Diagnosis of acid-base derangements and mortality prediction in the trauma intensive care unit: the physiochemical approach.

BACKGROUND: Conventional measures such as anion gap and base deficit can be inadequate for defining and managing complex acid-base derangements. Physiochemical analysis is an alternative approach based on the principles of electroneutrality and conservation of mass, and may be more accurate for defining the presence and type of acidosis and unmeasured anions.
METHODS: We retrospectively analyzed 2,152 sets of laboratory data from 427 trauma patients admitted to the intensive care unit. All data sets included simultaneous measurements of an arterial blood gas with base deficit (BD), serum electrolytes, albumin, lactate, and a calculated anion gap (AG). Physiochemical analysis was used to calculate the corrected anion gap (AGcorr), the apparent strong ion difference, the effective strong ion difference, the strong ion gap (SIG), and the base deficit corrected for unmeasured anions (BDua). Statistical analysis comparing AG and BD to the physiochemical measures was performed on all data and the subset of admission laboratory data only (n = 427).
RESULTS: Unmeasured anions as defined by an elevated SIG were present in 92% of patients (mean SIG, 5.9 +/- 3.3), whereas hyperlactatemia and hyperchloremia were present in only 18% and 21%, respectively. The physiochemical approach yielded a different clinical interpretation of the acid-base status than the conventional approach in 597 (28%) of the data sets. Lactate level was more strongly correlated with the physiochemical measures of SIG (r = 0.48) and AGcorr (r = 0.47) than with the conventional measures of AG (r = 0.24) and BD (r = 0.36, p < 0.01 for all). Both admission BD and BDua were significantly elevated in nonsurvivors, and logistic regression analysis for prediction of mortality revealed an area under the curve of 0.70 for BDua (p < 0.01) versus 0.65 for BD (p < 0.01). AGcorr and SIG did not differentiate survivors from nonsurvivors in the group as a whole. However, analysis of patients with a normal admission lactate level (n = 322) demonstrated a significant difference between survivors and nonsurvivors in SIG (7 vs. 5, p = 0.009), BDua (-4.2 vs. -2.0, p = 0.004), and AGcorr (21 vs. 19, p = 0.04), whereas the conventional measures of BD and AG showed no significant discriminatory ability.
CONCLUSION: Unmeasured anions are the most common component of metabolic acidosis in trauma intensive care unit patients. The physiochemical approach can significantly alter the acid-base diagnosis compared with conventional measures. The SIG, AGcorr, and BDua may be particularly helpful in predicting acid-base derangements and mortality in patients with normal serum lactate levels.