Evaluation of the non-calibrated pulse contour cardiac output monitor FloTrac/Vigileo against thermodilution in standing horses

Objective To evaluate the non-calibrated, minimally invasive cardiac output (CO) monitor FloTrac/Vigileo (FloTrac) against thermodilution (TD) CO in standing horses. Study design Prospective, experimental trial. Animals Nine adult horses weighing a median (range) of 535 (470-602) kg. Methods Catheters were placed in the right atrium, pulmonary artery and carotid artery under local anaesthesia. CO was measured 147 times by TD and FloTrac and indexed to body weight. Changes in CO were achieved with romifidine or xylazine and dobutamine constant rate infusions. Bland-Altman analysis, concordance and polar plot analysis were used to assess agreement and ability to track changes in CO. Results Mean +/- standard deviation COTD of 48 +/- 16 mL kg(-1) minute(-1) (range: 19-93 mL kg(-1) minute(-1)) and mean COFloTrac of 9 +/- 3 mL kg(-1) minute(-1) (range: 5-21 mL kg(-1) minute(-1)) were measured. Low agreement with a large mean bias of 39 mL kg(-1) minute(-1) and wide limits of agreement of 8-70 mL kg(-1) minute(-1) were found. The percentage error of 108% and precision of TD of +/- 18% resulted in an estimated precision of FloTrac of +/- 106%. Comparison of changes in COFloTrac with changes in COTD gave a concordance rate of 52% in the four-quadrant plot, and a mean polar angle of -11 degrees with radial limits of agreement of +/- 61 degrees in the polar plot. Mean arterial pressure (MAP) and COFloTrac were positively correlated (r = 0.5, p < 0.0001). No correlation of MAP with COTD was observed. Conclusions and clinical relevance The FloTrac system, originally designed for use in humans, neither measured absolute CO in standing horses accurately nor tracked relative changes in CO measured by TD correctly. The false dependence of COFloTrac on arterial blood pressure further discourages the use of this technique in horses.