Comparaison de deux techniques de monitorage du débit cardiaque : méthode du contour de pouls et thermodilution artérielle pulmonaire continue

Abstract

Two devices for continuous cardiac output (CO) monitoring at patient bedside were evaluated. Vigilance™ monitor (Baxter Lab.) provides CO measurements by intermittent bolus (TDCO Vig) and continuous thermodilution (CCO Vig) methods, curves are both collected using a pulmonary artery catheter ; PICCO™ monitor (Pulsion Lab.) also provides continuous CO by pulse contour analysis (CCO Pic) and intermittent trans-thoracic thermodilution CO measurements (TDCO Pic), which is used to calculate a calibration factor for further beat-to-beat estimations of CCO Pic. Both systemic arterial pressure and thermodilution curves are collected using a femoral arterial catheter. The objectives of the study were : 1) to compare both methods, 2) to identify the situations possibly responsible for a poor reliability of the pulse contour method. CO measurements ( n=301) were recorded simultaneously with both devices in 10 ICU patients : 44 conflicting data pairs, measured after vaso-active drugs initiation or following a calibration procedure performed during cardiac arythmia, were excluded. 257 data pairs for continuous CO measurements were compared using linear regressions and Bland-Altman plots method. Mean values for CCO Vig and CCO Pic were 7,3±2,63 and 7,8±3,04 l. min −1 respectively, and the correlation was significant ( r=0,94, p<10 −4). The average difference (bias) was 0,5± 1,06 l. min −1, i.e 7 % overestimation of CO by CCO Pic, and the 95 % confidence interval was large. For 9 among 10 patients the correlation between CCO Vig and CCO Pic was good (Spearman test). For CO values less than 8 l.min −1, the bias and 95 % confidence interval were smaller (0,2±0,66 l. min −1, n=180). Moreover, 108 data pairs for intermittent CO measurements were collected. The correlation between TDCO Vig (7,4±2,7 l. min −1) and TDCO Pic (8,1±1,06 l. min −1) was significant ( r=0,97, p<10 −4). The bias was 0,76±0,86 l. min −1 and could be related to a thermal loss between pulmonary and femoral arterial collection. On the whole, the agrement between both continuous CO measurments is acceptable. Since TDCO Pic is used for CCO Pic calibration, the difference between the two CO continuous methods could be partly explained by an overestimation of TDCO Pic. Consequently, the software calibration process should be refined. Moreover, we identified two situations that are responsible for erroneous values of CCO Pic and require a new calibration : cardiac arythmia during PICCO™ calibration, vaso-active drugs initiation.