Assessment of variables affecting the accuracy of extravascular lung water (EVLW) by single versus double indicator dilution techniques in ARDS patients

Introduction

Estimation of extravascular lung water (EVLW) by the single indicator dilution technique (SDT) implemented in the PICCO monitoring system relies on a relationship between intrathoracic blood volume (ITBV) and global end diastolic volume (GEDV), which has been derived from a large population of mixed critically ill patients using the double indicator dilution technique (thermal and indocyanine green dyes, DDT): ITBV_sdt = 1.25 × GEDV [1]. Since the difference between ITBV and GEDV corresponds to pulmonary blood volume (PBV), we can write: PBV_sdt = 0.25 × GEDV. Any factor influencing this relationship can affect SDT accuracy. The aims of our study were: 1) to compare SDT versus DDT for EVLW measurements in ARDS patients; and 2) to explore factors influencing the PBV and GEDV relationship, and affecting SDT accuracy.

Methods

We studied 21 ARDS patients (PaO₂/FiO₂ = 165 ± 65), monitored with a Swan–Ganz catheter, and a 4 F thermistore-tipped, fiberoptic catheter inserted through a femoral artery, both connected to a COLD monitoring system. SDT measurements were obtained by mean transit time and down slope time of thermal indicator using standard formula. Statistical analysis was performed by single and multiple linear regression analysis; accuracy was assessed according to Bland and Altman method.

Results

Agreement between EVLW_ddt and EVLW_sdt showed a bias of 15.3 ± 135.8, and a 95% confidence interval of -256 and 287; the correlation coefficient was r = 0.92 (P < 0.001). Correlations of GEDV and PBV_sdt with PBV_ddt, although significant, were poor (r = 0.394, P < 0.01). EVLW_ddt indexed to body weight (EVLWi) and cardiac index (CI) explained 64% of PBV indexed to body surface area (PBVi) variance (P < 0.001). EVLWi, CI, central venous pressure (CVP), and PaCO₂ explained 82% of PBVi to GEDVi ratio variance (P < 0.001). EVLW_ddt to EVLW_sdt differences were highly correlated with PBV_ddt (r = 0.88, P < 0.001). EVLWi, CI and PaCO₂ explained 65% of the differences between EVLW_ddt and EVLW_sdt variance (EVLW_ddt - EVLW_sdt = -263 + 18.64 × EVLWi - 50.9 × CI + 5.3 × PaCO₂, r² = 0.65, P < 0.001).

Conclusions

The SDT gives acceptable estimates of EVLW in ARDS patients. EVLWi, CI and PaCO₂ may influence the relationship between PBV and GEDV, and may affect accuracy of EVLW measured by SDT.