Sarah Walker, MD
Ann and Robert H Lurie Childrens Hospital of Chicago
Chicago, Illinois
Disclosure information not submitted.
Kyle Honegger, PhD
Data Scientist
Ann & Robert H. Lurie Children's Hospital of Chicago, United States
Disclosure information not submitted.
Michael Carroll, PhD
Research Assistant Professor
Ann & Robert H. Lurie Children's Hospital of Chicago, United States
Disclosure information not submitted.
Debra Weese-Mayer, MD
Professor of Pediatrics (Critical Care)
Ann & Robert H. Lurie Children's Hospital of Chicago, United States
Disclosure information not submitted.
L. Nelson Sanchez-Pinto, MD, MBI
Assistant Professor of Pediatrics (Critical Care)
Ann & Robert H. Lurie Children's Hospital of Chicago
Chicago, Illinois
Disclosure information not submitted.
Title: Utility of Arterial Waveform Analysis to Predict Fluid Responsiveness in Critically Ill Children.
Introduction:
Fluid boluses are commonly administered to children with hypotension, but response is variable. Inability to predict response to bolus may delay vasoactive infusion and result in fluid overload. Multiple physiologic variables derived from arterial waveforms have been studied to predict response to bolus in adults, but data in children are less robust. We hypothesize that the variables which predict increase in MAP in children will depend on degree of hypotension and clinical condition.
Methods:
This was a retrospective study of IV boluses ≥10 ml/kg given to children who had arterial waveform data captured in Bedmaster® during pediatric intensive care unit admissions from 3/2013-8/2020. At least 2hrs were required between boluses and every bolus had at least 30sec of adequate pre-bolus arterial waveform data. Stroke volume variation (SVV), pulse pressure variation (PPV), and dynamic arterial elastance (Eadyn) were calculated in 10sec intervals in the 20min pre-bolus. Fluid responsiveness (FR) was defined as increase by ≥ 10% in MAP from pre-bolus to average 20min post-bolus. Boluses were stratified by pre-bolus MAP as: < 10%ile, < 25%ile, and < 50%ile for age. Subgroup analysis compared those exposed to mechanical ventilation (MV) or vasoactive infusion (VI). Kruskal–Wallis test was used to assess associations.
Results:
A total of 1426 boluses administered to 609 children with recorded arterial waveforms were identified. Of these, 418 boluses in 310 patients met inclusion criteria. FR was more common in lower pre-bolus MAP cohorts (< 10%ile:48% vs. < 50%ile:30%). For the < 10%ile group, only Eadyn in patients without VI was associated with FR (p=0.02). For the < 25%ile, Eadyn was associated with FR (p=0.002) and this association persisted in subgroup analysis. For the < 50%ile, only SVV and PPV were associated with FR and only in intubated patients (PPV, p=0.04; SVV, p=0.03).
Conclusions:
Pre-bolus physiologic arterial waveform variables are inconsistently associated with FR in hypotensive children. The degree of hypotension and presence of MV or VI appear to modify this association. These variables could form the basis of a multivariable machine learning model to predict FR in a more consistent and reliable manner.