The influence of prone positioning on the accuracy of calibrated and uncalibrated pulse contour-derived cardiac index measurements.

Abstract

Patients with lung failure who undergo prone positioning often receive extended hemodynamic monitoring. We investigated the influence of modified prone positioning (135°) on the accuracy of pulse contour-derived calibrated cardiac index (CIPC) and uncalibrated cardiac index (CIVIG) in this patient population with transpulmonary thermodilution (TPTD) as reference technique. We studied 16 critically ill and mechanically ventilated patients (11 men, 5 women, aged 20-71 years) with acute lung injury or acute respiratory distress syndrome. Patients were monitored by TPTD with an integrated calibrated pulse contour technique (PiCCO®) and by uncalibrated pulse contour analysis (FloTrac/Vigileo™). Before prone positioning, cardiac index (given in L·min(-1)·m(-2)) was measured by TPTD (CITPTD) and CIPC was calibrated. After positioning, CIPC and CIVIG were read from the monitor and CITPTD was measured. After 8 to 10 hours, prone positioning was completed and measurements were performed analogously. Bland-Altman analysis based on a random-effects model was used to calculate limits of agreement (LOA) and percentage errors. Polar plots were used for trend analysis. Supine CITPTD was 3.3 ± 0.9 (mean ± SD) and CIVIG was 3.1 ± 0.8. After proning, CIPC was 3.5 ± 0.8, CIVIG 3.3 ± 0.8, and CITPTD 3.6 ± 0.8. Before repositioning, CITPTD was 3.5 ± 0.7 and CIVIG 3.3 ± 1.0. After repositioning, CITPTD was 3.1 ± 0.7, CIPC 3.3 ± 0.7, and CIVIG 2.9 ± 0.6. Mean bias pooled for proning and repositioning was -0.1 (LOA -0.7 to 0.6) for CIPC (percentage error 19%) and 0.3 (LOA -1.3 to 1.9) for CIVIG (percentage error 48%). Changes in CI were too small for trending analysis. Although calibrated CI measurements are only marginally influenced by prone positioning, according to the criteria of Critchley and Critchley, uncalibrated CI values show a degree of error, too high to be considered clinically acceptable.