Adrenal insufficiency in corticosteroids use: systematic review and meta-analysis.

OBJECTIVE: We aimed to estimate pooled percentages of patients with adrenal insufficiency after treatment with corticosteroids for various conditions in a meta-analysis. Secondly, we aimed to stratify the results by route of administration, disease, treatment dose and duration. METHODS: We searched seven electronic databases (PubMed, MEDLINE, EMBASE, COCHRANE, CENTRAL, Web of Science and CINAHL/Academic Search Premier) in February 2014 to identify potentially relevant studies. Original articles testing adult corticosteroid users for adrenal insufficiency were eligible. RESULTS: We included 74 articles with a total of 3753 participants. Stratified by administration form, percentages of patients with adrenal insufficiency ranged from 4.2% for nasal administration (95% CI: 0.5-28.9) to 52.2% for intra-articular administration (95% CI: 40.5-63.6). Stratified by disease, percentages ranged from 6.8% for asthma with inhalation corticosteroids only (95% CI: 3.8-12.0) to 60.0% for haematological malignancies (95% CI: 38.0-78.6). The risk also varied according to dose from 2.4% (95% CI: 0.6-9.3) (low dose) to 21.5% (95% CI: 12.0-35.5) (high dose), and according to treatment duration from 1.4% (95% CI: 0.3-7.4) (<28 days) to 27.4% (95% CI: 17.7-39.8) (>1 year) in asthma patients. CONCLUSIONS: 1) Adrenal insufficiency after discontinuation of glucocorticoid occurs frequently; 2) there is no administration form, dosing, treatment duration, or underlying disease for which adrenal insufficiency can be excluded with certainty, although higher dose and longer use give the highest risk; 3) the threshold to test corticosteroid users for adrenal insufficiency should be low in clinical practice, especially for those patients with nonspecific symptoms after cessation. Strengths and limitations of this study This is the first meta-analysis providing a broad view on the risk of adrenal insufficiency after use of various types of corticosteroids in several underlying diseases. Studies displayed heterogeneity in the type of corticosteroid used, underlying condition, treatment dose, treatment duration, and route of administration, thereby reflecting clinical practice. Our results were stratified by these factors. As no individual data were available, risk stratification at the level of the individual patient was not possible. Many articles with high level of bias were included in this meta-analysis, as there were only few articles with low level of bias available. This may have affected the results.