Abstract
OBJECTIVES: To study whether the reporting of clinical outcomes in arthritis trials measured on ordinal and interval scales is adequate in relation to meta-analysis. METHODS: Systematic review of randomised trials of non-steroidal anti-inflammatory drugs in patients with rheumatoid arthritis. Optimal reporting was defined as data in the original ordered categories for global evaluation and pain, and as mean and SD for number of tender joints and grip strength, and if a visual analogue scale had been used to measure pain. RESULTS: A total of 144 trials were included. The median sample size was 60 patients. The quality of the reporting increased over time for three of the four variables. Global evaluation was optimally reported in 52 of the 127 trials (41%) in which it was recorded. Pain was optimally reported in 27 of 98 trials (28%), number of tender joints in 41 of 123 trials (33%), and grip strength in 34 of 124 trials (27%). Even if rather broad criteria are adopted, only about half of the data were reported in a potentially useful way for a meta-analysis. CONCLUSIONS: Arthritis trials have been reported inadequately in relation to meta-analysis. As most trials are underpowered, meta-analysis is indispensable and the deficit therefore needs urgent improvement. Investigators should specify a priori what constitutes an important treatment effect and report numbers of patients improved.