fraction of inspired oxygen (0.21-0.8). Second, VO2, VCO2, and EE were measured in Sprague Dawley rats under mechanical ventilation and were compared to simultaneous Douglas bag collections. Third, the device was studied on n = 14 intubated, spontaneously breathing neonates and infants, comparing measured values to Douglas measurements. In all cases, we assessed for difference between the device and reference standard by linear regression and Bland-Altman analysis. RESULTS: In vitro, the mean ± standard deviation difference between the measured and reference standard VO2 was +0.04 ± 1.10 (95% limits of agreement, -2.11 to +2.20) mL/min and VCO2 was +0.26 ± 0.31 (-0.36 to +0.89) mL/min; differences were similar at each respiratory rate and tidal volume measured, but higher at fraction of inspired oxygen of 0.8 than at 0.7 or lower. In rodents, the mean difference was -0.20 ± 0.55 (-1.28 to +0.89) mL/min for VO2, +0.16 ± 0.25 (-0.32 to +0.65) mL/min for VCO2, and -0.84 ± 3.29 (-7.30 to +5.61) kcal/d for EE. In infants, the mean VO2 was 9.0 ± 2.5 mL/kg/min by Douglas method and was accurately measured by the device (bias, +0.22 ± 0.87 [-1.49 to +1.93] mL/kg/min). The average VCO2 was 8.1 ± 2.3 mL/kg/min, and the device exhibited a bias of +0.33 ± 0.82 (-1.27 to +1.94) mL/kg/min. Mean bias was +2.56% ± 11.60% of the reading for VO2 and +4.25% ± 11.20% of the reading for VCO2; among 56 replicates, 6 measurements fell outside of the 20% error range, and no patient had >1 of 4 replicates with a >20% error in either VO2 or VCO2. CONCLUSIONS: This device can measure VO2, VCO2, and EE with sufficient accuracy for clinical decision-making within the neonatal and pediatric size range, including in the setting of tachypnea or hyperoxia.

Physiologic effects of delayed sternal closure following stage 1 palliation.

BACKGROUND: Following stage 1 palliation, delayed sternal closure may be used as a technique to enhance thoracic compliance but may also prolong the length of stay and increase the risk of infection. METHODS: We reviewed all neonates undergoing stage 1 palliation at our institution between 2010 and 2017 to describe the effects of delayed sternal closure. RESULTS: During the study period, 193 patients underwent stage 1 palliation, of whom 12 died before an attempt at sternal closure. Among the 25 patients who underwent primary sternal closure, 4 (16%) had sternal reopening within 24 hours. Among the 156 infants who underwent delayed sternal closure at 4 [3,6] days post-operatively, 11 (7.1%) had one or more failed attempts at sternal closure. Patients undergoing primary sternal closure had a shorter duration of mechanical ventilation and intensive care unit length of stay. Patients who failed delayed sternal closure had a longer aortic cross-clamp time (123±42 versus 99±35 minutes, p=0.029) and circulatory arrest time (39±28 versus 19±17 minutes, p=0.0009) than those who did not fail. Failure of delayed sternal closure was also closely associated with Technical Performance Score: 1.3% of patients with a score of 1 failed sternal closure compared with 18.9% of patients with a score of 3 (p=0.0028). Among the haemodynamic and ventilatory parameters studied, only superior caval vein saturation following sternal closure was different between patients who did and did not fail sternal closure (30±7 versus 42±10%, p=0.002). All patients who failed sternal closure did so within 24 hours owing to hypoxaemia, hypercarbia, or haemodynamic impairment. CONCLUSION: When performed according to our current clinical practice, sternal closure causes transient and mild changes in haemodynamic and ventilatory parameters. Monitoring of SvO2 following sternal closure may permit early identification of patients at risk for failure.

Interfacial Nanoprecipitation toward Stable and Responsive Microbubbles and Their Use as a Resuscitative Fluid.

A new approach has been developed to prepare stable microbubbles (MBs) by interfacial nanoprecipitation of bioabsorbable polymers at air/liquid interfaces. This facile method offers robust control over the morphology and chemophysical properties of MBs by simple chemical modifications. This approach is amenable to large-scale manufacturing, and is useful to develop functional MBs for advanced biomedical applications. To demonstrate this, a MB-based intravenous oxygen carrier was created that undergoes pH-triggered self-elimination. Intravenous injection of previous MBs increased the risk of pulmonary vascular obstruction. However, we show, for the first time, that our current design is superior, as they 1) yielded no evidence of acute risks in rodents, and 2) improved the survival in a disease model of asphyxial cardiac arrest (from 0 to 100 %), a condition that affects more than 100 000 in-hospital patients, and carries a mortality of about 90 %.

Hemodynamic Effects of Lipid-Based Oxygen Microbubbles via Rapid Intravenous Injection in Rodents.

PURPOSE: Low oxygen levels, or hypoxemia, is a common cause of morbidity and mortality in critically ill patients. Hypoxemia is typically addressed by increasing the fraction of inspired oxygen, the use of mechanical ventilation, or more invasive measures. Recently, the injection of oxygen gas directly into the bloodstream by packaging it within lipid-based oxygen microbubbles (LOMs) has been explored. The purpose of this work is to examine the acute hemodynamic effects of intravenous injections of LOMs. METHODS: LOMs composed of 1,2-distearoyl-sn-glycero-3-phosphocoline and cholesterol were manufactured using a process of shear homogenization under an oxygen headspace. A 5 mL aliquot of either PlasmaLyte A, or low (37%) or high (55%) concentration LOMs (n = 10 per group) was injected over a 1 min period into Sprague Dawley rats instrumented for measurement of cardiac index and pulmonary (PVR) and systemic (SVR) vascular resistance during a 60 min observation period. Hemodynamics were compared between groups by linear mixed modeling. RESULTS: Approximately 1011 LOMs with mean diameter 3.77 ± 1.19 µm were injected over the 1 min period. Relative to controls, rodents treated with high concentration LOMs exhibited a higher pulmonary artery pressure (20 ± 0.4 mmHg vs 18 ± 0.4 mmHg, P < 0.001) and higher PVR (0.31 ± 0.01 vs 0.23 ± 0.01 mmHg/mL*min*kg, P < 0.001. Despite a stable cardiac index (62.2 ± 3.5 vs 62.3 ± 3.4 mL/min*kg, P < 0.001), mean arterial blood pressure decreased significantly in LOM-treated animals (46 ± 2 vs 60 ± 2 mmHg, P < 0.001) due to a decrease in SVR. Injections with aged LOM emulsions (>48 h since manufacture) resulted in a higher incidence of hemodynamic collapse during the observation period (P = 0.02). CONCLUSIONS: LOMs may be injected in quantities sufficient to deliver clinically meaningful volumes of oxygen but cause significant decrements in blood pressure and elevations in PVR.

Arch watch: current approaches and opportunities for improvement.

Coarctation of the aorta (CoA) is a ductus arteriosus (DA)-dependent form of congenital heart disease (CHD) characterized by narrowing in the region of the aortic isthmus. CoA is a challenging diagnosis to make prenatally and is the critical cardiac lesion most likely to go undetected on the pulse oximetry-based newborn critical CHD screen. When undetected CoA causes obstruction to blood flow, life-threatening cardiovascular collapse may result, with a high burden of morbidity and mortality. Hemodynamic monitoring practices during DA closure (known as an "arch watch") vary across institutions and existing tools are often insensitive to developing arch obstruction. Novel measures of tissue oxygenation and oxygen deprivation may improve sensitivity and specificity for identifying evolving hemodynamic compromise in the newborn with CoA. We explore the benefits and limitations of existing and new tools to monitor the physiological changes of the aorta as the DA closes in infants at risk of CoA.

Impact of Dexmedetomidine Infusion on Opioid and Benzodiazepine Doses in Ventilated Pediatric Patients in the Cardiac Intensive Care Unit.

INTRODUCTION: Dexmedetomidine (DEX) is frequently used as an adjunct agent for prolonged sedation in the intensive care unit (ICU), though its effect on concomitant opioids or benzodiazepines infusions is unclear. We explored the impact of DEX on concomitant analgosedation in a cohort of ventilated pediatric patients in a cardiac ICU, with stratification of patients according to duration of ventilation (< 5 versus ≥ 5 days) following DEX initiation. METHODS: We conducted a retrospective analysis on ventilated patients receiving a DEX infusion ≥ 24 h and at least one other sedative/analgesic infusion (January 2011-June 2021). We evaluated trends of daily doses of opioids and benzodiazepines from 24 h before to 72 h following DEX initiation, stratifying patients based on ventilation duration after DEX initiation (< 5 versus ≥ 5 days). RESULTS: After excluding 1146 patients receiving DEX only, 1073 patients were included [median age 234 days (interquartile range 90, 879)]. DEX was associated with an opioid infusion in 99% of patients and a benzodiazepine infusion in 62%. Among patients ventilated for < 5 days (N = 761), opioids increased in the first 24 h following DEX initiation [+ 1.12 mg/kg/day (95% CI 0.96, 1.23), P < 0.001], then decreased [- 0.90 mg/kg/day (95% CI - 0.89, - 0.71), P < 0.001]; benzodiazepines slowly decreased [- 0.20 mg/kg/day (95% CI - 0.21, - 0.19), P < 0.001]. Among patients ventilated for ≥ 5 days (N = 312), opioid administration doubled [+ 2.09 mg/kg/day (95% CI 1.82, 2.36), P < 0.001] in the first 24 h, then diminished minimally [- 0.18 mg/kg/day (95% CI - 0.32, - 0.04), P = 0.015] without returning to baseline; benzodiazepine administration decreased minimally [- 0.03 mg/kg/day (95% CI - 0.05, - 0.01), P = 0.010]. Similar trends were confirmed when adjusting for age, gender, surgical complexity, recent major invasive procedures, duration of mechanical ventilation before DEX initiation, extubation within 72 h following DEX initiation, mean hourly DEX dose, and use of neuromuscular blocking infusion. CONCLUSION: While in patients ventilated < 5 days opioids initially increased and then quickly decreased in the 72 h following DEX initiation, among patients ventilated ≥ 5 days opioids doubled, then decreased only minimally; benzodiazepines decreased minimally in both groups, although more slowly in the long-ventilation cohort. These findings may inform decision-making on timing of DEX initiation in ventilated patients already being treated with opioid or benzodiazepine infusions.

Predicting favorable response to intravenous morphine in pediatric critically ill cardiac patients.

INTRODUCTION: Analgesia and sedation are integral to the care of critically ill children. However, the choice and dose of the analgesic or sedative drug is often empiric, and models predicting favorable responses are lacking. We aimed to compute models to predict a patient's response to intravenous morphine. METHODS: We retrospectively analyzed data from consecutive patients admitted to the Cardiac Intensive Care Unit (January 2011-January 2020) who received at least one intravenous bolus of morphine. The primary outcome was a decrease in the State Behavioral Scale (SBS) ≥1 point; the secondary outcome was a decrease in the heart rate Z-score (zHR) at 30 min. Effective doses were modeled using logistic regression, Lasso regression, and random forest modeling. RESULTS: A total of 117,495 administrations of intravenous morphine among 8140 patients (median age 0.6 years [interquartile range [IQR] 0.19, 3.3]) were included. The median morphine dose was 0.051 mg/kg (IQR 0.048, 0.099) and the median 30-day cumulative dose was 2.2 mg/kg (IQR 0.4, 15.3). SBS decreased following 30% of doses, did not change following 45%, and increased following 25%. The zHR significantly decreased after morphine administration (median delta-zHR -0.34 [IQR-1.03, 0.00], p < 0.001). The following factors were associated with favorable response to morphine: A concomitant infusion of propofol, higher prior 30-day cumulative dose, being invasively ventilated and/or on vasopressors. Higher morphine dose, higher zHR pre-morphine, an additional analgosedation bolus ±30 min around the index bolus, a concomitant ketamine or dexmedetomidine infusion, and showing signs of withdrawal syndrome were associated with unfavorable response. Logistic regression (area under the receiver operating characteristic [ROC] curve [AUC] 0.900) and machine learning models (AUC 0.906) performed comparably, with a sensitivity of 95%, specificity of 71%, and negative predictive value of 97%. CONCLUSIONS: Statistical models identify 95% of effective intravenous morphine doses in pediatric critically ill cardiac patients, while incorrectly suggesting an effective dose in 29% of cases. This work represents an important step toward computer-aided, personalized clinical decision support tool for sedation and analgesia in ICU patients.