LAUSR

Introduction/Hypothesis: Pulmonary arterial and transpulmonary thermodilution (TPTD; PiCCO) are gold-standard techniques to derive cardiac index (CI). Due to their invasiveness they are limited to ICU and OR. Numerous less invasive techniques to estimate CI have been introduced. Most of these devices are based on pulse-contour-analysis (PCA). Although they are less invasive than indicator-dilution techniques, they still require expensive monitors and disposables. CapstesiaTM is an inexpensive smartphone-application to analyze arterial wave-forms from an invasive arterial line based on a snapshot of the arterial waveform on the original monitor. It provides estimates of CI, stroke volume index (SVI), pulse pressure variation (PPV) and dPmax. So far, only one study validated CI derived from Capstesia (CI_Cap) vs. an uncalibrated PCA-based estimate of CI by the FloTracTM. Therefore, we compared CI_Cap and SVI_Cap vs. the gold-standard values CI_PiCCOand SVI_ PiCCO derived from TPTD with PiCCO (Pulsion, Germany). Methods: Interim analysis after 39 simultaneous measurements of CI and SVI were performed in 11 ICU-patients (SOFA-score 13±5). Statistics: Wilcoxon-test; Bland-Altman-analysis; IBM SPSS 26. Results: CI_Cap and CI_PiCCO were not significantly different (3.2±1.3 vs. 3.7±1.7L/min/m; p=0.077). However, CI_Cap negatively (!) correlated with CI_PiC (r=-0.318; p=0.049) and was strongly imprecise (percentage error 137%; limits of agreement -5.29 and +4.21L/min/m). The bias (CI_Cap-CI_PiCCO) strongly correlated with CI_PiCCO (r=-0.871; p<0.001). This means that CI_Cap overestimates low values and understimates high values. Similar findings were made for SVI_Cap vs. SVI_PiCCO (39±18 vs. 43±15mL/m; p=0.125; bias -4±43mL/ m) which did not correlate (r=-0.158; p=0.336). SVI_Cap was strongly imprecise vs. SVI_PiCCO (percentage error 134%; limits of agreement -59 and +52mL/m). Conclusions: CI_Cap and SVI_Cap have moderate accuracy vs. CI_PiCCO and SVI_PiCCO. Yet, they are inacceptably imprecise vs. TPTD-derived values. This does not necessarily question the capstesia-technology, since imprecision is a well-known drawback of most other uncalibrated PCAestimates of CI. The poor performance regarding TPTD-derived parameters does not exclude an appropriate estimation of PCA-derived parameters (e.g. PPV; see CAPUCCINO-II).