Clinical validation of a computerized algorithm to determine mean systemic filling pressure

Abstract

Mean systemic filling pressure (Pms) is a promising parameter in determining intravascular fluid status. Pms derived from venous return curves during inspiratory holds with incremental airway pressures (Pms-Insp) estimates Pms reliably but is labor-intensive. A computerized algorithm to calculate Pms (Pmsa) at the bedside has been proposed. In previous studies Pmsa and Pms-Insp correlated well but with considerable bias. This observational study was performed to validate Pmsa with Pms-Insp in cardiac surgery patients. Cardiac output, right atrial pressure and mean arterial pressure were prospectively recorded to calculate Pmsa using a bedside monitor. Pms-Insp was calculated offline after performing inspiratory holds. Intraclass-correlation coefficient (ICC) and assessment of agreement were used to compare Pmsa with Pms-Insp. Bias, coefficient of variance (COV), precision and limits of agreement (LOA) were calculated. Proportional bias was assessed with linear regression. A high degree of inter-method reliability was found between Pmsa and Pms-Insp (ICC 0.89; 95%CI 0.72-0.96, p = 0.01) in 18 patients. Pmsa and Pms-Insp differed not significantly (11.9 mmHg, IQR 9.8-13.4 vs. 12.7 mmHg, IQR 10.5-14.4, p = 0.38). Bias was -0.502 ± 1.90 mmHg (p = 0.277). COV was 4% with LOA -4.22 - 3.22 mmHg without proportional bias. Conversion coefficient Pmsa ➔ Pms-Insp was 0.94. This assessment of agreement demonstrates that the measures Pms-Insp and the computerized Pmsa-algorithm are interchangeable (bias -0.502 ± 1.90 mmHg with conversion coefficient 0.94). The choice of Pmsa is straightforward, it is non-interventional and available continuously at the bedside in contrast to Pms-Insp which is interventional and calculated off-line. Further studies should be performed to determine the place of Pmsa in the circulatory management of critically ill patients. ( www.clinicaltrials.gov ; TRN NCT04202432, release date 16-12-2019; retrospectively registered).Clinical Trial Registration www.ClinicalTrials.gov , TRN: NCT04202432, initial release date 16-12-2019 (retrospectively registered).

Keywords: Cardiac output; Inspiratory hold; Mean systemic filling pressure; Right atrial pressure; Venous return.

Fig. 1

Example of venous return curve. Data points plotted represent consecutive cardiac index (CI) (y-axis) and corresponding right atrial pressure (RAP; x-axis) values during 12 s inspiratory hold maneuvers. With each increment of airway plateau pressure (Pplateau), CI (or venous return; VR; as in steady state conditions VR determines CI) will decrease, whereas RAP will increase. Pms-Insp (Pms calculated after inspiratory hold maneuver) is calculated by extrapolation of the VR-curve with linear regression (least squares method). The intersect of the VR-curve with the x-axis (at zero CI or VR) represents true Pms-Insp

Fig. 2

Intraclass correlation of Pmsa and Pms-Insp. Association between realtime Pms calculated by computerized algorithm (Pmsa) and Pms calculated after inspiratory hold maneuver (Pms-Insp). Intraclass correlation coefficient (ICC) is presented in the lower right corner of the scatter plot (95% CI 0.72–0.96, p ≤ 0.01)

Fig. 3

Bland-Altman analysis of Pmsa and Pms-Insp. Bland-Altman analysis showing the comparison between measurements of realtime Pms calculated by computerized algorithm (Pmsa; test-method) and Pms calculated from venous return curves during inspiratory hold maneuvers (Pms-Insp; reference method). The dashed horizontal line represents the mean of the differences (bias) which was found to be −0.502 ± 1.90 mmHg, p = 0.277. The upper and lower dotted horizontal lines represent the 95% limits of agreement (LOA) which are −4.22 and 3.22 mmHg