Predictive factors of adrenal insufficiency in patients admitted to acute medical wards: a case control study
© Oboni et al.; licensee BioMed Central Ltd. 2013
Received: 20 August 2012
Accepted: 24 January 2013
Published: 26 January 2013
Adrenal insufficiency is a rare and potentially lethal disease if untreated. Several clinical signs and biological markers are associated with glucocorticoid failure but the importance of these factors for diagnosing adrenal insufficiency is not known. In this study, we aimed to assess the prevalence of and the factors associated with adrenal insufficiency among patients admitted to an acute internal medicine ward.
Retrospective, case-control study including all patients with high-dose (250 μg) ACTH-stimulation tests for suspected adrenal insufficiency performed between 2008 and 2010 in an acute internal medicine ward (n = 281). Cortisol values <550 nmol/l upon ACTH-stimulation test were considered diagnostic for adrenal insufficiency. Area under the ROC curve (AROC), sensitivity, specificity, negative and positive predictive values for adrenal insufficiency were assessed for thirteen symptoms, signs and biological variables.
32 patients (11.4%) presented adrenal insufficiency; the others served as controls. Among all clinical and biological parameters studied, history of glucocorticoid withdrawal was the only independent factor significantly associated with patients with adrenal insufficiency (Odds Ratio: 6.71, 95% CI: 3.08 –14.62). Using a logistic regression, a model with four significant and independent variable was obtained, regrouping history of glucocorticoid withdrawal (OR 7.38, 95% CI [3.18 ; 17.11], p-value <0.001), nausea (OR 3.37, 95% CI [1.03 ; 11.00], p-value 0.044), eosinophilia (OR 17.6, 95% CI [1.02; 302.3], p-value 0.048) and hyperkalemia (OR 2.41, 95% CI [0.87; 6.69], p-value 0.092). The AROC (95% CI) was 0.75 (0.70; 0.80) for this model, with 6.3 (0.8 – 20.8) for sensitivity and 99.2 (97.1 – 99.9) for specificity.
11.4% of patients with suspected adrenal insufficient admitted to acute medical ward actually do present with adrenal insufficiency, defined by an abnormal response to high-dose (250 μg) ACTH-stimulation test. A history of glucocorticoid withdrawal was the strongest factor predicting the potential adrenal failure. The combination of a history of glucocorticoid withdrawal, nausea, eosinophilia and hyperkaliemia might be of interest to suspect adrenal insufficiency.
KeywordsAdrenal insufficiency Prevalence Risk factors ACTH-stimulation test Case-control study
Adrenal insufficiency is a relatively rare but potentially lethal disease if missed during acute settings [1, 2]. Symptoms, signs and biological markers associated with adrenal failure are well known by clinicians [3, 4]. However the relative importance of these symptoms, signs and markers has not been fully studied in patients admitted in acute medical wards. Adrenal insufficiency is defined as primary or secondary. Autoimmune and tuberculous adrenalitis are the principal etiologies for primary adrenal failure, which is characterized by low cortisol levels and elevated plasma concentrations of ACTH. Impairment of the hypothalamic-pituitary corticotropic axis is responsible for secondary causes of adrenal insufficiency. These situations are characterized by low circulating levels of cortisol and ACTH. The most frequent cause of secondary adrenal insufficiency is a tumour of the hypothalamic-pituitary region but administration of supraphysiologic doses of glucocorticoids may alter a normal hypothalamic response with secondary adrenal failure once individuals are weaned from the glucocorticoid treatment. Bilateral adrenalectomy or drug-induced adrenal insufficiency may be considered as iatrogenic etiologies for adrenal failure.
Symptoms commonly associated with adrenal insufficiency are “fatigue” (lack of energy or stamina), abdominal pain, nausea, and dizziness (hypotension symptoms). The patient history may include a key element related to previous glucocorticoid treatment, thus increasing the risk of secondary adrenal failure related to glucocorticoid withdrawal. Signs associated with adrenal insufficiency are low blood pressure, vitiligo and/or skin changes. Biological markers of the disease include hyperkalemia, hyponatremia, acidosis, hypercalcaemia and eosinophilia. Nevertheless, the proportion of individuals with one or more symptoms or signs associated with adrenal insufficiency is not clearly described in the literature.
Evaluating adrenal function is a difficult task in clinical practice . Cortisol response to ACTH administration may depend upon the biological stress of the patients (intensive care versus stable conditions), the cut-off values for cortisol levels and the use of low (1 μg) versus high (250 μg) dose ACTH-stimulation test [6–9]. Basal cortisol levels inaccurately assess adrenal function . Conversely, the high dose ACTH-stimulation test (“Synacthen® test”, cosyntropin stimulation test or tetracosactide test) is generally accepted as reliable to evaluate adrenal function in everyday practice, and considered as the easiest to perform [11–13]. If this test is recognized as gold standard in various clinical situations, it may be more difficult to define cut-off values for cortisol levels in intensive care. Patel and co-authors described expected values of basal serum cortisol (> 250 nmol/l) and peak cortisol (> 600 nmol/l) after 250 μg intramuscular tetracosactrin in acute hospital admissions . However, the optimal cut-off for peak cortisol levels avec ACTH challenge has been questioned by several authors and current recommendations seem to include a minimum of > 18-20 μg/dl (> 500–550 nmol/l) to consider the ACTH response as adequate. These normal values vary dependent on laboratory and assay. In our department, the 550 nmol/l cut-off is considered as a normal response to the ACTH challenge. The minimum increment in serum cortisol is considered by Patel  and others as invalid to diagnose adrenal failure, because individuals who have a high basal concentration, due to normal circadian rhytmicity or acute stress, may be unable to increase further cortisol secretion.
The aim of this study was to identify the clinical and biological determinants associated with adrenal insufficiency in patients admitted to an acute medical ward.
All patients admitted in a general internal ward of the Lausanne University Hospital between 1st of January 2008 and 31st of December 2010 and requiring an ACTH-stimulation test based upon clinical judgment were included. This study was approved by the Institutional Ethics Committee of the University of Lausanne.
All patient files were reviewed, and information related to a potential adrenal insufficiency as well as reasons to order the ACTH-stimulation test were included in the analysis. Symptoms collected were: fatigue, abdominal pain, nausea, generalized weakness, hypotension symptoms such as diziness and/or history of glucocorticoid withdrawal. Physical signs collected were: low blood pressure (<100 mmHg systolic and/or <50 mmHg diastolic blood pressure), vitiligo and/or skin changes. Biological markers collected were: hyperkalemia (> 5 mmol/l), hyponatremia (< 135 mmol/l), acidosis (pH < 7.35), hypercalcaemia (total and ionized, corrected calcium > 2.5 mmol/l) and eosinophilia (> 0.5 G/l). Except for age, all other variables were coded as yes/no. ACTH-stimulation test was considered as normal if any value of cortisol was ≥ 550 nmol/l.
Statistical analysis was conducted using R (http://www.r-project.org/) and Stata version 12 (Stata corp, College Station, TX, USA). Results were expressed as number of patients and (percentage) for qualitative data and as mean ± standard deviation for quantitative data. Chi -square analysis was used to compare categorical data between patients with abnormal ACTH-stimulation test and control group. A multivariate logistic regression analysis was performed to determine which variables were independently associated with an abnormal ACTH-stimulation test. Stepwise forward regression was used for model selection. Two models were used: a “restricted” model, with a significance level for entry < 0.05, and a “broad” model, with a significance level for entry <0.10. Results were expressed as Odds ratio (OR) and 95% confidence interval (CI). The performance of each model in discriminating between patients and controls was assessed using the area under the Receiver Operating Characteristic curve (AROC), the sensitivity, the specificity, the positive and the negative predictive values, with corresponding 95% CIs. Statistical significance was established with P less than 0.05.
During the 36 months of the survey, 11’310 patients were hospitalized in the general internal ward. Among them, 339 (2.9%) were clinically suspected of adrenal insufficiency and had an ACTH-stimulation test. Unexplained signs such as hypotension, hyperkalemia, metabolic acidosis or hyponatremia were the key signs used by clinicians to order the ACTH-stimulation test in most clinical situations. If adrenal failure was suspected, the ACTH-test was performed and in most cases, stress doses of glucocorticoid administrated for 48 hours. If the short ACTH stimulation test was normal, the glucocorticoid treatment was immediately stopped. In the proven 32 cases of adrenal failure, stress doses of hydrocortisone or prednisone were administrated and tapered. All patients with adrenal insufficiency left the hospital under glucocorticoid treatment. Among these 339 patients, 6 (1.8%) were considered as false-positives as they were under glucocorticoid treatment and 52 (15.3%) were excluded because of missing information for at least one study variable. The ACTH-stimulation test was normal in these 52 individuals. The remaining 281 (82.9%) patients (149 women and 132 men) were retained for analysis.
Clinical characteristics of the patients with adrenal insufficiency, versus control group
No adrenal insufficiency
Number of patients
67 ± 16
68 ± 16
63 ± 15
Low blood pressure
On bivariate analysis, history of glucocorticoid withdrawal was the only variable that was significantly different between responders and non-responders to ACTH: OR 6.71, 95% CI [3.08; 14.62]. The duration and the dose of the treatment and the time elapsed between the end of the treatment and the withdrawal was very variable between the patients, and sometimes the information was lacking. None of the abnormal responders were severly stressed, based on C reactive protein, albumin or leucocytosis values.
Multivariate analysis of factors associated with adrenal insufficiency
Variables associated with adrenal insufficiency
[2.94 ; 15.09]
[1.09 ; 11.14]
[3.18 ; 17.11]
[1.03 ; 11.00]
[1.02 ; 302.3]
[0.87 ; 6.69]
Discrimination capacity of the restrictive and the broad models
Positive predictive value (%)
Negative predictive value (%)
Correctly classified (%)
In order to assess whether the stepwise model selection was effective, we then performed a model comparison between our models and new models created by adding each time an extra variable (Additional file 1: Table S1). There was not any substantial increase of the AROC, and the differences were not significant, indicating that our model selection was efficient.
The administration of supraphysiologic doses of ACTH (250 μg) is the standard challenge to test the adrenal responsiveness. This test has been widely used and several studies have reported an excellent agreement between peak cortisol concentrations obtained during the test and in the gold standard insulin tolerance test [15, 16]. The low-dose (1 μg) ACTH test has been proposed as a more sensitive test to detect secondary adrenal failure. When compared to the usual high dose ACTH test, the 1 μg stimulation test has demonstrated a slightly improved sensitivity. However, the handling of the commercially available ampoule of 250 mg of ACTH makes the test more difficult to perform. Furthermore, it has been suggested that the low dose ACTH may bind to the surface of the injection devices and may blunt the anticipated response .
Slightly over one out of ten (11.4%) patients with suspected adrenal insufficient admitted in an acute medical ward had an abnormal response to high-dose ACTH stimulation test defined as a 550 nmol/l cut-off for cortisol. A low baseline cortisol (< 165 nmol/l) not responding to ACTH is sometimes a recognized criteria to establish adrenal insufficiency. Partial secondary adrenal insufficiency might be present in critically ill patients, characterized by a poor cortisol response to ACTH despite normal baseline cortisol. A low cortisol increment (< 248 nmol/l) was measured in 19.2% (54/281 patients) of the entire cohort and in 68.8% (22/32 patients) of the individuals with proven adrenal failure. We cannot rule out that some patients with low cortisol increment to the ACTH challenge may have some functional adrenal insufficiency. However, this was probably not the case in our selected population hospitalized in a general internal medicine ward. All critically ill patients were excluded from the study since these patients are directly admitted in the intensive care unit of our institution. Low blood pressure was present in half of patients in both cases with adrenal insufficiency and controls. This sign was – in the design of our study - not very useful to discriminate between patients with or without adrenal insufficiency. Low blood pressure remains a well recognized typical sign of acute adrenal insufficiency but clinicians should of course rule out other clinical diagnosis responsible for low blood pressure [1, 4].
Two multivariate models were analyzed. The “restricted” model was composed of history of glucocorticoid withdrawal and nausea, while the “broad” model included two more variables, eosinophilia and hyperkalemia. Still, the AROCs of both models did not differ, as they did not differ from a model including all signs and symptoms. Although the more parsimonious “restricted” model might be preferred, its screening capacity is not optimal. Hence, in order to maximise the likelihood of correctly diagnosing adrenal insufficiency, clinicians should not rely their judgement on the presence or absence of these two criteria only. The diagnosis of adrenal failure is usually obvious in patients with full-blown symptoms and signs associated with acute glucocorticoid deprivation. However, the diagnosis may be much more difficult in patients admitted in an acute medical ward where multiple co-morbidities are often present. Therefore, the clinician judgement remains a critical step in the evaluation and the “broad model” may therefore be more accurate.
Our study has several limitations. First, it was limited to an acute general internal medicine ward and the estimated prevalence might be different from other wards such as intensive care units. Second, some cases might have been missed if the diagnosis was omitted, namely where the test was not ordered. Finally, the plasma ACTH values were not collected for all individuals and we could not discriminate between primary and secondary adrenal insufficiencies. We also cannot rule out that partial adrenal failure may have been missed in the controls. The high dose ACTH test may identify only profound adrenal failure but not partial and subtle corticotropic failure that could have been detected in a low-dose (1 μg) ACTH stimulation challenge.
11.4% of suspected adrenal insufficient patients admitted to acute medical ward actually present with adrenal insufficiency, defined by an abnormal response to high-dose (250 μg) ACTH-stimulation test.
Our results suggest that history of glucocorticoid withdrawal is the factor most associated with adrenal insufficiency. The combination of a history of glucocorticoid withdrawal, nausea, eosinophilia and hyperkalemia might be of interest to suspect adrenal insufficiency.
GW is the Professor heading the internal medicine ward of the University Hospital in Lausanne, Switzerland. FP is the Professor heading the endocrinology ward of the University Hospital in Lausanne, Switzerland. PMV is a MD/PhD and has a heading position in the University institute for preventive and social medicine of the University Hospital in Lausanne, Switzerland. JBO is studying in medical school, in the 6th year.
Area under the Receiver Operating Characteristic curve
- Arlt W, Allolio B: Adrenal insufficiency. Lancet. 2003, 361 (9372): 1881-1893. 10.1016/S0140-6736(03)13492-7.View ArticlePubMedGoogle Scholar
- Willis AC, Vince FP: The prevalence of Addison's disease in Coventry, UK. Postgrad Med J. 1997, 73 (859): 286-288. 10.1136/pgmj.73.859.286.View ArticlePubMedPubMed CentralGoogle Scholar
- Dunlop D: Eighty-Six Cases of Addison's Disease. Br Med J. 1963, 2 (5362): 887-891. 10.1136/bmj.2.5362.887.View ArticlePubMedPubMed CentralGoogle Scholar
- Neary N, Nieman L: Adrenal insufficiency: etiology, diagnosis and treatment. Curr Opin Endocrinol Diabetes Obes. 2010, 17 (3): 217-223. 10.1097/MED.0b013e328338f608.View ArticlePubMedPubMed CentralGoogle Scholar
- Dorin RI, Qualls CR, Crapo LM: Diagnosis of adrenal insufficiency. Ann Intern Med. 2003, 139 (3): 194-204.View ArticlePubMedGoogle Scholar
- Kappeler H, Schwyzer R: Synthetic peptides related to the corticotropins (ACTH) and the melanophore stimulating hormones (MSH) possessing corticotropin releasing activity (CRF-activity). Experientia. 1960, 16: 415-417. 10.1007/BF02178842.View ArticlePubMedGoogle Scholar
- Wood JB, Frankland AW, James VH, Landon J: A Rapid Test of Adrenocortical Function. Lancet. 1965, 1 (7379): 243-245.View ArticlePubMedGoogle Scholar
- Crowley S, Hindmarsh PC, Holownia P, Honour JW, Brook CG: The use of low doses of ACTH in the investigation of adrenal function in man. J Endocrinol. 1991, 130 (3): 475-479. 10.1677/joe.0.1300475.View ArticlePubMedGoogle Scholar
- Rasmuson S, Olsson T, Hagg E: A low dose ACTH test to assess the function of the hypothalamic-pituitary-adrenal axis. Clin Endocrinol. 1996, 44 (2): 151-156. 10.1046/j.1365-2265.1996.600482.x.View ArticleGoogle Scholar
- Thorn GW, Goetz FC, Streeten DH, Dingman JF, Arons WL: Use of the intravenous ACTH test in clinical practice. J Clin Endocrinol Metabol. 1953, 13 (5): 604-613. 10.1210/jcem-13-5-604.View ArticleGoogle Scholar
- Lindholm J, Kehlet H, Blichert-Toft M, Dinesen B, Riishede J: Reliability of the 30-minute ACTH test in assessing hypothalamic-pituitary-adrenal function. J Clin Endocrinol Metabol. 1978, 47 (2): 272-274. 10.1210/jcem-47-2-272.View ArticleGoogle Scholar
- Lindholm J, Kehlet H: Re-evaluation of the clinical value of the 30 min ACTH test in assessing the hypothalamic-pituitary-adrenocortical function. Clin Endocrinol. 1987, 26 (1): 53-59. 10.1111/j.1365-2265.1987.tb03638.x.View ArticleGoogle Scholar
- Clark PM, Neylon I, Raggatt PR, Sheppard MC, Stewart PM: Defining the normal cortisol response to the short Synacthen test: implications for the investigation of hypothalamic-pituitary disorders. Clin Endocrinol. 1998, 49 (3): 287-292. 10.1046/j.1365-2265.1998.00555.x.View ArticleGoogle Scholar
- Patel SR, Selby C, Jeffcoate WJ: The short Synacthen test in acute hospital admissions. Clin Endocrinol. 1991, 35 (3): 259-261. 10.1111/j.1365-2265.1991.tb03532.x.View ArticleGoogle Scholar
- Cunningham SK, Moore A, McKenna TJ: Normal cortisol response to corticotrophin in patients with secondary adrenal failure. Arch Intern Med. 1983, 143: 2276-2279. 10.1001/archinte.1983.00350120066016.View ArticlePubMedGoogle Scholar
- Oelkers W: The role of high- and low-dose corticotrophin tests in the diagnosis of secondary adrenal insufficiency. Eur J Endocrinol. 1998, 139: 567-570. 10.1530/eje.0.1390567.View ArticlePubMedGoogle Scholar
- Murphy RJ, Livesey J, Espiner EA, Donald RA: The low dose ACTH test: a further word of caution. J Clin Endocrinol Metabol. 1998, 83: 712-713. 10.1210/jc.83.2.712.View ArticleGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1472-6823/13/3/prepub
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.