The repeatability of transpulmonary thermodilution measurements

Introduction

Recently, less invasive cardiovascular monitoring with transpulmonary thermodilution with the PiCCO system using a central venous line and an arterial thermodilution catheter got increasingly popular. In the last issue of the Yearbook of Intensive Care and Emergency Medicine, the opinion was stated that a single cold saline injection is sufficient to adequately measure cardiac output and derived thermodilution parameters [1]. Since there are no reported investigations in the literature on the subject, we wanted to examine this hypothesis.

Methods

We retrospectively examined the data of the PiCCO system (Pulsion, Munich, Germany) from 18 patients treated in our neurosurgical intensive care unit. Neurosurgical diagnosis was mainly severe subarachnoid hemorrhage; 10 patients were additionally diagnosed with systemic inflammatory response syndrome, acute respiratory distress syndrome or cardiac failure.

Thermodilution measurements consisted of up to five single injections with a bolus of 20 cm³ iced saline. Data was automatically stored on a laptop connected to the PiCCO system. From a raw data volume of 200 MB, the thermodilution measurements were extracted and analysed for repeatability. An analysis of variance (ANOVA) was performed to quantify the disagreement between single measurements.

Results

A total of 417 thermodilution procedures consisting of 1465 single bolus injections were analyzed for indexed cardiac output (CI), intrathoracic blood volume (ITBI) and extrapulmonary lung water (ELWI). The median difference between the lowest and highest value in a single series was 0.3 l/min/m² (CI), 80 ml/m² (ITBI) and 1 ml/kg (ELWI), respectively. Calculated from the within-subject variance of the ANOVA, the 95% repeatability coefficient for two measurements of the same thermodilution sequence was 0,72 l/min/m² (CI), 270 ml/m² (ITBI) and 3.5 ml/kg (ELWI). This translates to 48% (CI), 180% (ITBI) and 87% (ELWI) of the measured parameters' normal ranges.

Discussion

According to our data, we disadvise the use of a single injection for the measurement of transpulmonary thermodilution. Especially for preload parameters, the mean of at least three repeated measurements reflects the patients status more appropriately. It remains a clinical judgement how much imprecision of measured data a clinician is willing to accept. This may vary in different situations and is, as a matter of principle, not solvable by statistics [2].