Are clinical trials in rheumatoid arthritis generalizable to routine practice? A re-evaluation of trial entry criteria.

OBJECTIVE Trials of disease-modifying anti-rheumatic drugs (DMARDs) enrol active rheumatoid arthritis patients identified using standard criteria (three out of four of: >/=6 tender joints, >/=6 swollen joints, ESR >/= 28 mm/h, >/=45 min morning stiffness). Concern has been expressed about generalizability, as many patients in routine practice have less active disease. Furthermore, these criteria do not map onto standard disease activity and treatment response measures. We examined how many routine patients were sufficiently active to meet trial recruitment criteria and whether alternative definitions of active disease were more appropriate. METHODS We studied 504 patients in a cross-sectional study, 156 in a longitudinal study and 94 starting new DMARDs or biologics. Patients were classified as 'trial active' (met entry criteria), in remission or 'intermediately active' (between the two). We also evaluated the effect of amendments to criteria. RESULTS Cross-sectionally only 38% patients were 'trial active', but longitudinally 68% were 'trial active' at least once. Thus, many clinic patients do have disease activity below the level required for trial entry, but over time most reach eligibility levels. More (62%) of the cohort starting new treatment were 'trial active', suggesting that recruitment criteria relate to clinical decisions. Criteria omitting morning stiffness and a disease activity score (DAS28) >/=5.4 replicated the classification given by current criteria. CONCLUSIONS Trial results can be generalized to routine practice because most clinic patients are 'trial active' when their therapy is changed and most become 'trial active' over time. As DAS-based criteria are simpler and relate directly to response measures, their use should be considered in future.