In a recent European survey of outcome measures for rheumatoid arthritis over 60 different instruments were found to be in use. For low back pain, over 100 different instruments were found to be in use. If, for example, measuring disability is considered important, what are the characteristics of a good measure of disability? How do we chose between competing instruments?
Recent advances in our understanding of outcome measurement have identified a number of fundamental criteria that should be met, prior to more established tests of reliability, validity and responsiveness. The two most important characteristics are unidimensionality and absence of Differential Item Functioning (DIF).
The ICIDH helps us to think conceptually about outcomes, and contributes to measurement by identifying unidimensional constructs. A new mathematical approach (Rasch analysis) then allows for formal tests of unidimensionality in ordinal scales. Questions, or tasks which are summated in some fashion to give an overall score must be unidimensional. Furthermore, at the same level of ability (for example in walking), the probability of a patient doing a task must be the same irrespective of whether they are male or female, young or old, underlying pathology (cross-diagnostic validity) or country of use (cross-cultural validity) – the absence of DIF.
Once these fundamental requirements of measurement are met, tests of reliability, validity and responsiveness may proceed. In this way, those working in the field of arthritis can be confident that evaluation of interventions are supported by high quality measurement.