Adrenal insufficiency during septic shock.

OBJECTIVE To determine whether a baseline (random) cortisol concentration <25 microg/dL in patients with septic shock was a better discriminator of adrenal insufficiency than the standard (250 microg) and the low-dose (1 microg) corticotropin stimulation tests as assessed by the hemodynamic response to steroid replacement. SETTING Intensive care unit. PATIENTS Fifty-nine patients with septic shock. Their mean age was 57 +/- 16.7 yrs; 29 were male. INTERVENTIONS A baseline cortisol concentration was obtained. Patients then received an intravenous injection of 1 microg of corticotropin (low-dose test) followed 60 mins later by an injection of 249 microg of corticotropin (high-dose test). Cortisol concentrations were obtained 30 and 60 mins after low- and high-dose corticotropin. All patients were administered hydrocortisone (100 mg every 8 hrs) for the first 24 hrs while awaiting results of cortisol assessment. Patients were considered steroid responsive if the pressor agent could be discontinued within 24 hrs of the first dose of hydrocortisone. MEASUREMENTS AND MAIN RESULTS Forty-seven percent of patients died. Twenty-two percent of patients met the diagnostic criteria of adrenal insufficiency by the low-dose test and 8% by the high-dose test. However, 61% of patients met the criteria of adrenal insufficiency when we used a baseline cortisol concentration of <25 microg/dL. Twenty-two patients (37%) were steroid responsive; the baseline serum cortisol was 14.1 +/- 5.2 microg/dL in the steroid-responsive patients compared with 33.3 +/- 18 microg/dL in the steroid-nonresponsive patients (p <.0001). Ninety-five percent of steroid-responsive patients had a baseline cortisol concentration <25 microg/dL. Fifty-four percent of steroid responders had a diagnostic low-dose test and 22% a diagnostic high-dose test. Receiver operating characteristic curve analysis revealed that a stress cortisol concentration of 23.7 microg/dL was the most accurate diagnostic threshold for determination of the hemodynamic response to glucocorticoid therapy. CONCLUSIONS Adrenal insufficiency is common in patients with septic shock, the incidence depending largely on the diagnostic test and criteria used to make the diagnosis. There is clearly no absolute serum cortisol concentration that distinguishes an adequate from an insufficient adrenal response. However, we believe that a random cortisol concentration of <25 microg/dL in a highly stressed patient is a useful diagnostic threshold for the diagnosis of adrenal insufficiency.