The adrenocorticotropic hormone (ACTH) stimulation test is the definitive standard for assessing adrenal insufficiency (AI), but it can be resource-intensive. The utility of using morning basal serum cortisol levels as a predictive screening tool is an area of active investigation. This study aimed to determine optimal lower and upper cutoff values for morning basal cortisol, measured by immunoassays employing specific monoclonal antibody, to accurately predict AI and reduce the need for dynamic testing in non-critically ill patients.

We conducted a single-center, retrospective analysis of 221 non-critically ill patients admitted to a tertiary hospital in Taiwan from January 2017 to December 2024. All patients underwent a standardized morning basal serum cortisol measurement collected between 06:00 and 10:30. This standardization included instructions for patients to avoid staying up late, maintain fasting after midnight, and ensure steroid medications were withheld according to their half-life, if possible, or switched to cortisone and withheld for at least one day prior to blood draw. Subsequently, a standard dose (250 μg) ACTH stimulation test was performed. Cortisol levels were measured using Roche Elecsys® Cortisol II. AI was diagnosed if the post-ACTH 30-minute cortisol was <14.6 ug/dl and the 60-minute cortisol was <14.7 ug/dl. The diagnostic performance of the morning basal cortisol level was evaluated using receiver operating characteristic (ROC) curve analysis.

Of the 221 patients analyzed, 36.2% (n=80) were diagnosed with AI. The morning basal serum cortisol level demonstrated high diagnostic accuracy, with an area under the ROC curve (AUC) of 0.91. A "rule-in" cutoff for diagnosing AI was established at <2.72 ug/dl, which provided a high specificity of 96.4% and a sensitivity of 62.2%. Conversely, a "rule-out" cutoff of ≥5.99 ug/dl safely excluded AI with a high sensitivity of 95.1% and a specificity of 64.8%.

Standardized morning basal serum cortisol level is a highly predictive and valuable screening tool for AI in non-critically ill patients. The established high-specificity and high-sensitivity cutoffs can be effectively used to stratify patients, thereby potentially obviating a significant number of unnecessary ACTH stimulation tests. This streamlined diagnostic approach can improve clinical efficiency and resource utilization.