Agreement between venous and arterial blood gas analysis of acid-base status in critical care and ward patients: a retrospective cohort study.

PURPOSE: To determine whether the use of venous blood gases can be a suitable alternative to arterial sampling to evaluate acid-base status.

METHODS: The database of the clinical laboratory in a large academic hospital was searched for records of venous blood gas analysis and an arterial sample taken within ten minutes from the same patient. Bland-Altman analyses of pH, pCO2 , and lactate were performed for samples obtained from patients separately from within and outside the intensive care unit (ICU).

RESULTS: In 2,296 paired arterial-venous samples from 351 ICU patients, the bias was 0.044, -6.2 mmHg, and -0.07 mEq·L-1 for pH, pCO2 , and lactate, respectively. The range of agreement centred on this bias (upper minus lower level of agreement) was 0.134, 16.7 mmHg, and 1.35 mEq·L-1 for pH, pCO2 , and lactate, respectively. Intraclass correlation coefficients (ICCs) were 0.79, 0.76, and 0.99 for pH, pCO2 , and lactate, respectively, indicating excellent agreement. Multiple samples obtained from the same patient had a median standard deviation of 0.02, 2.77 mmHg, and 0.18 mEq·L-1 for pH, pCO2 , and lactate, respectively. Similar agreement was observed in samples from patients outside the ICU, although the ICC was only 0.53 for pCO2 .

CONCLUSIONS: Venous gases are suitable for initial evaluation of acid-base status in critically ill patients. Based on clinical evaluation, an arterial sample may then be considered for confirmation, and thereafter, venous blood gases could be sufficient for monitoring response to treatment.