CTRI

Acid-base disequilibrium is common among critically ill patients and quantifying these abnormalities may be of prognostic significance. Various approaches such as the Boston, Copenhagen and Stewartâ€™s physicochemical approach have been proposed to interpret these abnormalities. Among these approaches, the empirical Boston method is widely used, though it has inherent limitations in complex clinical settings like hyperchloraemic acidosis.

In 1981, Stewart proposed what was described as a non-empirical method based on physicochemical fundamentals of acid-base physiology. Although Stewartâ€™s approach is based on sound principles, solving the complex polynomial equations involved in it at the bedside was difficult and resulted in poor adoption of the method by clinicians. Hence, in early 1990â€™s, Fencl proposed simplified equations, which were easily applicable at the bedside. Unfortunately, even this has not lead to its increased applicability. Fundamentally, the physicochemical and empirical approaches aim at measuring unmeasured anions (UA) in an attempt to grade the severity of disease process. Blood lactate is another easily measured and is one of the components of the UA.

The Fencl approach at the bedside has not been well described in ICU populations. We propose to compare the ICU mortality predictive ability of admission blood lactate and the unmeasured anions measured by Stewart-Fencl method.

METHODOLOGY

Step 1: This study includes all patients admitted to St. Johnâ€™s Medical College Medical ICU from the Emergency medicine department for a period of one year prospectively. Demographic data, admission severity and mortality prediction scores (APACHE-II, SOFA, MPM II, MODS) of all the patients will be recorded. Routine admission blood gas values including standard base excess (SBE), electrolytes, albumin and lactate levels will be captured. Unmeasured anions will be calculated using simplified Stewart-Fencl approach.16

Step 2: Patients will be followed till the discharge from ICU. Patients will be categorized into dead or alive. It will then be tested whether the unmeasured anions measured by Stewart-Fencl approach is superior to arterial blood lactate levels at admission to predict the mortality at 48 hours and ICU mortality. It is also tested whether Stewart-Fencl approach can unmasks the hidden acid-base abnormalities, which are not detected, by SBE and AG.

Step 3: Summary statistics will be reported using mean and standard deviation for continuous variables and median and IQR for non-normally distributed variables. Bivariate analyses using Chi- square test will be used to compare the predictors. Logistic regression analysis will be used with ICU mortality as the dependent variable and UA, lactate and other co-variates as the independent variables. Based on the discriminative ability of the predictive models the predictive power of the co-variated would be compared using ROC curves. All the analyses will be carried out using STATA (version:13.0).