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 mLkg-1 minute-1 ) and mean COF loTrac of 9 ± 3 mLkg-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 COF loTrac with changes in COTD gave a concordance rate of 52% in the four-quadrant plot, and a mean polar angle of -11° with radial limits of agreement of ±61 ° in the polar plot. Mean arterial pressure (MAP) and COF loTrac 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 COF loTrac on arterial blood pressure further discourages the use of this technique in horses.