Developing physical frailty specifications for investigation of frailty pathways in older people.

Different frailty definitions are suitable for different purposes. When investigating its key multidimensional predictors and effects, narrower definitions of frailty that exclude these elements may be more desirable. For this purpose, candidate physical frailty specifications are constructed and then evaluated on their construct and concurrent validity. For 4638 participants aged 65 to 89 years from wave 2 (2004) of the English Longitudinal Study of Ageing, confirmatory factor analysis is performed to create physical frailty specifications with four indicators (slowness, weakness, exhaustion, and weight loss) and with three indicators (slowness, weakness, and either exhaustion or weight loss). Using derived factor scores, their convergent, discriminant, and concurrent validity are compared. For specifications with four indicators and with three indicators including exhaustion, slowness contributes dominantly to the physical frailty factor. However, with three indicators including weight loss, weakness contributes most. Where represented, weight loss only contributes minimally. Higher factor scores are significantly associated with chronic diseases, functional impairment, and poor self-rated health, although less so for the third specification. Factor scores for the first two specifications have low correlation with psychological and social frailty while those for the third have negligible correlation. Factor scores increase with higher Frailty Index although again less so for the third specification. Minor differences are seen across gender. On account of their convergent, discriminatory, and concurrent validity, physical frailty specifications with four indicators and with three indicators including exhaustion hold promise for use in investigation of frailty pathways involving multidimensional predictors and effects.