OBJECTIVE: To determine factors that predict decline in manual perform
ance using a multivariate model of determinants of functional limitati
on. DESIGN: Longitudinal observational study. SETTINGS: Ambulatory gen
eral medicine clinics, residences of homebound individuals, and a cont
inuing care retirement community. PARTICIPANTS: Subjects were 485 pers
ons more than 60 years of age and included continuing care retirement
community (CCRC) residents (n = 215), chronically homebound older pers
ons (n = 65), and ambulatory older adults (n = 205). Mean age at basel
ine was 78 years. MEASUREMENT: Independent variables included demograp
hics, physician measures of upper-extremity joint impairment, comorbid
ities derived from physical examination and chart abstract, self-asses
sed arthritis pain, depression, and anxiety. The major dependent varia
ble was 2-year decline in timed manual performance below a threshold a
ssoci ated with need for long-term care services. RESULTS: The proport
ion of subjects who exceed a Timed Manual Performance Test threshold o
f 350 seconds increased slowly from baseline through Year 4 for all ag
e groups but rose rapidly from Year 4 to Year 6 for the oldest group (
>85 years at baseline). Using a discrete survival model, we found that
age, education, grip strength, and psychological status predicted cro
ssing the manual performance threshold within a 2-year period. CONCLUS
IONS: The findings, coupled with earlier findings that upper extremity
joint impairment predicted both grip strength and manual performance,
suggest that joint impairment may be an important risk factor for fut
ure functional limitation. Since diminished hand function has been sho
wn to predict dependency, development and testing of interventions to
maintain or restore upper extremity joint function and reduce pain wou
ld appear to be a high research priority.