RELIABILITY OF ANION GAP AS AN INDICATOR OF BLOOD LACTATE IN CRITICALLY-ILL PATIENTS

Objective: To evaluate the sensitivity, specificity, and predictive va lues of an elevated anion gap as an indicator of hyperlactatemia and t o assess the contribution of blood lactate to the serum anion gap in c ritically ill patients. Design: Prospective study. Setting: General in tensive care unit of a university hospital. Patients: 498 patients, no ne with ketonuria, severe renal failure or aspirin, glycol, or methano l intoxication. Measurements and results: The anion gap was calculated as [Na+] - [Cl-] - [TCO2]. Hyperlactatemia was defined as a blood lac tate concentration above 2.5 mmol/l. The mean blood lactate concentrat ion was 3.7 +/- 3.2 mmol/l and the mean serum anion gap was 14.3 +/- 4 .2 mEq/l. The sensitivity of an elevated anion gap to reveal hyperlact atemia was only 44 % [95 % confidence interval (CI) 38 to 50], whereas specificity was 91 % (CI 87 to 94) and the positive predictive value was 86 % (CI 79 to 90). As expected, the poor sensitivity of the anion gap increased with the lactate threshold value, whereas the specifici ty decreased [for a blood lactate cut-off of 5 mmol/l: sensitivity 67 % (CI 58 to 75) and specificity = 83 % (CI 79 to 87)]. The correlation between the serum anion gap and blood lactate was broad (r(2) = 0.41, p < 0.001) and the slope of this relationship (0.48 +/- 0.026) was le ss than 1 (p < 0.001). The serum chloride concentration in patients wi th a normal anion gap (99.1 +/- 6.9 mmol/l) was comparable to that in patients with an elevated anion gap (98.8 +/- 7.1 mmol/l).Conclusions: An elevated anion gap is not a sensitive indicator of moderate hyperl actatemia, but it is quite specific, provided the other main causes of the elevated anion gap have been eliminated. Changes in blood lactate only account for about half of the changes in anion gap, and serum ch loride does not seem to be an important factor in the determination of the serum anion gap.