Validated Criteria for the Interpretation of a Single Measurement of Serum Cortisol in the Investigation of Suspected Adrenal Insufficiency

Scott D. Mackenzie; Robert M. Gifford; Luke D. Boyle; Mike S. Crane; Mark W. J. Strachan; Fraser W. Gibb


Clin Endocrinol. 2019;91(5):608-615. 

In This Article

Subjects and Methods

As this was a retrospective study of de-identified patient data, ethics approval was not required. Analysis of 1624 SSTs, performed between 2011 and 2014, was undertaken in 1222 patients attending general endocrine clinics at the Western General Hospital, Edinburgh, UK (derivation cohort). Demographic data and information on the indication for the SST were collected by review of electronic case records. Serum cortisol was measured before and 30 minutes after intramuscular administration of ACTH1–24 (Synacthen®) 250 μg. For individuals prescribed oral glucocorticoid replacement, this was withheld on the day of testing. Criteria were proposed to define upper cut-offs for morning (8 AM-12 PM) and afternoon (12 PM-6 PM) baseline serum cortisol which were >95% sensitive for the detection of a subnormal-stimulated cortisol. Individuals with a basal cortisol <50 nmol/L were assumed to have a subnormal-stimulated cortisol in keeping with AI, and the accuracy of this assumption was tested. The proportion of individuals lying between these two thresholds, and thus requiring confirmatory testing, was used to assess the impact of such a strategy on the number of SSTs required.

To assess the external validity of the proposed cut-offs, we tested these criteria in two validation cohorts. The criteria were applied retrospectively to 873 SSTs undertaken in 770 individuals attending general endocrine clinics at the Royal Infirmary, Edinburgh, UK, over the same time period (outpatient validation cohort). The third cohort (inpatient validation cohort; n = 804 SSTs in 776 individuals) comprised patients admitted to acute medical units and general medical wards in both hospitals over the period 2011–2015. For inpatients, data were included where the baseline sample from the SST was taken between 6 AM and 6 PM.

Analysis was undertaken to: (a) compare the performance of the test in the morning to that in the afternoon; and (b) assess whether performance of the test in the outpatient and inpatient validation cohorts matched that of the derivation cohort, across both morning and afternoon periods.

Serum cortisol was measured using the same methodology across all cohorts (Abbott Architect® i2000 immunoassay system), according to the manufacturer's protocol. The assay consistently demonstrated an inter-assay precision below the manufacturer stated CV of 10%. An adequate response to Synacthen was defined as a 30 minute cortisol of ≥430 nmol/L.[20]

Statistical Methods

Sensitivity was defined as the probability that a patient failing the SST is below a given threshold for basal cortisol, and specificity the probability that a patient passing the SST is above that threshold. Positive predictive value (PPV) denotes the proportion of patients below the threshold who fail the SST and negative predictive value (NPV) the proportion of patients above the threshold who pass the SST.

Sensitivity, specificity, NPV and PPV were compared using Fisher's exact test. Area under curve (AUC) for ROC curve analysis and linear regression analysis were performed using SPSS version 23.