The DCO2 measured by gastric tonometry predicts survival in children receiving extracorporeal life support. Comparison with other hemodynamic and biochemical information. Royal Children's Hospital ECMO Nursing Team.

STUDY OBJECTIVE: To assess the role of gastric tonometry in monitoring children receiving extracorporeal life support (ECLS) and to determine if DCO2 or pHi in the weaning phase of ECLS predicts survival.

DESIGN: A prospective study of consecutive patients treated with ECLS.

SETTING: A tertiary pediatric ICU that is the ECLS referral center for Australia.

PATIENTS: Twenty consecutive children receiving ECLS for cardiovascular or respiratory failure.

INTERVENTIONS: All children were monitored throughout their ECLS course using a tonometer inserted into the stomach via the orogastric route. The PCO2 in the tonometer balloon was measured every 4 to 6 h and the pHi was calculated using the Henderson-Hasselbalch equation. The DCO2, which is the difference between PCO2 in tonometer saline solution and arterial blood, was calculated. We compared the ability of pHi, DCO2, heart rate, mean arterial pressure, arterial pH, base deficit, and blood lactate to predict death or survival during the weaning phase. Measurements were taken on the lowest level of support, which for veno-arterial extracorporeal membrane oxygenation and ventricular assist device was defined as the lowest ECLS pump flows, and on veno-venous extracorporeal membrane oxygenation was defined as the time of lowest ECLS gas flow. Predictive power was assessed using the receivor operating characteristic (ROC) analysis on the data collected at these times.

RESULTS: In the weaning phase of ECLS, the pHi was significantly lower in children who died (pHi = 7.21; 95% confidence intervals, 7.14 to 7.28) than in those who survived (pHi = 7.38; 95% confidence intervals, 7.28 to 7.47). The DCO2 was significantly higher in children who died (23.6 mm Hg; 95% confidence intervals, 14.3 to 33.1) compared with survivors (4.7 mm Hg; 95% confidence intervals, -0.78 to 10.1). The area under the ROC curve was 0.95 for DCO2 (and 0.88 for pHi). pHi and DCO2 predicted survival better than base deficit (area under ROC curve, 0.82), blood lactate level (0.29), arterial pH (0.65), heart rate (0.62), and mean arterial pressure (0.74).

CONCLUSIONS: DCO2 is a clinically meaningful measurement in children receiving ECLS. A high DCO2 was a good predictor of death in this series. Gastric tonometry may provide a useful measure of the adequacy of regional perfusion and oxygenation in this group of patients.