Objective: This study aimed to evaluate the performance of screening tests (serum cortisol and 24-h urinary free cortisol) and the human-corticotrophin releasing hormone (h-CRH) test in the assessment of adrenal function in patients with hypothalamic-pituitary disorders.
Design: Summary receiver operating characteristics (SROC) curve analysis was applied with the insulin tolerance test (ITT) as reference test. A peak serum cortisol response to ITT > or = 500 nmol/l indicated adrenal sufficiency. The sensitivity at the intersect of the diagonal between sensitivity = 1 and (1-specificity) = 1 with the SROC curve, where sensitivity and specificity are equal, and the corresponding weighted kappa, an estimate of agreement with the ITT, served as parameters of test performance. The diagnostic yield, representing the proportion of tests obviating the need for an ITT, was also calculated.
Measurements: Serum cortisol at 0800 h (n = 122), at 1600 h (n = 116), 24-h urinary free cortisol (n = 115) and the peak serum cortisol to h-CRH (n = 129) were compared with the peak serum cortisol to ITT.
Patients: Eighty patients with hypothalamic-pituitary disorders in whom 75 ITT's were performed pre- and 57 post-operatively.
Results: Sensitivity at the intersect and weighted kappa were higher for 0800 h serum cortisol (0.873 and 0.763 respectively) than for 1600 h serum cortisol (0.769 and 0.561) and 24-h urinary free cortisol (0.777 and 0.576). These parameters were 0.868 and 0.756 for the h-CRH test. The diagnostic yield was 63.9% for 0800 h serum cortisol compared to 25.9% for 1600 h serum cortisol (P < 10(-8)), 23.5% for 24-h urinary free cortisol (P < 10(-8)) and 60.5% for the h-CRH test (NS).
Conclusions: Serum cortisol measurement at 0800 h is better than 1600 h and 24-h urinary free cortisol to evaluate adrenal function in this patient category. The diagnostic applicability of the h-CRH test is not superior to 0800 h serum cortisol measurement.