The measurement of locomotor disability in epidemiological studies

• Main Author: Muller, Sara Nicole
• Published: Keele University 2010
• Subjects: 614.5983; R Medicine (General)
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.528365

This thesis is concerned with the measurement of locomotor disability (LMD) in epidemiological studies. The central hypothesis was that LMD is a continuous phenomenon and research into this important health indicator, with specific reference to its relationship to pain in community-dwelling adults aged 50+ years, could be improved by interval-level measurement, rather than binary definitions. A systematic search and narrative review of the literature revealed a range of concepts and content of previous self-complete LMD instruments, and an absence of interval-level measures. A brief, self-complete scale of physical functioning, the PF-10, commonly used in epidemiological studies, and suggested as a measure of LMD, was taken as the starting point for empirical work in this thesis. A subset of five items mapped onto the LMD construct and possessed acceptable psychometric properties. Analysis of cross-sectional data from 18,497 adults using ordinal regression models and individual item responses illustrated one, albeit relatively inefficient, approach to moving beyond binary outcomes for investigating the association between pain and LMD. An interval-level measure of LMD was derived using the Rasch model and combining the five items into two super-items (walking, stair-climbing). The scoring mechanism was externally verified in local, national and international datasets, and the psychometric properties confirmed. Data from 680 initially pain-free adults were used to demonstrate the potential of the new measure for longitudinal analysis. This suggested a right-shift (worsening) in the distribution of LMD at three and six years. Pain onset resulted in a more rapid increase in LMD, and recovery from pain led to only a partial return to pre-pain levels. Locomotor disability exists on a continuum and its measurement should reflect this. An interval-level measure was derived from a set of commonly used items. This measure offers several advantages (brevity, application to retrospectively gathered data) but also has limitations (ceiling/floor effects).