Critical review of clinical assessment techniques for rheumatoid arthritis trials: new developments

Since the early work of Taylor (1937) and Steinbrocker (1949), clinical methods of assessing the response of patients to anti-rheumatic drugs have undergone considerable revision. The evolution of clinical metrology has seen simplicity yield to complexity, and in some instances resulted in controversy. Clinical assessment techniques (indices) for rheumatoid arthritis (RA) trials should fulfil ten fundamental criteria: the index should be designed for a specific purpose (clinical versus radiographic, short versus long-term, unidimensional versus multidimensional outcomes); the index should have been validated on individuals or populations of patients having similar characteristics to future study populations; reliability (test-retest, intra-class correlation, observer agreement) should be adequate for achieving measurement objectives; validity (face, content, criterion, construct) should be adequate for achieving measurement objectives; the index must be sufficiently responsive, i.e. able to detect significant change in the underlying variable; index performance should have been maintained in subsequent applications under similar study conditions; the method of deriving scores, particularly in composite indices, should be both credible and comprehensive; the feasibility of data collection and instrument application should not be constrained by time or cost; the measurement process must be ethical; and finally, utilisation of the index should have been adopted by other clinical investigators. Despite progress in clinical assessment techniques in RA trials, there is still insufficient standardisation. This situation could be improved with further outcome conferences and consensus development exercises.