|
Low-dose corticosteroids in rheumatoid arthritis: a meta-analysis of their moderate-term effectiveness |
Saag K G, Criswell L A, Sems K M, Nettleman M D, Kolluri S |
|
|
Authors' objectives To compare the effectiveness of prednisolone to placebo and active drug controls, and to compare the relative effectiveness of prednisolone to second line agents in the treatment of rheumatoid arthritis (RA).
Searching Clinical studies published in the English language were identified by searching MEDLINE from 1966 to 1994 using the subject headings 'corticosteroids' and 'rheumatoid arthritis', and by handsearching Arthritis and Rheumatism and the Scandinavian Journal of Rheumatology from inception through 1994; all abstracts of Arthritis and Rheumatism over the past 15 years were also examined for unpublished studies. Reference lists of all identified studies were reviewed and the authors of all unpublished manuscripts were contacted.
Study selection Study designs of evaluations included in the reviewThe studies included those of randomised controlled and crossover design, which lasted at least 3 months, used prednisolone (or a comparative corticosteroid preparation) at a mean dosage of at most 15 mg/day with either placebo or active drug control, and reported at least one of the outcome measures in a quantitative manner.
Specific interventions included in the reviewThe drug treatments given included prednisolone (or a comparable corticosteroid preparation) at a mean dosage equal to or less than 15 mg/day, deflazacort, chloroquine, aspirin and the following second line agents: antimalarials, gold salts and penicillamine. Background use of second-line agents and/or non-steroidal anti-inflammatory agents was allowed.
Participants included in the reviewAdult patients over the age of 18 years were studied. These included patients who were documented as having met either the 1958 or the 1987 American Rheumatism Association criteria for RA, others who were treated before these criteria were in use but were considered to have RA, and 9 patients with an inflammatory disorder other than RA. The mean age of the patients studied was 53.4 years (standard deviation, SD= 9.7), the mean proportion of females was 60.2% (SD= 8.1), and the mean duration of RA was 4.8 years (SD=3.1).
Outcomes assessed in the reviewThe main outcomes assessed were joint tenderness, joint swelling, grip strength and erythrocyte sedimentation rate (ESR).
How were decisions on the relevance of primary studies made?Two reviewers, blinded to the results and any information from which the study could be identified, abstracted information on preselected inclusion criteria.
Assessment of study quality The studies were rated for quality using the method developed by Chalmers et al. (see Other Publications of Related Interest), which includes evaluation of the degree of blinding, choice of analytical techniques and the control of bias. The authors do not state how the papers were assessed for quality, or how many of the authors performed the quality assessment.
Data extraction At least two reviewers abstracted the data, with any ambiguities being resolved by consensus of the reviewers or by contacting the authors.
Methods of synthesis How were the studies combined?A summary mean, weighted for sample size and study variance, was calculated for each outcome and a summary effect size was calculated by combining the results across trials. A random-effects model was used. To compare steroids with second-line agents, a comparative meta-analysis was performed by separating the corticosteroid treatment arm from the control arm in each study and statistically combining all studies within each treatment arm. The combined effect size was calculated as the mean of the joint tenderness, joint swelling, ESR and grip strength effect sizes.
How were differences between studies investigated?Heterogeneity was assessed using the Q statistic.
Results of the review Nine studies (N=472) were included.
The quality rating for the included studies ranged from 15 to 56% of the perfect score.
The Q statistic for the assessment of heterogeneity between studies for effect sizes gave P<0.001.
Number of tender joints. Active drug and placebo control was assessed using 6 studies: summary mean 1.78 (95% confidence interval, CI: 0.12, 3.43), effect size 0.60 (95% CI: -0.31, +1.51). Placebo control only was assessed using 4 studies: summary mean 2.43 (95% CI: 0.27, 4.58), effect size 0.90 (95% CI: -0.18, +2.00).
Number of swollen joints. Active drug and placebo control was assessed using 4 studies: summary mean 2.08 (95% CI: -0.60, +4.77), effect size 0.71 (95% CI: -0.41, +1.83). Placebo control only was assessed using 3 studies: summary mean 3.09 (95% CI: 0.51, 5.66), effect size 1.05 (95% CI: -0.30, +2.38).
ESR: active drug and placebo control was assessed using 5 studies: summary mean 8.02 (95% CI: 1.75, 14.3), effect size 0.72 (95% CI: -0.17, +1.61). Pacebo control only was assessed using 3 studies: summary mean 11.8 (95% CI: 4.28, 19.4), effect size 1.20 (95% CI: -0.26, +2.65).
Relative effectiveness of prednisolone to second-line agents commonly used for the treatment of RA: the unadjusted combined effect size for the corticosteroid arm was 0.82. For placebo-controlled studies, the combined effect size was 0.86.
Authors' conclusions The findings are based on very few controlled studies, but generally they uphold the widely held belief that corticosteroids are effective for the treatment of RA when administered for periods of around 6 months. Further investigation is needed to carefully weigh the relative effectiveness of steroids against their known toxicity.
CRD commentary This clearly-written review included predefined inclusion criteria, details of the methods used to select trials for inclusion, quality rating of the primary studies, and a discussion on the limitations of the review. As the authors state, the information gained from this meta-analysis was restricted by the small number of relevant trials fulfilling the criteria for inclusion. Limiting the literature search to English language studies may have omitted some relevant studies, but many other potential sources were searched for both unpublished and published studies. More comprehensive details of the individual primary studies including study design, quality assessment scores, sample size, baseline comparability of treatment groups, intention to treat analysis and results from individual studies, would have been helpful. It is unclear if analysis of the crossover trials excluded the treatment/period effect and carry-over effects, prior to estimating the effect size. As the authors state, the primary studies had numerous methodological problems: lack of double-blinding; small sample size, thus limiting statistical adjustment for important patient characteristics and background therapy; data beyond one year were unavailable in sufficient quantities to perform a longer-term meta-analysis; and few studies included radiographic information, thus limiting the outcome measures used. Significant heterogeneity was found among studies, and the stated intention of the authors, i.e. to stratify the studies based on quality specific estimates and to conduct sensitivity analysis or adjust the effect sizes for important covariates (e.g. RA disease duration, age, co-morbidity and baseline measures of RA disease severity), was not considered possible given the small number of studies. The comparison of steroids with second-line agents by combining all studies within each treatment arm does, as the authors acknowledge, disrupt the randomisation intention and may diminish the validity of any results. The authors correctly acknowledge the possible limitations on the generalisability of the results to older patients or those with a longer duration of the disease. The included studies ranged from 1955 to 1995, a period during which, as the authors state, drug therapies for RA have altered. Newer drugs such as methotrexate and sulphasalazine are now used in the treatment of RA. Given the small number of poor quality studies, lack of baseline characteristics, lack of adjustment for covariates, lack of investigation of heterogeneity and use of drugs which have now been displaced by others, it is unclear what the place of steroids would be in a treatment regime for RA today.
Implications of the review for practice and research There is a need for ongoing evaluation, using relevant outcome measures, of the most cost-effective therapies for RA.
Bibliographic details Saag K G, Criswell L A, Sems K M, Nettleman M D, Kolluri S. Low-dose corticosteroids in rheumatoid arthritis: a meta-analysis of their moderate-term effectiveness. Arthritis and Rheumatism 1996; 39(11): 1818-1825 Other publications of related interest Chalmers TC, Smith H, Blackburn B, Silverman B, Schroeder B, Reitman D, et al. A method for assessing the quality of a randomized control trial. Control Clin Trials 1981;2:31-49.
Indexing Status Subject indexing assigned by NLM MeSH Anti-Inflammatory Agents /therapeutic use; Anti-Inflammatory Agents, Non-Steroidal /therapeutic use; Arthritis, Rheumatoid /drug therapy; Aspirin /therapeutic use; Chloroquine /therapeutic use; Dose-Response Relationship, Drug; Humans; Prednisone /administration & Pregnenediones /therapeutic use; Treatment Outcome; dosage /therapeutic use AccessionNumber 11996001893 Date bibliographic record published 30/06/1998 Date abstract record published 30/06/1998 Record Status This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn. |
|
|
|