Background Non-invasive, continuous monitoring of cardiac output (CO) could transform care of sick neonates through earlier detection and improved targeted management of cardiovascular compromise. Whole body electrical bioimpedance (WBEB) has been developed for non-invasive CO measurement but has yet to be validated for use in neonates. WBEB may have significant advantages over intermittent, operator-dependent echocardiography.

Objectives This study aimed to validate WBEB (NICaS monitor, NI Medical, Israel) for use in healthy, term and late preterm neonates, compared to echocardiography.

Methods Well neonates <12 hours old born to healthy mothers at ≥35 weeks gestational age were recruited. Two NICaS monitor pads were applied to supine babies in a left wrist-right ankle configuration for two hours; an echo was performed during this time by a consultant neonatologist trained in echocardiography. Left ventricular (LV) stroke volume (SV) was measured using standardised echo techniques. The NICaS monitor uses fluctuations in WBEB to calculate SV at 20-second intervals using a proprietary algorithm. The median of 15 minutes of NICaS data prior to the start of the echo for each baby was used in the analysis, since babies were more likely to become distressed during the echo, reducing the quality of the NICaS data through movement artefact. Extreme, non-physiological outlier values when babies were unsettled were excluded from NICaS SV data. R (R Core Team, 2019) software was used for data analysis, including descriptive statistics, Bland-Altman analysis and Pearson correlation.

Results 35 neonates were recruited (20 females), with a median (range) gestational age of 39⁺¹ weeks (35⁺⁶ – 42⁺²) and birth weight of 3.34kg (2.2–4.4kg). Monitoring was performed on day one for all babies, and additionally on day two for four babies who remained in the hospital. Five babies did not have NICaS data immediately prior to the echo due to the need to feed: therefore, we included 34/39 paired measurements in the final analysis. The mean (SD) echo LVSV was higher than that of NICaS SV (1.90±0.44 vs 1.52±0.38ml/kg; 95% CI: -0.57 to -0.29; p <0.0001). Bland-Altman bias was 0.43ml/kg, with limits of agreement from -0.36 to 1.21ml/kg. Mean percentage error was 40%, but when corrected for the percentage error of echo, the true precision was 27%. The Pearson correlation between the two measures was r=0.54 (p=0.001; 95% CI: 0.24 to 0.74).

Conclusions We postulate that the higher echo LVSV compared to NICaS may be because echo LVSV measurements were made pre-ductus arteriosus (patent in 28/34 measurements), while NICaS calculates SV from peripheral signals (post-ductus arteriosus). The NICaS’ true precision was 27%, which is within the clinically acceptable percentage error for new devices (30%), and there was a significant correlation between the NICAS SV and echo LVSV measurements. These results indicate that the NICaS monitor may be reliable for SV monitoring in healthy term and late-preterm neonates. If validity is confirmed in term and preterm infants, we envisage that WBEB could be used as a complementary clinical tool for continuous haemodynamic monitoring in neonatal intensive care, resulting in a step-change in clinical practice.