Multicenter comparison of cortisol as measured by different methods in samples of patients with septic shock

To compare inter-laboratory and inter-assay measurements of total cortisol in patients with septic shock and to evaluate current recommendations for diagnosis of corticosteroid insufficiency in septic shock. In the multinational CORTICUS study duplicate serum samples were taken before and after corticotropin stimulation tests in patients with septic shock. Serum cortisol was measured in paired samples, one being measured by the chemical laboratory of each participating site and the other by a central laboratory using an electrochemiluminescence immunoassay. Cortisol levels measured by tandem mass spectrometry were used as a 'gold standard' reference method in a subset of samples. A total of 425 corticotropin tests (850 cortisol samples) were available for comparison of local and central laboratory measurements. The concordance correlation coefficient between central laboratoty immunoassay and local hospital assays was 0.98 (CI 0.97-0.99) when the immunoassay of one manufacturer was used and 0.60 (CI 0.54-0.65) when immunoassays of different manufacturers were used. The comparison with the reference method of mass spectrometry showed concordance correlation coefficients ranging from 0.43 to 0.97 depending on the assay under study. Diagnosis of corticosteroid insufficiency was diverging due to inter-assay variations in up to 27% of cases. In samples taken from patients in septic shock, there was a high inter-assay variation of total serum cortisol. Comparisons with a reference method revealed both over- and underestimations of true cortisol levels. These inter-assay variations in samples of patients with septic shock complicate the diagnosis of corticosteroid insufficiency.