Review: Corticosteroid therapy does not induce peptic ulcer
ACP J Club. 1995 May-June;122:81. doi:10.7326/ACPJC-1995-122-3-081
Conn HO, Poynard T. Corticosteroids and peptic ulcer: meta-analysis of adverse events during steroid therapy. J Intern Med. 1994 Dec;236:619-32.
To determine whether corticosteroid therapy induces peptic ulcer and other complications.
A search was done in the MEDLARS database (1966 to 1982) using the terms clinical trials, corticotropin, cortisone, dexamethasone, hydrocortisone, methylprednisolone, paramethasone, prednisolone, prednisone, triamcinolone, and adrenal cortex hormones; the Quarterly Cumulative Index of Medicine was searched (1950 to 1963) using the terms therapy, adrenal cortex preparations, and hormones in addition to the terms for individual hormones used in the MEDLARS search; and bibliographies of previous meta-analyses and of the identified studies were reviewed.
Selected studies were double-blind, randomized controlled trials of steroid therapy. Investigations including other therapies, such as nonsteroidal anti-inflammatory drugs (NSAIDs) or antacids, were included if the drugs were administered systematically to both study and control group patients. Studies in which medications such as chemotherapy were given to only 1 group were excluded. Of 1857 studies reviewed, 93 that included a total of 6602 patients were eligible for inclusion.
The study design, inclusion criteria, dosage and duration of steroid therapy, and complications were evaluated for each study by 2 independent reviewers. The complications considered were peptic ulcer, diabetes, sepsis, psychosis, tuberculosis, hypertension, osteoporosis, and cosmetic effects. The number of events for each complication was compared for patients in the steroid and control groups using 3 statistical techniques.
The steroid and control groups did not differ for number of patients with peptic ulcer (0.4% vs 0.3%; P > 0.25, respectively), nor were differences between groups noted for different treatment durations. A small difference existed, however, in symptoms compatible with peptic ulcer (0.9% vs 0.3%, P < 0.01). Of the 7 other complications, 3 occurred 4 times more frequently among patients receiving steroids than among controls: cosmetic effects (10.5% vs 2.6%, P < 0.001), diabetes (1.2% vs 0.3%, P < 0.02), and hypertension (0.9% vs 0.2%, P < 0.01). The other 4 complications did not differ between the 2 treatment groups.
No association exists between the administration of steroid therapy and the development of peptic ulcer. More occurrences of dermatologic effects, diabetes, and hypertension are noted among patients receiving corticosteroids than among those not receiving them.
Source of funding: Not stated.
For article reprint: Not available.
Whether corticosteroid therapy is associated with peptic ulcer disease remains controversial, but the systematic review by Conn and Poynard clarifies earlier discrepancies (1, 2). Although the incidence of ulcer increased with longer-term steroid therapy, this increase was not significant. Dyspepsia, however, occurred more frequently in patients treated with steroids (0.9%) than in those receiving placebo (0.3%). This may help to explain the impression that peptic ulcers are associated with steroid use.
Combining the results of trials (meta-analysis) is useful for determining overall effect sizes and for ascertaining whether low study power could explain previous negative results. These analyses, however, are still subject to bias. For example, ulcers may have been missed because only symptomatic persons were investigated and most of the studies did not apply endoscopy. It is also possible that steroid use puts only some patients at risk for developing ulcers, for example, those receiving a total dose > 1000 mg prednisone equivalent (1) or those with a past history of ulcer. Another key issue is the possible interaction between steroids and NSAIDs. A case-control study reported that NSAIDs accounted for ulcer development in patients receiving steroids (3). Several potential confounding variables were considered but not adjusted for by Conn and Poynard because no overall difference in ulcer development between control and steroid groups was observed.
The take-home message is that clinically important peptic ulcer disease is unlikely to be associated with steroid therapy; thus, coprescription of prophylactic agents to prevent peptic ulcer in patients receiving steroids alone need not be recommended.
Nicholas J. Talley, MD, PhD
Wayne H.C. Hu, MBBS, MRCPUniversity of SydneySydney, New South Wales, Australia