BACKGROUND: In recent years, there has been a growing interest in measuring quality of life (QoL) in people with osteoarthritis (OA). While several measures exist that evaluate pain and function, no tool currently exists to address the overall impact of OA on QoL. The aim of this study was to develop and validate a needs-based QoL measure for OA.
METHODS: The OAQoL was developed in three phases. In Phase 1, in-depth, semi-structured interviews were conducted with 44 OA patients to explore issues related to living with OA. Direct quotes made by participants formed the items, with a true/not true response format. Cognitive debriefing interviews held with 17 people with OA explored the relevance, clarity and ease of completion of the measure. In Phase 2, the draft questionnaire was mailed to 635 patients to test the psychometric properties of the questionnaire using Rasch analysis. Test-retest assessment of the revised questionnaire was undertaken in Phase 3 by mailing the OAQoL to a further 201 participants on two occasions, two weeks apart. In each phase of the study, participants who had single and multiple site OA, including OA at the knee, hip, hand and foot were included.
RESULTS: Participant profiles for each phase of the study are presented in Table 1. A 39 item draft measure devised during Phase 1 was mailed out in Phase 2. Rasch analysis of the draft questionnaire (n=259) indicated initial misfit to the model and differential item functioning (DIF) for several items, including, I feel like a burden to others and pain controls my life. Once 14 problematic items were deleted the data showed good fit to the model (χ2[75]=83.60, p=0.23). The test-retest (n=60) assessment indicated that all items were answered consistently across time. However, while the questionnaire was unidimensional, there was a high fit residual for one item and DIF associated with gender for the item, I worry I hold people back and for joint location for the item, It takes me longer to complete household tasks. These items were removed, leaving a 22 item OAQoL that demonstrated good fit to the Rasch model (χ2[44]=44.56, p=0.53), with excellent test re-test agreement (ρ=0.925, p<0.001; z=-0.06, p=0.995).
CONCLUSION: The OAQoL is a simple and easy to use 22 item measure of OA-specific QoL. It has been developed as a true patient-based questionnaire and demonstrates good psychometric properties, including invariance by age, gender and site, and test-retest reliability.
