Early lactate changes improve the outcome prediction for extracorporeal membrane oxygenation.

Abstract

Serial lactate (clearance) data are commonly used for risk stratification in patients receiving veno-arterial extracorporeal life support (ECLS). We retrospectively analysed 855 patients who had undergone ECLS due to cardiac (n = 578) and non-cardiac (n = 277) aetiologies between 2002 and 2013 at National Taiwan University Hospital. Serial lactate (clearance) data were collected before ECLS and at 8, 16, 24, 48 and 72 h after ECLS. To investigate the impact of lactate (clearance) levels on 180-day survival, we performed linear mixed model and joint model analyses using the Bayesian approach. Among the 855 patients, 564 (65.9%) patients died within 180 days after ECLS cannulation. The joint model showed that the effect of lactate on survival was null in both the reduced model and the fully adjusted model. However, an effect of lactate clearance on survival was observed in the reduced model [estimate 0.004; 95% confidence interval (CI) 0.002-0.006] and the fully adjusted model (estimate 0.003; 95% CI 0.001-0.005). In a further secondary analysis, lactate clearance (hazard ratio 0.861; 95% CI 0.813-0.931) at 16 h after ECLS cannulation was determined to be a risk factor for mortality. According to a receiver operating characteristic curve analysis, the SAVE score combined with lactate clearance (area under curve = 0.881) showed good outcome discrimination. Incorporating lactate clearance at 16 h after ECLS cannulation into the SAVE system improved the predictive value for mortality in patients receiving ECLS.