OBJECTIVE: Disability has been identified as a core outcome measure in ankylosing spondylitis (AS). The Dougados Functional Index (DFI) and the Bath Ankylosing Spondylitis Functional Index (BASFI) have been selected as core measures of function in this disease. However, neither of these instruments has undergone rigorous psychometric testing. METHODS: The psychometric properties of 2 measures of disability, the BASFI and the revised Leeds Disability Questionnaire (RLDQ), were compared in a cohort of 208 outpatients with AS. Rasch analysis was used to examine the properties of each measure and to compare them on a common scale. Test-retest was assessed in a cohort of 149 subjects who completed each instrument twice over an interval of 2 weeks. RESULTS: Both instruments gave an even spread of scores across the study group, but BASFI responses were positively skewed and RLDQ responses negatively skewed. There was a highly significant difference between perceived severity groups for both instruments (Kruskal-Wallis chi-squared: RLDQ, 75.1; BASFI, 80.4; both p < 0.0001). Both instruments gave acceptable test-retest scores (RLDQ ICC = 0.95, 95% CI 0.93-0.97; BASFI ICC = 0.94, 95% CI 0.92-0.96). Both instruments were found to be unidimensional according to the Rasch model, but the BASFI had more items displaying differential item functioning. Category disordering was apparent with the BASFI but not the RLDQ. However, both instruments displayed disordered item thresholds. Neither instrument can be used as an interval measure. Both measures had "towers" of thresholds whereby several thresholds were marking the same point on the underlying disability construct. This was particularly notable in the case of the BASFI. CONCLUSION: Both the BASFI and RLDQ provide a unidimensional measure of function in AS that is in accord with patient perception of disease severity. Neither instrument can be used as an interval measure. Changing the way that the instruments are scored, for example by collapsing categories, may improve their performance.
OBJECTIVE: Disability has been identified as a core outcome measure in ankylosing spondylitis (AS). The Dougados Functional Index (DFI) and the Bath Ankylosing Spondylitis Functional Index (BASFI) have been selected as core measures of function in this disease. However, neither of these instruments has undergone rigorous psychometric testing. METHODS: The psychometric properties of 2 measures of disability, the BASFI and the revised Leeds Disability Questionnaire (RLDQ), were compared in a cohort of 208 outpatients with AS. Rasch analysis was used to examine the properties of each measure and to compare them on a common scale. Test-retest was assessed in a cohort of 149 subjects who completed each instrument twice over an interval of 2 weeks. RESULTS: Both instruments gave an even spread of scores across the study group, but BASFI responses were positively skewed and RLDQ responses negatively skewed. There was a highly significant difference between perceived severity groups for both instruments (Kruskal-Wallis chi-squared: RLDQ, 75.1; BASFI, 80.4; both p < 0.0001). Both instruments gave acceptable test-retest scores (RLDQ ICC = 0.95, 95% CI 0.93-0.97; BASFI ICC = 0.94, 95% CI 0.92-0.96). Both instruments were found to be unidimensional according to the Rasch model, but the BASFI had more items displaying differential item functioning. Category disordering was apparent with the BASFI but not the RLDQ. However, both instruments displayed disordered item thresholds. Neither instrument can be used as an interval measure. Both measures had "towers" of thresholds whereby several thresholds were marking the same point on the underlying disability construct. This was particularly notable in the case of the BASFI. CONCLUSION: Both the BASFI and RLDQ provide a unidimensional measure of function in AS that is in accord with patient perception of disease severity. Neither instrument can be used as an interval measure. Changing the way that the instruments are scored, for example by collapsing categories, may improve their performance.