DEVELOPING A NEW METRIC FOR ADLS

BACKGROUND: Most ADL summary measures add up the number of dependencies. They assume an equal weighting among items and require an arbitrary definition of dependency. METHODS: A panel of experts in geriatrics was asked to rate a set of ADL and IADL components in terms of the contribution of each to overall dependency using a magnitude estimation (ME) technique. The ratings were done for both domains and for levels of function within each domain. The resultant scores were compared with the usual dependency count approach. RESULTS: The distribution of weighted items appears to have face validity. The pattern of dependency traced by the ME model was intermediate between those created by simple counts with two different definitions of dependency. The ME approach has stronger statistical properties than the ordinal scoring approach. CONCLUSIONS: The ME approach appears to be a reasonable and workable method for creating a measure that permits meaningful statements about the mean value of functional dependencies. This result can be used for any comparison of means such as across individuals, across aggregates of individuals, or across time.