Meta-analysis: Short course oral corticosteroids improve lung function in 10% of patients with chronic obstructive pulmonary disease
ACP J Club. 1991 May-Jun;114:80. doi:10.7326/ACPJC-1991-114-3-080
Callahan CM, Dittus RS, Katz BP. Oral corticosteroid therapy for patients with stable chronic obstructive pulmonary disease. A meta-analysis. Ann Intern Med. 1991;114:216-23.
To evaluate the effect of oral corticosteroids in patients with stable chronic obstructive pulmonary disease.
English-language studies published from 1966 to 1989 that evaluated corticosteroid therapy in patients with chronic obstructive pulmonary disease were identified using MEDLINE. Bibliographies from identified studies were searched by hand.
Of the 43 studies initially identified, 33 passed an initial screening and were submitted to 3 nonstudy reviewers who assessed study quality using 9 criteria. Based on the judgment of these reviewers, 18 studies were removed from further analysis because they failed either to exclude patients with asthma or acute chronic obstructive pulmonary disease exacerbations or to report pretreatment and post-treatment spirometry results. Of the remaining 15 studies, 5 met the minimal criteria for inclusion in the meta-analysis and 10, that included 299 patients met all criteria.
Data were abstracted for each of the 15 studies by 1 of the authors using a standardized data abstraction form. Response to therapy was defined as ≥ 20% improvement from baseline forced expiratory volume in 1 second (FEV1). Sensitivity analysis was done by relaxing the 9 quality criteria and redefining patient response within the range of a 15% to 30% improvement in baseline FEV1.
The dropout rate ranged from 0% to 26%. All study treatments were similar, typically equivalent to prednisone, 40 mg/d for 14 days. Treatment effect size for each study was calculated as the difference between the proportion of patients responding to steroid therapy and the proportion of patients responding to placebo, and ranged from 0% to 38%. For the 10 studies meeting all quality criteria, the weighted mean treatment effect size was 10% (95% CI 2% to 18%) for an improvement of ≥ 20% from baseline FEV1. For an improvement of ≥ 15% and ≥ 30% from baseline FEV1 the treatment effect sizes were 10% (95% CI 2% to 18%) and 11% (95% CI 3% to 19%), respectively. No evidence of association existed between treatment effect size and study sample size, average FEV1, age, dropout rate, or publication date.
For patients with stable chronic obstructive lung disease, an improvement ≥ 20% from the FEV1 occurred 10% more often in those receiving oral corticosteroid therapy than in those receiving placebo. Individual response could not be predicted from FEV1 or age.
Sources of funding: Health Resources and Services Administration and the National Institutes of Health.
Address for article reprint: Dr. C.M. Callahan, Indiana University School of Medicine, Department of Medicine, Regenstrief Health Center, 5th floor, 1001 West 10th Street, Indianapolis, IN 46202-2859, USA.
Corticosteroid therapy has documented benefit for the treatment of acute and chronic asthma, and, without strong data to support efficacy, it is also used by some clinicians to treat chronic obstructive pulmonary disease. This meta-analysis clarifies the role of this therapy: Improvement occurs in only 10% of patients with stable disease. Additionally, if the doses required to effect this improvement must be given for extended periods, as might be the case with this chronic condition, unacceptable complications become likely (1). The aggregate risk/benefit ratio weighs against the use of this therapy for most patients with chronic obstructive pulmonary disease.
The studies included in this meta-analysis, however, excluded patients with asthma. How reliable was this exclusion? Patients with chronic obstructive pulmonary disease who have a "component of asthma" are seen regularly. Perhaps the documented variation in response to corticosteroid therapy is really the variation in the occurrence of subclinical asthma found in the study samples. If so, then an individual patient may be served best by an n-of-1 randomized trial of corticosteroid therapy, an alternative that is discussed by the authors.
Meta-analyses designed to deal critically with original analyses that have fallen short of answering clinical questions are being published regularly. This is a good one (2).
Stephen R. Jones, MD
Good Samaritan Hospital & Medical CenterPortland, Oregon, USA.