Regional variation in gastric intramucosal pCO₂ may confound tonometric measurement: an evaluation using two tonometers and continuously recirculating gas tonometry

Background

Gastric tonometry is used to assess splanchnic perfusion in critically ill patients, and relies on measurement of the pCO₂ in the tonometer balloon. Saline is most frequently used as the CO₂ vehicle, but phosphate-buffered saline and continuously recirculating gas tonometry (CRGT) have been used in an attempt to improve the accuracy of the tonometric pCO₂ measurements. Comparisons between these methods using two tonometers placed together have been made in vitro and in vivo (animal and human), but the assumption that two closely situated tonometer balloons containing the same vehicle produce the same pCO₂ reading has not been tested, which questions the validity of these comparisons. Using a single technique (CRGT) we investigated the reproducibility of paired tonometric measurements and their precision both in the laboratory and in patients.

Method

The study comprised in vitro and in vivo components. An equilibration chamber containing 0.9% saline was kept at 37°C and a constant pCO₂ of either 4.5% or 9.5%, monitored using a Paratrend 7 probe (Biomedical Sensors, High Wycombe, UK). Two Tonocap monitors (Datex, Helsinki, Finland) were calibrated and connected to two tonometers (Tonometric Division, Instrumentarium Division Helsinki, Finland) within the chamber and, after a 30 min equilibration period, 24 paired pCO₂ measurements were recorded at 10 min intervals at each CO₂ concentration. 10 intubated, ventilated and sedated critically ill patients were then studied, using two tonometers positioned in the stomach and connected to the two tonocap monitors. After a 30 min equilibration period 12 paired pCO₂ measurements were recorded per individual at 10 min intervals.

Results

During the in vitro study the pCO₂ of the saline was maintained at two steady means of 4.52 (4.48–4.54) and 9.47 (9.37–9.5) kPa. There was excellent agreement between tonometers, with a mean bias of 0.16 kPa (95% confidence interval -0.09 to 0.058), and limits of agreement -0.51 to 0.47 kPa. Agreement between each tonometer and Paratrend pCO₂ measurements was also good with biases of 0.110 and 0.103 kPa (95% confidence intervals -0.210 to -0.028 and -0.194 to -0.012 kPa) and limits of agreement -0.54 to 0.23 and -0.52 to 0.32 kPa. Agreement between tonometers in vivo was poor. Although the mean bias was 0.132 kPa (95% confidence interval 0.286 to 0.418 kPa), the limits of agreement were -2.83 to 3.09 kPa.

Conclusion

The limits of agreement between two tonometers in the stomachs of critically ill patients using CRGT were unacceptably wide. This may be due to inherent flaws in the technique in vivo, or to regional differences in intraluminal pCO₂.