OBJECTIVE: To determine factors that predict change in walking velocit
y in older people using a multivariate model. DESIGN: Longitudinal obs
ervational study. SUBJECTS: A total of 588 persons older than age 60,
including subjects residing in a continuing care retirement community
(CCRC) (n = 248), and homebound (n = 79) and ambulatory (n = 261) subj
ects. Mean age at baseline = 77. MEASUREMENT: Independent variables in
cluded demographics, physician measures of lower-extremity joint impai
rment and other musculoskeletal and neurological variables, comorbidit
ies derived from physical examination and chart abstract, self-assesse
d arthritis pain, depression, and anxiety. The major dependent variabl
es were 2- and 4-year decline in walking velocity below a threshold as
sociated with nursing home placement. MAIN RESULTS: From baseline to Y
ear 4, median walking velocity declined from 61.8 to 53.0 m/min, and t
he proportion oi subjects above a threshold value of 11.5 m/min declin
ed from 95.3% to 80.4%. Age, joint impairment, and weakness of quadric
eps, measured at baseline, predicted 2-year and 4-year decline in walk
ing velocity. CONCLUSIONS: The findings indicate that joint impairment
and quadriceps strength contribute significantly to crossing a clinic
ally significant threshold in walking velocity among older people over
time. Future research is needed to determine whether these risk facto
rs can be modified through preventive interventions such as muscle-str
engthening exercises and pain medication.