ABC | Volume 113, Nº2, August 2019

Original Article Coll et al Non-invasive cardiac output measurement Arq Bras Cardiol. 2019; 113(2):231-239 Table 1 – Parameters calculated by the Cheetah NICOM ® system Parameter Equation measuring unit Stroke Volume (SV) CO/HR x 1000 ml/beat Stroke Volume Index SV/BSA ml/m 2 /beat Cardiac Output (CO) HR x SV/1000 l/min Cardiac Index (CI) CO/BSA l/min/m 2 Mean arterial pressure (MAP) (SBP + (2 x DBP))/3 mmHg Total Peripheral Resistance 80 x (MAP)/CO dynes x sec/cm 5 Total Peripheral Resistance Index 80 x (MAP)/CI dynes x sec/cm 5 /m 2 HR: heart rate; BSA: body surface area; SBP: systolic blood pressure; DBP: diastolic blood pressure. Table 1 shows the parameters that were calculated by theCheetah NICOM®system. TheNICOM®systemuses four sensors applied to the right and left sides of the chest. Each sensor consists of an outer transmitting electrode and an inner receiving electrode. The outer electrodes transmit a low amplitude alternating electrical current with a frequency of 75 kHz to the thoracic cavity. The electrical properties of the thorax cyclically change due to the pulsatile volume of blood ejected from the heart. The pulsatile blood flow in the large thoracic arteries causes time delays (phase shifts) between the applied alternating electrical current and the thoracic voltage measured by the inner electrodes. Based on the measured phase shift the maximum aortic flow (dX/dtmax) and the ventricular ejection time (time from aortic valve opening to aortic valve closure, VET) were measured. Finally, the SV was obtained as SV=DX/DT×VET. Thereon, the CO, and finally the CPO, was derived. 43 The SV data were measured beat‑by‑beat and averaged over 60 s. Statistical analysis In a first step, participants were excluded from statistical analyses due to measurement errors (outliers), which were defined as ≥mean ± twofold pooled standard deviation. 44 The test-retest reliabilityof cardiopulmonaryandhemodynamic parameters was analyzed by (1) the difference inmeans to detect systematic bias, (2) intraclass correlation coefficients (ICC) to examine the relative reliability, and (3) typical error (TE) of measurements to quantify the absolute reliability. 45 To examine the difference in means, a progressive statistical approach using magnitude-based inferences for practical significance were computed. 46 Compared to traditional null-hypothesis testing, that is influenced by the sample size, magnitude-based inferences ground an analysis, how big the observed effect is, and if the effect is lower, similar, or higher than the smallest worthwhile difference (SWD). 46 Therefore, means and 90% confidence intervals (CIs) was computed first. Then, the disposition of the mean differences in relation to the SDWs were investigated. While the SDW for the maximal workload was calculated from the pooled standard deviation multiplied by 0.2, the SWD for all other physiological variables were calculated from the pooled standard deviations multiplied by 0.6, because it is well known that physiological variables showed a clearly higher spontaneous variability than biomechanical measures. 47 Finally, the likelihoods for test 2 showing “true” higher, similar, or lower values than test 1 were determined and qualitatively described using the following probabilistic scale: <1%, most unlikely ; 1 to <5%, very unlikely ; 5 to <25%, unlikely ; 25 to <75%, possibly ; 75 to <95%, likely ; 95 to<99%, very likely , and≥99%, most likely . If the likelihoods for having both higher and lower values were ≥5%, the differences were described as unclear . Otherwise, the differences were interpreted according to the observed likelihoods. To clarify the meaningfulness of the differences, standardized differences labeled as effect sizes (ESs) were calculated and interpreted accordingly: 0.2 to <0.6, small ; 0.6 to <1.2, moderate ; 1.2 to <2.0, large ; 2.0 to<4.0, very large ; and≥4.0, extreme large . To express the relative reliability, ICCS and 90%CIs were computed. The coefficients were described as follows: <0.20, very low ; 0.20 to <0.50, low ; 0.50 to <0.75, moderate ; 0.75 to <0.90, high ; 0.90 to <0.99, very high ; and ≥0.99, extremely high . To quantify the absolute reliability, TEs and 90% CIs were calculated. The meaningfulness of the TEs was expressed via standardization for which the aforementioned scale for standardized differences was applied. 47 The relationships between the CPO and measures of cardiac structure and function as well as traditional cardiopulmonary exercise parameters were investigated using Pearson correlation coefficients (r) that were interpreted accordingly: <0.1, trivial ; 0.1 to <0.3, small , 0.3 to <0.5, moderate ; 0.5 to <0.7, large ; 0.7 to <0.9, very large ; 0.9 to 1.0, almost perfect. 47 Lastly, common variances from coefficients of determinations (R 2 ) were computed. Thereby, a cutting-off value of 50% was defined to clarify, if two variables are dependent or independent from each other. 48 Results 25 participants completed both exercise tests. 17 participants (10 male, 7 female) were finally included. 8 participants were excluded due to measurement errors (outliers). Anthropometric, echocardiographic, and spiroergometric data of the participants are presented in Table 2. Reliability Data concerning systematic bias are presented in Table 3. It shows the differences in means between test 1 and test 2 for all hemodynamic and cardiopulmonary parameters measured at rest and during submaximal and peak exercise conditions. For all parameters, there were unclear to very likely trivial differences with small to moderate ESs (ES: 0.2-0.6). 233