Secondary Logo

Journal Logo

Cardiovascular anaesthesia

The influence of haemodialysis on haemodynamic measurements using transpulmonary thermodilution in patients with septic shock: an observational study

Pathil, Anita; Stremmel, Wolfgang; Schwenger, Vedat; Eisenbach, Christoph

Author Information
European Journal of Anaesthesiology: January 2013 - Volume 30 - Issue 1 - p 16-20
doi: 10.1097/EJA.0b013e328358543a


This article is accompanied by the following Invited Commentary:

McGrath BA, Columb MO. Thermodilution cardiac output during haemodialysis: what are we measuring? Eur J Anaesthesiol 2013; 30:7–8.


The measurement of flow-based haemodynamic variables such as cardiac index (CI) has become increasingly important to guide and optimise therapeutic strategies in critically ill patients in the ICU. Pulmonary artery thermodilution was commonly regarded as the clinical ‘gold standard’ in the past,1 but less invasive techniques with comparable accuracy such as thermodilution by the PiCCO system have emerged as promising, alternative tools for haemodynamic monitoring.2–4 In addition to measurement of CI, the PiCCO technique provides additional information on volumetric variables such as global end-diastolic volume index (GEDVI) and extravascular lung water index (EVLWI), which are useful for determination of volume preload, especially in patients with septic shock.5 Earlier studies have defined GEDVI of 640 and 800 ml m−2 as the lower and upper limits of normovolaemia.6,7 Furthermore, EVLWI more than 10 ml kg−1 has been reported to serve as an indicator of fluid overload and pulmonary oedema.8,9 Thus, volume depletion can be assumed in patients with GEDVI less than 640 ml m−2 and EVLWI less than 10 ml kg−1, whereas volume overload is identified in patients with EVLWI more than 10 ml kg−1.

GEDVI and EVLWI are estimated by analysis of the mean transit time and the downslope time of the thermodilution curve, whereas CI is determined by the Stewart–Hamilton method. To compensate for individual differences in terms of compliance and resistance of the arterial system or changing clinical conditions, periodic manual calibration of the PiCCO system using transpulmonary thermodilution is required. Stable conditions during the measurements are necessary for correct calibration of the PiCCO system and accurate determination of CI, GEDVI and EVLWI. Therefore, in certain conditions, fluctuations in blood flow or temperature may disturb the processing of the thermodilution signal and result in altered measurements of haemodynamic variables.10,11

In this prospective study, we enrolled patients with septic shock who were receiving haemodynamic monitoring by the PiCCO system, and intermittent haemodialysis. The aim was to determine the effect of haemodialysis on the measurement of haemodynamic variables by transpulmonary thermodilution.


The study protocol (S-259/2009) was approved by the Ethics Committee of the University of Heidelberg (chairperson Professor Dr T. Strowitzki) on 19 October 2009 and written informed assent was obtained from the legal representative of each patient. For this prospective observational study, we included 30 patients from an internal medicine ICU at the University of Heidelberg with septic shock according to the definition of the International Sepsis Definitions Conference in 2001 (i.e. persistent arterial hypotension despite adequate fluid resuscitation unexplained by other reasons)12 and acute renal failure (i.e. oliguria <0.5 ml kg−1 h−1 and/or increase in serum creatinine of more than 27 μmol l−1 or percentage increase in serum creatinine >50%) with the need for intermittent haemodialysis. All patients received mechanical ventilation and inotropic therapy (e.g. norepinephrine, epinephrine, dobutamine). Ventilator settings, and infusion rates of inotropes and fluids, were determined by the attending physician according to clinical requirements. All patients had a central venous catheter and a 5 FG thermistor-tipped arterial catheter (Pulsion Medical Systems, Germany), which was inserted through the femoral artery for haemodynamic monitoring. Haemodialysis was performed as slow extended daily dialysis (SLEDD; Genius, Fresenius Medical Care, Germany)13 using a three-lumen 12 FG central venous catheter inserted via the internal jugular, subclavian or femoral vein.

Haemodynamic measurements

Transpulmonary thermodilution was performed every 8 h using the PiCCO technology (Infinity, PiCCO SmartPod, Dräger, Germany). CI (reference range 3–5 l min−1 m−2), GEDVI (reference range 680–900 ml m−2) and EVLWI (reference range 3–7 ml kg−1) were measured by transpulmonary thermodilution with triplicate injections of 15 ml of ice-cold saline 0.9% through a separate central venous catheter (i.e. not used for haemodialysis) inserted via the internal jugular, subclavian or femoral vein as described elsewhere.3,14 Triplicate measurements of CI, GEDVI and EVLWI were performed during periods of haemodynamic stability off haemodialysis and immediately after the patient started haemodialysis (CIHD, GEDVIHD and EVLWIHD). All measurements were obtained within a maximum time period of 15 min in order to compare haemodynamic variables measured by the PiCCO system before and during haemodialysis. Ventilator settings, administration of fluids and inotrope infusion rates were kept constant between measurements.

Statistical analysis

Data are presented as mean (SD). Paired Student's t-test was used to determine the significance of differences of haemodynamic measurements before and during haemodialysis. Linear regression analysis and Pearson correlation coefficients were used for analysis of data pairs. A two-tailed P value less than 0.05 was considered significant. Statistical analysis was performed with IBM SPSS statistics version 19 (IBM Corporation, Armonk, New York, USA).


We analysed 60 measurements (30 data pairs) in 30 patients, with each being the mean of triplicate measurements. The mean age of the patients was 56.7 (SD 11.4, range 26–79) years (Table 1).

Table 1
Table 1:
Individual ages and haemodynamic variables

The mean CI was 4.71 (SD 1.63, range 2.00–9.16) l min−1 m−2 and mean CI determined during haemodialysis (CIHD) was 4.18 (SD 1.38, range 1.05–7.44) l min−1 m−2. The difference between CI and CIHD was significant (P < 0.01) at −0.54 [SD 0.70, 95% confidence interval (CI) −0.80 to −0.28] l min−1 m−2 (Fig. 1a). The correlation between CI and CIHD was also significant (r = 0.91, P < 0.01; Fig. 1b).


The mean GEDVI was 864.8 (SD 173.7, range 500–1128) ml m−2 and mean GEDVI with haemodialysis (GEDVIHD) was 775.3 (SD 220.1, range 400–1593) ml m−2. The difference between GEDVI and GEDVIHD was significant (P = 0.02) at −89.5 (SD 191.8, 95%CI −161.2 to −17.9) ml m−2 (Fig. 1c). GEDVI and GEDVIHD were also significantly correlated (r = 0.55, P < 0.01; Fig. 1d).

The mean EVLWI was 10.3 (SD 4.2, range 2–20) ml kg−1 and mean EVLWI during haemodialysis (EVLWIHD) was 10.0 (SD 4.5, range 2–22) ml kg−1. EVLWI and EVLWIHD did not differ significantly (P = 0.42) at −0.3 (SD 2.0, 95% CI −1.1 to 0.5) ml kg−1 (Fig. 1e). The correlation between EVLWI and EVLWIHD was significant (r = 0.90, P < 0.01; Fig. 1f).


Although our results showed significant correlations between measurements before and during haemodialysis for all variables, statistically significant differences were observed for CI and GEDVI in our patients with septic shock, with lower values for both variables during haemodialysis. Although the 95% CI of the mean difference for CI measurements is narrow (−0.80 to −0.28 l min−1 m−2) and possibly clinically acceptable, within-patient differences are wider. Given a mean difference of −0.54 and a SD of 0.7 l min−1 m−2 in our cohort, these imply that 95% of differences occur within a range from −1.94 to 0.86 l min−1 m−2 in a situation where 2.0 represents the span of the reference range. Although extreme changes in classification or diagnosis (cardiogenic shock or hyperdynamic state) are unlikely between the two states, opposing marginal reclassifications from within the reference range are possible and vice versa. The significant correlations and the significant small mean differences in these variables suggest that they may be used to follow trends, but the range of within-patient differences suggests that actual values should not be relied on absolutely.

To date, there is limited information about potential effects of haemodialysis on PiCCO monitoring, a clinically relevant situation because patients with septic shock frequently present with the need for intermittent renal replacement therapy. PiCCO measurements have been used to document cardiovascular stability in patients undergoing SLEDD.15 However, the authors did not report whether or not they stopped the treatment for the PiCCO measurements, which, based on the presented data, might have made a difference. It may also be hypothesised that increased recirculation due to concomitant haemodialysis results in an overestimation of the mean transit time and an underestimation of the downslope time. This could lead to an overestimation of GEDVIHD and an underestimation of EVLWIHD. As GEDVIHD in our cohort was significantly underestimated, recirculation does not seem to have been a major influence.

A case report has suggested that thermodilution during haemodialysis produced measurements which resulted in a smaller calculated CI.11 The authors hypothesised that turbulences in the blood flow due to the haemodialysis catheter were causative and concluded that the pump must be stopped for the period of calibration. Because our data revealed significant underestimation in the measurement of CIHD compared to CI, we can confirm this observation. Processing of the thermodilution signals can be compromised further by fluctuations in blood temperature. Impairment of cardiac output measurements because of ‘thermal noise’ have been reported previously for the pulmonary artery thermodilution method.16,17 and were explained by respiratory variations in pulmonary artery blood temperature. As haemodialysis also alters average blood temperature, this could be a reasonable explanation for the slightly lower CI estimates that were also observed in our study.18 It is also possible that because the paired measurements could not be taken simultaneously and were separated by a short interval, the differences in measurements may represent genuine changes in haemodynamic state or changes due to haemodialysis.

Although not statistically significant, haemodialysis reduced some EVLWI values in a discrete but potentially clinically relevant manner. EVLWI serves as a measure for fluid overloading and lung oedema8,9 and a threshold of more than 10 ml kg−1 is commonly used to guide goal-directed volume therapy. In the current study, three of the 30 patients had lower EVLWI values which might have altered decisions on fluid management (Table 1). With a mean difference of −0.3 and a SD of 2.0 ml kg−1, these imply that 95% of differences occur within a range from −4.3 to 3.7 ml kg−1 in a situation where 4.0 ml kg−1 represents the reference range, or 3.0 ml kg−1 if less than 7.0 to more than 10 is taken as the clinically relevant range. Here, extreme changes in classification or diagnosis are possible. Therefore, measurements of EVLWI during haemodialysis should be interpreted with caution and determination of EVLWI without haemodialysis or additional volumetric information using transthoracic or transoesophageal echocardiography should be considered.

In conclusion, although our results showed significant correlations for CI, GEDVI and EVLWI with and without haemodialysis, we found that haemodialysis did significantly reduce CI and the cardiac preload marker GEDVI, but not EVLWI, when measured by the PiCCO system in patients with septic shock. Although differences are small, the variability of within-patient differences may be clinically important and care should be taken in relying solely on such measurements.


Assistance with the study. The authors are grateful to Tom Bruckner (Department of Medical Biometrics, University of Heidelberg) for assistance with statistical analysis.

Financial support and sponsorship: this work was supported by the Department of Gastroenterology, University of Heidelberg, Germany.

Conflicts of interest: none declared.


1. Swan HJ, Ganz W, Forrester J, et al. Catheterization of the heart in man with use of a flow-directed balloon-tipped catheter. N Engl J Med 1970; 283:447–451.
2. Della Rocca G, Costa MG, Pompei L, et al. Continuous and intermittent cardiac output measurement: pulmonary artery catheter versus aortic transpulmonary technique. Br J Anaesth 2002; 88:350–356.
3. Godje O, Hoke K, Goetz AE, et al. Reliability of a new algorithm for continuous cardiac output determination by pulse-contour analysis during hemodynamic instability. Crit Care Med 2002; 30:52–58.
4. Gondos T, Marjanek Z, Kisvarga Z, Halasz G. Precision of transpulmonary thermodilution: how many measurements are necessary? Eur J Anaesthesiol 2009; 26:508–512.
5. Michard F, Alaya S, Zarka V, et al. Global end-diastolic volume as an indicator of cardiac preload in patients with septic shock. Chest 2003; 124:1900–1908.
6. Combes A, Berneau JB, Luyt CE, Trouillet JL. Estimation of left ventricular systolic function by single transpulmonary thermodilution. Intensive Care Med 2004; 30:1377–1383.
7. Reuter DA, Kirchner A, Felbinger TW, et al. Usefulness of left ventricular stroke volume variation to assess fluid responsiveness in patients with reduced cardiac function. Crit Care Med 2003; 31:1399–1404.
8. Fernandez-Mondejar E, Castano-Perez J, Rivera-Fernandez R, et al. Quantification of lung water by transpulmonary thermodilution in normal and edematous lung. J Crit Care 2003; 18:253–258.
9. Sakka SG, Klein M, Reinhart K, Meier-Hellmann A. Prognostic value of extravascular lung water in critically ill patients. Chest 2002; 122:2080–2086.
10. Sami A, Rochdil N, Hatem K, Salah BL. PiCCO monitoring accuracy in low body temperature. Am J Emerg Med 2007; 25:845–846.
11. Martinez-Simon A, Monedero P, Cacho-Asenjo E. Erroneous measurement of haemodynamic parameters by PiCCO monitor in a critically ill patient with renal replacement therapy: a case report. Crit Care 2006; 10:410.
12. Levy MM, Fink MP, Marshall JC, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 2003; 31:1250–1256.
13. Fliser D, Kielstein JT. Technology insight: treatment of renal failure in the intensive care unit with extended dialysis. Nat Clin Pract Nephrol 2006; 2:32–39.
14. Morgan P, Al-Subaie N, Rhodes A. Minimally invasive cardiac output monitoring. Curr Opin Crit Care 2008; 14:322–326.
15. Kielstein JT, Kretschmer U, Ernst T, et al. Efficacy and cardiovascular tolerability of extended dialysis in critically ill patients: a randomized controlled study. Am J Kidney Dis 2004; 43:342–349.
16. Johnson RW, Normann RA. Central venous blood temperature fluctuations and thermodilution signal processing in dogs. Ann Biomed Eng 1989; 17:657–669.
17. Latson TW, Whitten CW, O’Flaherty D. Ventilation, thermal noise, and errors in cardiac output measurements after cardiopulmonary bypass. Anesthesiology 1993; 79:1233–1243.
18. Sakka SG, Hanusch T, Thuemer O, Wegscheider K. The influence of venovenous renal replacement therapy on measurements by the transpulmonary thermodilution technique. Anesth Analg 2007; 105:1079–1082.

cardiac index; extravascular lung water index; global end-diastolic volume index; haemodialysis; PiCCO

© 2013 European Society of Anaesthesiology