Age-related deficit accumulation and the risk of late-life dementia

Abstract

Introduction: Many age-related health problems have been associated with dementia, leading to the hypothesis that late-life dementia may be determined less by specific risk factors, and more by the operation of multiple health deficits in the aggregate. Our study addressed (a) how the predictive value of dementia risk varies by the number of deficits considered and (b) how traditional (for example. vascular risks) and nontraditional risk factors (for example, foot problems, nasal congestion) compare in their predictive effects.

Methods: Older adults in the Canadian Study of Health and Aging who were cognitively healthy at baseline were analyzed (men, 2,902; women, 4,337). Over a 10-year period, 44.8% of men and 33.4% of women died; 7.4% of men and 9.1% of women without baseline cognitive impairment developed dementia. Self-rated health problems, including, but not restricted to, dementia risk factors, were coded as deficit present/absent. Different numbers of randomly selected variables were used to calculate various iterations of the index (that is, the proportion of deficits present in an individual. Risks for 10-year mortality and dementia outcomes were evaluated separately for men and women by using logistic regression, adjusted for age. The prediction accuracy was evaluated by using C-statistics.

Results: Age-adjusted odds ratios per additional deficit were 1.22 (95% confidence interval (CI), 1.18 to 1.26) in men and 1.14 (1.11 to 1.16) in women in relation to death, and 1.18 (1.12 to 1.25) in men and 1.08 (1.04 to 1.11) in women in relation to dementia. The predictive value increased with the number (n) of deficits considered, regardless of whether they were known dementia risks, and stabilized at n > 25. The all-factor index best predicted dementia (C-statistics, 0.67 ± 0.03).

Conclusions: The variety of items associated with dementias suggests that some part of the risk might relate more to aberrant repair processes, than to specifically toxic results. The epidemiology of late-life illness might best consider overall health status.

Figure 1

Flow diagram showing the sample. The CSHA assembled a representative cohort of 10,263 participants aged 65 years and older in 1991/1992 (CSHA-1) in all Canadian provinces, with follow-ups occurring in 1996/1997 (CSHA-2) and 2001/2002 (CSHA-3) [CSHA 2000]. At baseline, community-dwelling older adults were screened, and self-reported health evaluation data were available in 8,940 participants who completed the baseline survey. Global cognitive assessment was made with use of the 100-point Modified Mini-Mental State Examination (3MS) [Teng EL, Chui HC. The Modified Mini-Mental State (3MS) examination.J Clin Psychiatry 1987, 48; 314–318.]. People who had 3MS total scores ≤78 were invited to have a detailed clinical cognitive examination, at which time, a clinical diagnosis was made. Cognitive status of all participants was assessed at baseline and at both 5-year and 10-year follow-ups. Subjects who were assessed as cognitively intact at baseline, based on negative screening (3MS >78) and/or a negative clinical diagnosis (2,902 men and 4,337 women) were further analyzed.

Figure 2

Risk-factor indices in relation to age. The indices containing vascular risk factors (VRFIs; n = 4), nontraditional risk factors (NTRFIs; n = 19), and all risk factors (ARFIs; n = 42) are presented as a function of age in men (green) and in women (red). Symbols represent the observational data as means for 3-year age groups; lines represent curve fitting. The nontraditional and all factor indices increased exponentially with age (for example, R² = 0.981 in men and R² = 0.939 in women for the NTRFI; R² = 0.982 in men and R² = 0.987 in women for the ARFI; P < 0.001 in each case).

Figure 3

Ten-year death (left panels) and dementia (right panels) rates in men (upper panels) and women (lower panels) as a function of deficit accumulation for the 19-item nontraditional risk-factor index (NTRFI), the four-item vascular risk-factor index (VRFI), and the 42-item all-risk-factor index (ARFI). Data represent the mean and the variance of the population for the tertiles with the lowest (open bars), medium (light-grey bars), and the highest (dark-grey bars) index scores.

Figure 4

C-statistics (areas under the curve) of the indices in the prediction of death (left panel) and dementia (right panel) as a function of the number of deficits considered in the indices, in men (green) and in women (red). Symbols represent the observational data, and lines represent the curve fitting to a reversed exponential function (R² > 0.960; P < 0.001). Data were obtained by using randomly selected variables with 1,000 simulations. Solid symbols on the left show the performance with use of each individual deficit (traditional risk factors, stars; nontraditional, asterisks; others, dots). Performance of the nontraditional, traditional, and all risk-factor indices are shown by the larger down-triangles, up-triangles, and diamond symbols.