Modelflow underestimates cardiac output in heat-stressed individuals

Abstract

An estimation of cardiac output can be obtained from arterial pressure waveforms using the Modelflow method. However, whether the assumptions associated with Modelflow calculations are accurate during whole body heating is unknown. This project tested the hypothesis that cardiac output obtained via Modelflow accurately tracks thermodilution-derived cardiac outputs during whole body heat stress. Acute changes of cardiac output were accomplished via lower-body negative pressure (LBNP) during normothermic and heat-stressed conditions. In nine healthy normotensive subjects, arterial pressure was measured via brachial artery cannulation and the volume-clamp method of the Finometer. Cardiac output was estimated from both pressure waveforms using the Modeflow method. In normothermic conditions, cardiac outputs estimated via Modelflow (arterial cannulation: 6.1 ± 1.0 l/min; Finometer 6.3 ± 1.3 l/min) were similar with cardiac outputs measured by thermodilution (6.4 ± 0.8 l/min). The subsequent reduction in cardiac output during LBNP was also similar among these methods. Whole body heat stress elevated internal temperature from 36.6 ± 0.3 to 37.8 ± 0.4°C and increased cardiac output from 6.4 ± 0.8 to 10.9 ± 2.0 l/min when evaluated with thermodilution (P < 0.001). However, the increase in cardiac output estimated from the Modelflow method for both arterial cannulation (2.3 ± 1.1 l/min) and Finometer (1.5 ± 1.2 l/min) was attenuated compared with thermodilution (4.5 ± 1.4 l/min, both P < 0.01). Finally, the reduction in cardiac output during LBNP while heat stressed was significantly attenuated for both Modelflow methods (cannulation: -1.8 ± 1.2 l/min, Finometer: -1.5 ± 0.9 l/min) compared with thermodilution (-3.8 ± 1.19 l/min). These results demonstrate that the Modelflow method, regardless of Finometer or direct arterial waveforms, underestimates cardiac output during heat stress and during subsequent reductions in cardiac output via LBNP.

Fig. 1.

Bland-Altman plots comparing cardiac output measured by thermodilution and Modelflow from arterial cannulation (top) and Finometer (bottom) before and during lower-body negative pressure (LBNP) in normothermic subjects. The plots indicate relatively good agreement between the methods regardless of the perturbation (before and during LBNP) and the source of the arterial waveform (i.e., catheterization or Finometer). Black lines depict mean bias and ± 2 SD for the pre-LBNP period, whereas the gray lines depict these values during LBNP.

Fig. 2.

Reductions in cardiac output due to LBNP in normothermic subjects. Before the application of LBNP, while normothermic, cardiac outputs were similar between devices (values at the top of each bar). Likewise, the subsequent reductions in cardiac output to LBNP were similar between the three approaches.

Fig. 3.

Increases in cardiac output due to whole body heating. Heat stress increased cardiac output measured from all three approaches. However, the magnitude of this increase as estimated by Modelflow was significantly less, regardless of the origin of the arterial waveform, relative to thermodilution-derived cardiac outputs. Values above each bar depict the cardiac output for that device before the heat stress. *Significant difference from either Modelflow method.

Fig. 4.

Reductions in cardiac output due to LBNP while subjects were heat stressed. Before the application of LBNP, Modelflow cardiac outputs were significantly lower than thermodilution cardiac outputs (values at the top of each bar). The magnitude of the reduction in cardiac output due to LBNP was significantly smaller when evaluated via Modelflow, regardless of waveform source, compared with thermodilution. *Significantly different relative to either Modelflow method during LBNP. †Significantly different from thermodilution method for the pre-LBNP period.

Fig. 5.

Bland-Altman plots comparing cardiac output measured by thermodilution and Modelflow methods from direct cannulation (top) and Finometer (bottom) before and during LBNP in the heat-stressed condition. Heat stress alone caused a large average bias between thermodilution- and Modelflow-based measures of cardiac output. This bias was reduced during LBNP; however, this occurred primarily because of a large reduction in thermodilution-derived cardiac outputs coupled with a small reduction in Modelflow-derived cardiac outputs (see Fig. 4). Black lines depict mean bias and ± 2 SD for the pre-LBNP period, whereas the gray lines depict these values during LBNP.