The predictive value of cognitive impairment together with demographic and
health factors on long-term survival was evaluated. The population sample c
omprised 389 subjects, all 62 years old. Cognitive performances were measur
ed using verbal, visuomotor and memory tests. Cognitive impairment was dete
rmined by comparing performances with norms derived from healthy controls.
Ten years after testing, the probability of survival was 89% for the cognit
ively preserved subjects, 80% for those with mild impairment, and 71% for t
hose with moderate impairment (p = 0.009). Relative risk (RR) for shortened
survival was 1.7 (95% Cl 0.9-3.2) for the mildly, and 2.6 (95% Cl 1.4-4.8)
for the moderately impaired. Perceived health problems were, as expected.
related to reduced survival (p < 0.001, RR 3.6, 95% Cl 2.1-6.0), and there
was an association between cognitive impairment and impaired perceived heal
th (p = 0.040). Multivariate analyses with Cox's regression models showed t
hat cognitive impairment, in particular, impaired episodic memory had an as
sociation with survival, in addition to the expected effects. Thus, memory
impairment may reflect very early signs of underlying disease, and so the f
indings provide predictive validity for the cognitive methods used.