Comparison of the diagnostic accuracy of measured and calculated free cortisol in acutely ill patients using the Coolens equation.

Objective: To investigate the agreement between two methods of measurement of plasma free cortisol in acutely ill patients; an indirect method using the Coolens equation, and direct measurement using high-performance liquid chromatography-tandem mass spectrometry, which is the gold standard.

Design, Participants And Setting: Prospective observational study among patients with septic shock in a tertiary intensive care unit and patients with liver failure attending a hospital outpatient clinic while awaiting transplantation. Paired values of free cortisol levels obtained from direct measurement and from calculation were analysed to provide estimates of bias and precision for the two methods.

Outcome Measures: Free and total plasma cortisol and corticosteroid binding globulin concentrations.

Results: 102 samples were analysed. The overall bias was -17%± 50%, with 95% limits of agreement of - 115% to 80%. Bias was noted to be greater in specimens with higher albumin concentration, and was proportional to free cortisol concentration.

Conclusions: The observed bias between the two methods is of a magnitude that would be expected to produce clinically relevant discrepancies. Due to the proportional nature of the error, adding a correction factor is not feasible. Results obtained from using the Coolens method to calculate free cortisol concentration in acutely ill patients should be interpreted with caution.