Systematic review of rheumatoid arthritis clinical studies: Suboptimal statistical analysis of radiological data.

INTRODUCTION: The distribution of progression scores in rheumatoid arthritis is highly skewed, requiring advanced statistical analysis techniques, with different techniques resulting in different outcomes. METHODS: Three databases were searched to identify rheumatoid arthritis clinical trials and observational studies that described radiographic analysis techniques, comparing at least two groups. RESULTS: Of 5980 identified papers, 225 were eligible for data extraction. Parametric techniques (t-tests, ANOVA or linear regression) were used in 39 studies, of which 18% took the skewed distribution into account. In 53 studies, continuous data was categorized and analyzed with binomial or ordinal methods (chi-square tests or logistic regression). Two studies treated the outcome as a 'count' outcome variable (applying a Poisson). CONCLUSION: There is large heterogeneity in the analysis strategy of radiographic progression in recent rheumatoid arthritis clinical trials and observational studies, with the majority of studies applying simple, suboptimal or inappropriate methods.