Stewart’s and Fencl’s methods have recently been proposed to interpret acid–base disorders where traditional theory has proven inadequate. Our objectives were to evaluate: (1) the occurrence of acid–base disturbances in critically ill patients and their trend over the first 3 intensive care unit days, (2) whether Stewart’s theory offers advantages over the traditional theory in the diagnosis of acid–base metabolic disturbances and (3) whether variables derived from Stewart’s and Fencl’s methods offer advantages over the traditional method to predict patient mortality.

A prospective cohort study in a general intensive care unit. Blood samples were analysed for arterial blood gases, electrolytes and proteins. PaCO2, pH, bicarbonate, base excess, standard base-excess, sodium, potassium, chloride, phosphorous, calcium, magnesium and lactate were measured. Anion gap, Stewart’s and Fencl’s variables were calculated.

When using Stewart’s method, metabolic acidosis and metabolic alkalosis were found in 92.9% and 93.4% of samples, respectively. Corresponding figures obtained with the traditional method were 15% and 18.7%. In 245 (64.5%) samples, Stewart’s method revealed that metabolic acidosis and alkalosis were simultaneously present, whereas the traditional method revealed a normal acid–base status. Strong ion gap increased significantly over the first 3 intensive care unit days. Strong ion gap and lactate were independent predictors of 28-day mortality.

Metabolic acidosis by unmeasured anions is a clinically relevant phenomenon, which is correlated with mortality. Progressive metabolic acidosis may be ongoing in the early phase of critical illness despite the absence of acidaemia.

To determine if peripheral venous blood gas values for pH, partial pressure of carbon dioxide (PCO2) and the resultant calculated bicarbonate (HCO3) predict arterial values accurately enough to replace them in a clinical setting.

This prospective observational study was performed in a university tertiary care emergency department from June to December 1998. Patients requiring arterial blood gas analysis were enrolled and underwent simultaneous venous blood gas sampling. The following data were prospectively recorded: age, sex, presenting complaint, vital signs, oxygen saturation, sample times, number of attempts and indication for testing. Correlation coefficients and mean differences with 95% confidence intervals (CIs) were calculated for pH, PCO2 and HCO3. A survey of 45 academic emergency physicians was performed to determine the minimal clinically important difference for each variable.

The 218 subjects ranged in age from 15 to 90 (mean 60.4) years. The 2 blood samples were drawn within 10 minutes of each other for 205 (96%) of the 214 patients for whom data on timing were available. Pearson’s product–moment correlation coefficients between arterial and venous values were as follows: pH, 0.913; PCO2, 0.921; and HCO3, 0.953. The mean differences (and 95% CIs) between arterial and venous samples were as follows: pH, 0.036 (0.030–0.042); PCO2, 6.0 (5.0–7.0) mm Hg; and HCO3, 1.5 (1.3–1.7) mEq/L. The mean differences (± 2 standard deviations) were greater than the minimum clinically important differences identified in the survey.

Arterial and venous blood gas samples were strongly correlated, and there were only small differences between them. A survey of emergency physicians suggested that the differences are too large to allow for interchangeability of results; however, venous values may be valid if used in conjunction with a correction factor or for trending purposes.