Introduction
Transpulmonary thermodilution (TPTD) is increasingly used in the clinical area [
1], but its precision for measuring CI and the number of cold boluses that must be replicated is a matter of debate. Indeed, a recent study concluded that calculating the mean of two TPTD measurements was enough for reaching an acceptable level of precision [
2] but it included a limited number of patients. Additionally, for another transpulmonary dilution technique using the lithium and not the cold dilution, it has been recently shown that at least three dilution measurements were required for reaching an acceptable precision. In addition to CI, TPTD also allows estimating the global end-diastolic volume (GEDV, that is, the volume of the cardiac cavities at end-diastole) and the extravascular lung water (EVLW, that is, the volume of the pulmonary edema) [
3]. The precision of EVLW measurement derived from TPTD has been reported by some studies [
2,
4‐
6], but again, the number of cold injectates that is needed for obtaining an acceptable reproducibility of the measurements remains to be determined. As for GEDV, its precision has been investigated in one study only [
2].
Thus, we attempted to answer the important practical question to know the number of thermal injections that must be performed for assessing CI, GEDV and EVLW with an acceptable precision. In particular, we evaluated the number of thermodilution measurements that must be replicated for detecting changes in CI, GEDV and EVLW ≥15% with an acceptable confidence. We also analyzed the factors influencing the precision of the measurements.
Discussion
This study shows that three thermodilution measurements are required for estimating CI, GEDVi and EVLWi through TPTD with an acceptable level of precision. If three thermodilution measurements are averaged, the technique allows detecting changes in CI, GEDVi and EVLWi of more than 15% with an acceptable confidence.
There have been a number of studies reporting a good agreement between the measurement of CI by TPTD and by bolus pulmonary artery thermodilution [
10‐
16] or by the Fick method [
17,
18]. By showing bias and limits of agreement of the TPTD CI compared to the reference CI, these previous studies investigated the technique accuracy [
1], that is, its ability to give a value of CI that is close to the reference value [
9]. Nevertheless, validation of a technique monitoring CI should not only be based upon accuracy. Another key criterion to consider is precision, that is, the ability of the technique to provide values of CI that are close to each other [
8]. Indeed, precision is a very important criterion for at least two reasons. First, it determines the least change that can be trusted with confidence as being significant in clinical practice. This is especially important for techniques measuring CI [
19], since one is much more interested in variations of CI values over time than to a given CI value. Second, since precision increases with the number of measurements, it conditions the minimal number of measurements that must be replicated in clinical practice.
Our result concerning the precision of the TPTD measurement of CI is in accordance with two previous studies [
20,
21] but in discrepancy with another one [
2]. In particular, we found that averaging two measurements only might be insufficient for reaching an acceptable precision and that three thermodilution measurements are actually needed. However, the latter study [
2] included a much more limited number of patients than our study. Interestingly, the precision of CI measurement we found for TPTD was similar to the precision reported by Cecconi
et al. for another transpulmonary dilution device, the LiDCO system [
22]. As the PiCCO device, this technique uses transpulmonary dilution but it uses lithium rather than cold saline as a diluted indicator. The similarity of precision between the two techniques suggests that their precision is more influenced by the transpulmonary dilution technique itself than the nature of the indicator.
We observed that, provided that three thermodilution measurements were averaged, the precision of the TPTD measurement of CI was within the 10% limit that is usually admitted as being desirable [
8]. Comparison with the precision of the bolus pulmonary artery thermodilution is somewhat difficult, since the latter was evaluated by only a few studies. In a 20-year-old study, Stetz
et al. [
23] found that precision of pulmonary artery catheter was similar to that we report with TPTD. More recently, Nilsson
et al. reported a precision as low as 6% when averaging only three pulmonary artery thermodilutions [
24]. If confirmed, these results would suggest that pulmonary artery thermodilution is more precise than TPTD.
The precision was almost similar for CI, GEDVi and EVLWi. This is not surprising since their measurement by TPTD share some common components. We could not analyze some constituents on which the measurement of CI, GEDVi and EVLWi is based, such as the mean transit time and down slope time of the thermodilution curve, what must be considered as a limitation of our study. Importantly, we found that the precision of TPTD measurements was not different when patients with cardiac arrhythmias, continuous veno-venous hemofiltration or spontaneous breathing were excluded from analysis. The absence of significant correlation between the dose of norepinephrine and the precision of CI, GEDVi and EVLWi suggests that vasopressors also do not influence the precision of TPTD measurements.
The good precision of EVLWi and GEDVi measurements is important as an increasing number of studies have proposed to consider EVLW as a prognostic factor [
4,
5,
25] or as a criterion of judgement for therapy [
6] in acute respiratory distress syndrome, even though concerns have been raised concerning its reliability in this condition [
26]. The results of the present study might have some important implications. The first ones are for daily clinical practice. By showing that injecting three cold boluses is sufficient for obtaining an acceptable precision of the technique, the study supports the manufacturer's guidelines of averaging three thermodilution measurements [
7], a recommendation that, until now, was not based upon published data. If one considers that a precision of 8% is sufficient for CI, GEDVi and EVLWi, what is highly reasonable [
27], repeating more than three boluses is useless. This might be important when considering that repeating cold injections is time consuming and that thermodilution must be frequently repeated in case of hemodynamic instability [
28]. The second implication of the results is for clinical research purposes. For instance, it indicates that if three thermodilution measurements are averaged, a change in CI, GEDVi or EVLWi of 15% or more can be considered as the cut-off defining a positive response to a therapeutic intervention [
21,
29,
30].
We acknowledge some limitations to our study. First, we did not exclude any measurement from analysis, while less skilled operators should do it when obtaining less reliable thermodilution curves. Second, the measurements were performed by only one skilled operator, precluding the assessment of inter-observer variability. Third, we did not investigate whether injecting more than 15 mL for performing each bolus could increase the precision of the technique. Fourth, the body temperature, CI, GEDVi and EVLWi values were within the normal range, such that the precision of TPTD for extreme values of these variables remains to be determined. In addition, we did not include patients with slow atrial fibrillation, which may affect the precision of TPTD measurements. Also, ideally, the measurements should have consisted of a series of five boluses compared to another series of five boluses for which one to five are randomly selected. Finally the precision of TPTD when using a humeral arterial catheter remains to be determined.
Competing interests
Profs. Jean-Louis Teboul and Xavier Monnet are members of the Medical Advisory Board of Pulsion Medical Systems. As consultants for this company, they received honoraria. The company did not finance the manuscript. The company was not involved in any part of the conception or performance of the study. The other authors declare that they have no conflict of interest.
Authors' contributions
XM conceived the study, performed analysis and interpretation of the data, and drafted the manuscript. RP performed the collection of data, contributed to analysis and interpretation of the data and helped draft the manuscript. MK performed the collection of data, contributed to analysis and interpretation of the data, and helped draft the manuscript. MJ contributed to the collection of data, CR participated in the design of the study, contributed to analysis and interpretation of the data and helped draft the manuscript. J-LT conceived the study, participated in its design, contributed to analysis and interpretation of the data and helped draft the manuscript. All authors read and approved the final manuscript.