Physicochemical acid-base parameters and mortality in 1,070 ICU patients: a retrospective cohort study
- 505 Downloads
KeywordsRetrospective Cohort Study Subclavian Vein Clinical Information System Standard Statistical Technique Influence Mortality
Acid-base disorders are common in ICU patients and are associated with increased mortality and morbidity. A physicochemical (PHYSCO) approach to acid-base abnormalities based on principles described by Stewart and Figge may explain various derangements better than traditional approaches. Recent data suggest an increased inorganic ion difference is associated with lower mortality in septic patients. We determined the effects of baseline and changes in PHYSCO parameters suggested by Stewart and Figge on 30-day mortality in a large cohort of unselected ICU patients.
We performed a retrospective cohort study using data obtained from the electronic clinical information system. All patients who underwent lactate and SvO2 measurement (from internal jugular or subclavian vein) during the first 24 hours after ICU admission were included. Baseline, daily follow-up and 30-day mortality data were recorded. The worst recorded were used to calculate, for each day and for the change from day 1 to day 3, the following PHYSCO parameters: apparent strong ion difference (SIDa), inorganic SIDa (iSIDa; excludes lactate), effective SID (SIDe: includes pCO2, albumin, pH and phosphate) and strong ion gap (SIG; SIDa minus SIDe). Descriptive analysis used standard statistical techniques and univariate logistic regression (dependent variable 30-day mortality) assumed statistical significance at P < 0.05.
A total of 1,544 patients were included. Complete data to calculate day 1 and day 3 parameters were available in 1,070 and 656 patients, respectively. Overall 30-day mortality was 21.3%. Mean (SD) age and APACHE II score were 63.0 (16.4) and 21.6 (6.9), respectively. For survivors versus nonsurvivors there were significant differences in baseline SIDa (P = 0.03), iSIDa, SIDe, lactate and base excess (all P < 0.001), but not in SIG (P = 0.1). An increased SIDa (P = 0.015) and SIDe (P < 0.001), and decreased iSIDa and lactate (all P < 0.001) but not SIG (P = 0.09) were predictive of mortality. For day 1 to day 3, only Δ-SIG was weakly predictive of mortality (OR 1.03 per 1 unit increase, P = 0.04, 95% CI = 1.00 to 1.05, n = 656) but there was no difference between survival groups (P = 0.051).
This study confirms recent data suggesting baseline differences (but not changes over 3 days) in the relationships between strong and weak ions and pCO2 (reflected in SIDa, lactate, and impacts on mortality). The different effects observed for SIG and BE imply a significant unmeasured acid load influencing mortality. Bias and confounding may affect these findings and they should therefore be confirmed prospectively.