Objective: Although disease-specific measures of impairment and disability are available for use in AS, no instrument exists for assessing the impact on quality of life (QoL). Our aim was to produce an AS-specific QoL measure that would be relevant, acceptable, valid and reliable.
Methods: The Instrument (the ASQoL) was developed in parallel in the UK and the Netherlands (NL). The theoretical foundation of the ASQoL is the needs-based model of QoL. Content was derived from analysis of in-depth qualitative interviews with relevant patients in both countries. Face and content validity were tested by interview with AS patients (UK and NL). An initial postal survey was then conducted in the UK (121 patients - 92M, 29F). Rasch analysis resulted in a unidimensional 26-item questionnaire. A second postal survey was conducted (UK and NL). to explore test-retest reliability (2 weeks), internal consistency, validity and to confirm scaling properties. Comparator measures were the Nottingham Health Profile (NHP) and the BASFI (in UK and NL), the Leeds Disability Questionnaire (LDQ) in the UK, and Dougados Functional Index (DFI) in the NL.
Results of second postal survey: Questionnaires were obtained from n=164 (112M, 52F, mean age = 48y, mean disease = 20y) in UK, n=154 (110M, 44F mean age = 48Sy, mean disease = 21y) in NL. Rasch analysis showed that the items formed an hierarchical order, had minimal item misfit and stability over time. However, certain individual items were found to be problematic and these were removed resulting in an 18-item scale with raw scores at the ordinal level of measurement. For the IS-item scale test-retest reliability (Spearman's Rho) was 0.98 in UK and 0.91 in NL. Intcmal consistency (Cronbach's Alpha) 0.90 in UK 0.89 in NL. Convergent/divergent validity: good correlalion with sections of NHP (0.52 - 0.80 in UK; 0.50·0.79 in NL), BASFI(0.72). LDQ (0.67) and DFI (0.76).
Conclusions: The ASQoL presents a valuable tool for assessing the impact of treatment on the quality of life of AS patients. It is well accepted by users, relates well to existing measures of disability and has excellent psychometric and scaling properties.
