The difference between vision and hallucination is that other people can see the vision. – Marc Andreessen.
Background.
Non-invasive cardiac output monitoring (NICOM) using bioreactance (BRT) in pregnancy is gaining interest but lacks validation in the obstetric population, where changes in circulating blood volume, stroke volume and abdominal fluid content could significantly alter values. The authors compare simultaneous cardiac output (CO) measurements obtained using the NICOM® (BRT-CO) device and echocardiography (echo-CO), and also assessed the relationship between maternal characteristics and myocardial performance.
Methods.
Paired stroke volume (SV) and CO readings were obtained using NICOM® and echocardiography, in a group of healthy nulliparous women throughout a 15 min period, at rest. Agreement between NICOM® and echocardiography was assessed using Bland–Altman analysis and the intraclass correlation coefficient (ICC). Left ventricular (LV) function was assessed using systolic strain and tissue Doppler velocities (S′, E′, and A′ waves).
Results. Thirty-five women with a median [interquartile range] age, weight, and gestation of 29 [26–34] yr, 71 [64–79] kg, and 28 [21–29] weeks, respectively, were enrolled. There was good agreement between NICOM®-measured and echocardiographically measured SV [mean bias 6 ml (limits of agreement −18 to 29); ICC 0.8 (95% confidence interval 0.6–0.9), P<0.001] and CO [mean bias 0.2 litres (limits of agreement −1.3–1.7); ICC 0.8 (95% confidence interval 0.7–0.9), P<0.001; mean percentage error ±26%; coefficient of error (precision)=3.4%]. No confidence intervals were given for the limit of agreement values, and the comparator used for the Bland-Altman analysis was the mean between the NICOM-measured and echocardiographically measured values.
The mean (sd) LV S′ was 9.7 (2.3) cm s−1. The mean (sd) LV strain was −18.6 (2.6)%. There was a negative relationship between BMI and LV diastolic function measured using the E′:A′ ratio (r = −0.51, P<0.01).
Discussion.
The authors concluded that stroke volume and CO measurements obtained using NICOM® were comparable to those obtained using echocardiography, with acceptable limits of agreement. They further concluded that increased maternal BMI negatively impacts LV diastolic function measured using tissue Doppler imaging.
The authors however failed to demonstrate an unmet need for this technology in their literature review, and they failed to demonstrate literature to justify their use of echocardiography as a substitute for a “gold standard” measure of cardiac output.
The statistical tests applied were appropriate, however no confidence intervals were given for the limits of agreement arrived at by Bland-Altman analysis, and the comparator used here was the mean between NICOM values and echocardiography values, which introduces a bias favouring NICOM to both correlation and mean percentage error values. Furthermore, the authors did not state what they would consider an acceptable level of precision for the NICOM values obtained, although they reported a measure of this (the intraclass coefficient) in their results.
The attempt to assess the relationship between maternal characteristics and myocardial performance did not add to the understanding of NICOM as a technique to measure cardiac output in the obstetric population. There was no power calculation, and although it reported a correlation between BMI and decreasing LV diastolic function which was statistically significant, is a question that would have been better addressed by a separate study. This raises the possibility that it may represent “data-mining”.
Summary by Dr Fausto Morrell-Ducos
