Background: The precision of bolus thermodilution cardiac output measurements in patients with atrial fibrillation (AF) has not previously been determined. A priori we suspected that the precision would be lower in patients with AF than in patients with sinus rhythm (SR). Consequently, we also determined if the precision could be improved by injecting the thermal indicator into the right ventricle instead of the right atrium.
Methods: Cardiac output was determined as the average result of four injections of 10 ml of iced saline. Replicate measurements were performed with thermal indicator injections into the right atrium and ventricle. The coefficients of variation and the precisions were calculated.
Results: In the 25 patients with AF, mean cardiac output was 3.96 l min(-1) (range 2.4-7.4), the coefficient of variation 0.073 (95% CI +/- 0.011), and the precision 0.38 l min(-1) (95% CI +/- 0.14) with injection into the right atrium. In the 25 patients with SR, mean cardiac output was 4.73 l min(-1) (range 2.4-7.3), the coefficient of variation 0.047(95% CI +/- 0.006), and the precision 0.38 l min(-1) (95% CI +/- 0.14). In both groups, an agreement analysis demonstrated that the injection of indicator into the right ventricle resulted in a significantly higher cardiac output [AF+0.25 (95% CI +/- 0.15) l min(-1), SR+0.29 ( +/- 0.20) l min(-1)].
Conclusion: The coefficient of variation for cardiac output determinations is 55% higher in patients with AF. Two measurements, separated by time or intervention, must differ by 15% in AF patients and 9% in SR patients before one can be 95% confident that a real change has taken place.