Development and validation of an arterial blood gas analysis interpretation algorithm for application in clinical laboratory services.

Background: Arterial blood gas analysis (ABGA) is a useful test that estimates the acid-base status of patients. However, numerically reported test results make rapid interpretation difficult. To overcome this problem, we have developed an algorithm that automatically interprets ABGA results, and assessed the validity of this algorithm for applications in clinical laboratory services.

Methods: The algorithm was developed based on well-established guidelines using three test results (pH, PaCO₂ and [HCO₃⁻]) as variables. Ninety-nine ABGA test results were analysed by the algorithm. The algorithm's interpretations and the interpretations of two representative web-based ABGA interpretation programs were compared with those of two experienced clinicians.

Results: The concordance rates between the interpretations of each of the two clinicians and the algorithm were 91.9% and 97.0%, respectively. The web-based programs could not issue definitive interpretations in 15.2% and 25.3% of cases, respectively, but the algorithm issued definitive interpretations in all cases. Of the 10 cases that invoked disagreement among interpretations by the algorithm and the two clinicians, half were interpreted as compensated acid-base disorders by the algorithm but were assessed as normal by at least one of the two clinicians. In no case did the algorithm indicate a normal condition that the clinicians assessed as an abnormal condition.

Conclusions: The interpretations of the algorithm showed a higher concordance rate with those of experienced clinicians than did two web-based programs. The algorithm sensitively detected acid-base disorders. The algorithm may be adopted by the clinical laboratory services to provide rapid and definitive interpretations of test results.