Agreement between peripheral venous and arterial lactate levels

Study objective: To test the hypothesis that measurements of peripheral venous lactate (V-LACT) can be substituted for arterial lactate (A-LACT) in predicting arterial hyperlactacidemia. Methods: We conducted a prospective comparison of paired A-LACT and V-LACT measurements obtained from a convenience sample of 74 ED patients who presented to an urban, public teaching hospital, 70% of whom had abnormal A-LACT. Results: Mean A-LACT and V-LACT were 2.8 mmol/L and 3.0 mmol/L, respectively. A-LACT and V-LACT were strongly correlated (r(2)=.89). Simultaneous multivariate adjustment for tourniquet time and for time elapsed between drawing of A-LACT and V-LACT had no effect on this correlation. Although the mean difference between V-LACT and A-LACT was only .22 mmol/L, the range that included 95% of the disagreement between paired measurements in individual patients was -1.3 mmol/L to 1.7 mmol/L. When A-LACT and V-LACT levels were each divided into normal and abnormal (elevated) groups, V-LACT showed 94% sensitivity (95% confidence interval [CI], 83% to 99%), 57% specificity (95% CI, 34% to 78%), a positive likelihood ratio of 2.2, and a negative likelihood ratio of .1. A-LACT values were used as the criterion standard for these calculations. Conclusion: Correlation between A-LACT and V-LACT was high in this cohort of patients, but agreement is imperfect. The odds of arterial hyperlactacidemia appear to be reduced substantially by the finding of a normal V-LACT but are only marginally increased if the V-LACT is increased. Caution should be used in the routine substitution of V-LACT for A-LACT.