With Stewart approach, pH depends on strong ion difference (SIDe), PaCO2 and non volatile
weak acids (Atot). This approach might detect complex acid-base disorders undetected by conventional approach. We designed
this study to describe acid-base disorders with both Stewart and conventional approaches and to compare the
diagnostic performance in patients with chronic respiratory failure (CRF) in acute respiratory failure (ARF) or with acute
respiratory distress syndrome (ARDS) because they are expected to have complex acid-base disorders. Conventional approach
was based on standardized base excess (SBE), bicarbonate (HCO3
-) and anion gap (AG). Working hypotheses
were that in CRF, metabolic alkalosis was associated with respiratory acidosis and in ARDS, Stewart approach was
able to identify metabolic acidosis not detected with the conventional approach.
Observational study in a medical ICU of a University hospital on 36 patients with obstructive CRF (CRFo), 36
with non obstructive CRF (CRFno) and 28 with ARDS prospectively included over 8 months. Measurements were performed
on ICU admission, day 1 and day 2 after admission.
Metabolic alkalosis occurred in less than 5% of samples in CRF patients and in 0 ARDS patient. As compared to
CRF patients, ARDS patients exhibited lower SBE, higher AG, lower SIDe and lower Atot. In ARDS, low SIDe was primarily
due to elevated unidentified anions. Hypoalbuminemia was present in more than 75% of patients without difference
between groups. Normal values of SBE, HCO3
- and AG were very common. Stewart approach detected low
SIDe in 13% of samples with normal SBE, in 13.8% of samples with normal HCO3
-, and in 11% of samples with
normal AG corrected for abnormal albumin concentration, without difference between CRF and ARDS.
In these selected critically ill patients, Stewart approach exceeded the diagnostic performance of the conventional
approach even when AG corrected was taken into account. Further studies in CRF patients with chronic hypercapnia
and elevated bicarbonatemia are required to assess the incidence of associated metabolic alkalosis.