Functional decline among older adults: comparing a chronic disease cohort and controls when mortality rates are markedly different.

In studies of functional status decline in older persons, the strategy for handling deaths during follow-up may influence policy implications. We compared 301 older ESRD dialysis patients with 322 controls to determine whether functional decline over 3 years among dialysis patients exceeded that of "normal aging." We used two different statistical methods and, for each, compared results when deaths were excluded and then included in the analysis. Dialysis patients incurred a larger follow-up mortality rate and were more impaired at baseline. Findings based on functional transition over time, assessed by a nominal variable, were sensitive to whether or not deaths were included in the analysis. However, findings based on nonparametric methods for an ordinal scale (functional impairment) were not sensitive to whether or not deaths were included in the analysis. Analyzing data with and without deceased subjects may be the most comprehensive approach to comparing two cohorts over time.