Venous Physiology Predicts Dehydration in the Pediatric Population

Abstract

BackgroundNo standard dehydration monitor exists for children. This study attempts to determine the utility of Fast Fourier Transform (FFT) of a peripheral venous pressure (PVP) waveform to predict dehydration. Materials and methodsPVP waveforms were collected from 18 patients. Groups were defined as resuscitated (serum chloride ≥ 100 mmol/L) and hypovolemic (serum chloride < 100 mmol/L). Data were collected on emergency department admission and after a 20 cc/kg fluid bolus. The MATLAB (MathWorks) software analyzed nonoverlapping 10-s window signals; 2.4 Hz (144 bps) was the most demonstrative frequency to compare the PVP signal power (mmHg). ResultsAdmission FFTs were compared between 10 (56%) resuscitated and 8 (44%) hypovolemic patients. The PVP signal power was higher in resuscitated patients (median 0.174 mmHg, IQR: 0.079-0.374 mmHg) than in hypovolemic patients (median 0.026 mmHg, IQR: 0.001-0.057 mmHg), (P < 0.001). Fourteen patients received a bolus regardless of laboratory values: 6 (43%) resuscitated and 8 (57%) hypovolemic. In resuscitated patients, the signal power did not change significantly after the fluid bolus (median 0.142 mmHg, IQR: 0.032-0.383 mmHg) (P = 0.019), whereas significantly increased signal power (median 0.0474 mmHg, IQR: 0.019-0.110 mmHg) was observed in the hypovolemic patients after a fluid bolus at 2.4 Hz (P < 0.001). The algorithm predicted dehydration for window-level analysis (sensitivity 97.95%, specificity 93.07%). The algorithm predicted dehydration for patient-level analysis (sensitivity 100%, specificity 100%). ConclusionsFFT of PVP waveforms can predict dehydration in hypertrophic pyloric stenosis. Further work is needed to determine the utility of PVP analysis to guide fluid resuscitation status in other pediatric populations.