Abstract

METHODS

Details on the Adult Changes in Thought (ACT) study, a prospective cohort study of dementia, have been published previously.3 Participants are members of Group Health, an integrated health care delivery system in the Northwest United States. The original cohort consisted of 2581 people enrolled in 1994 to 1996. An expansion cohort (n = 811) was enrolled in 2000 to 2002. In 2004, the study began ongoing enrollment to replace people who die, develop dementia, or drop out. The analysis included 3605 participants who were enrolled in ACT and who had follow-up data available on September 30, 2010.

ACT participants are recruited from a random sample of community-dwelling members aged 65 years or older who are invited to an intake visit. Consenting individuals found to be free of dementia are enrolled. ACT participants are evaluated for dementia at baseline and every 2 years with the Cognitive Abilities Screening Instrument.14 Those with Cognitive Abilities Screening Instrument scores lower than 86 are referred for a complete diagnostic evaluation that includes physical and neurological examinations, neuropsychological testing, and laboratory and imaging studies if not already available. A consensus conference including the study physicians, a neuropsychologist, and a research nurse determines diagnoses of dementia based on Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), criteria15 and of Alzheimer’s disease based on DSM-IV and National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association criteria.16 Participants remain in the cohort until they are determined to have dementia or Alzheimer’s disease or are censored as a result of death or discontinuation of participation in the study. Only dementia diagnoses from ACT evaluations were used in analysis. Mortality was confirmed through death certificates, state death registry data, annual telephone contact with family members of study participants with dementia, quarterly mailings to participants, and thorough review of local obituaries.

We described the distribution of participant characteristics, including age, gender, race/ethnicity, education, health conditions detected during follow-up, and baseline body mass index (BMI; defined as weight in kilograms divided by the square of height in meters), stratified by observed dementia incidence. We calculated age-specific incidence of dementia and Alzheimer’s disease with the person-years approach17 by dividing the number of cases by the person-years at risk in 5-year age intervals starting from age 65 years. We multiplied these rates by 1000 to report rates per 1000 person-years. The number of person-years contributed by a participant without dementia was the time between baseline examination to last follow-up examination. The number of person-years contributed by a participant with dementia was the time from baseline to the midpoint between the visit at which dementia was diagnosed and the previous follow-up examination. For participants who died or dropped out of the study, the date of last follow-up examination was used as the end point. We calculated confidence intervals (CIs) for incidence rates with a Poisson distribution for the number of cases within each age interval.

We used a 3-state, piecewise homogeneous Markov model to estimate life expectancy with and without dementia.18,19 Life expectancy without dementia at a given age was defined as the average number of years a participant who attained that age without dementia was expected to live free of dementia.9 Life expectancy with dementia was the average number of years a participant who attained that age without dementia was expected to live following a subsequent dementia diagnosis. Thus, these measures incorporate both the risk of developing dementia and the average number of years lived following a dementia diagnosis. By contrast, other measures of survival include only those who currently have dementia.20 We also calculated percentage of total life expectancy without dementia. In our Markov model, we specified states of no dementia, dementia, and death. We assumed a progressive model structure, allowing transitions between no dementia and dementia or death and between dementia and death. The piecewise, homogeneous Markov model assumes constant instantaneous transition rates within each age stratum. Age strata used in our model were 70, 71, 72, 73, 74, 75 to 76, 77 to 79, 80 to 84, 85 to 89, and 90 years or older. On the basis of estimated age stratum–specific transition rates, we then estimated life expectancy and dementia-free life expectancy in a simulated cohort of 1 000 000 individuals. For each individual, we simulated a life history beginning at age 70 years based on maximum likelihood estimates from the Markov model. We then computed mean age at dementia diagnosis, mean age at death, and mean number of years between dementia diagnosis and death across the cohort. To construct 95% CIs, we repeated the cohort simulation with the 2.5th and 97.5th percentiles of the estimated transition rates.

To investigate the association between gender, education, and life expectancy, we estimated the Markov model allowing for gender-specific and education-specific transition intensities. For gender, we assumed a proportional hazards structure. Thus, transition intensities in all age strata were assumed to be multiplied by the same factor associated with gender. The proportional hazards structure for education resulted in model instability, so we allowed for separate education effects in each age stratum. Life expectancies for individuals of varying educational attainment were reported at age 75 years because these estimates were more stable than those for older or younger individuals because of the sample size. Trends with respect to educational attainment were similar for older and younger individuals.

All statistical analyses were performed with R 2.15.0 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

From 1994 to 2010, 4455 participants enrolled in the ACT study. Among the enrollees, 3605 participants had at least 1 follow-up examination. A consensus diagnosis of dementia was determined in 815 participants, with 639 diagnoses of possible or probable Alzheimer’s disease and 176 diagnoses of dementia attributable to other causes. The median length of follow-up was 6.3 years (interquartile range = 4.0–10.5 years, with a maximum of 16.4 years), accounting for 24 052 person-years.

Of the participants with Alzheimer’s disease, 68.9% died during follow-up, compared with 81.3% of those with dementia attributable to other causes and 43.7% of those with no dementia diagnosis. Participants who received a diagnosis of dementia during follow-up were more likely to be older, to be female, and to have achieved at most a high school education (Table 1). Obesity and proportion with detection of hypertension, stroke, and myocardial infarction during follow-up were similar regardless of dementia diagnosis. At the end of follow-up, 72.9% of those aged 65 to 69 years at baseline were still alive, whereas 14.7% of those aged 90 years or older at baseline were still alive. At the end of follow-up, 10.5% of those aged 65 to 69 years at baseline had a dementia diagnosis, whereas 5.6% of the respondents in this age group had an Alzheimer’s disease diagnosis. Of those aged 90 years or older at baseline, at the end of follow-up, 42.6% had a dementia diagnosis, and 33.8% had an Alzheimer’s disease diagnosis.

The incidence rate of dementia increased with each successive age group (Table 2), reflecting higher incidence rates of both Alzheimer’s disease and non–Alzheimer’s disease dementia. These patterns were apparent in men and women. The incidence rates for dementia were similar for men and women through ages 80 to 84 years. In the 2 older age groups, dementia incidence rates were higher in women than in men, with a larger gender difference for Alzheimer’s disease than for non–Alzheimer’s disease dementia.

Life expectancy without dementia (average number of years a participant who attained that age is expected to live without dementia) decreased over ages 70 to 90 years (Table 3). At age 90 years, life expectancy without dementia was 4.2 years (95% CI = 3.5, 5.1). Life expectancy with dementia (average number of years a participant who attained that age is expected to live with dementia) also decreased over age but at a more gradual rate. At age 90 years, life expectancy with dementia was 1.3 years (95% CI = 1.2, 1.5). The percentage of total life expectancy without dementia also decreased with age. At all ages, life expectancy without dementia was about 1 year longer for women than for men, and life expectancy with dementia was about 0.5 years longer for women than for men. However, at all ages, percentage of total life expectancy without dementia was slightly higher for men than for women.

Life expectancy without dementia was approximately 0.5 years longer for study participants who completed college or more education than for those with lower educational attainment among respondents aged 75 years (Table 4). Conversely, life expectancy with dementia was approximately 0.5 years shorter for participants who completed college or more education compared with those with lower educational attainment among respondents aged 75 years. Percentage of life expectancy free of dementia increased with greater educational attainment, and this gain was slightly larger for women than for men.

DISCUSSION

In a cohort free of dementia and cognitive impairment at baseline, dementia incidence increased after age 85 years in men and women. Although this result is not surprising, our study is notable for providing detailed information on the oldest Americans. Our findings are unique because of the use of longitudinal data and in estimating trends among the entire population, not just in patients who currently have dementia. Of the study participants aged 70 years or older, oldest adults, men, and those with greater educational attainment had shorter life expectancies with dementia (average years with dementia) than did their counterparts. Younger adults, women, and those with greater educational attainment had longer life expectancies without dementia (average years without dementia) than did their counterparts. Younger adults, men, and those with greater educational attainment also had higher percentages of total life expectancies without dementia than did their counterparts.

Our findings supported our hypotheses: dementia incidence increases with age; life expectancy with dementia decreases with age; women have longer life expectancy than men regardless of dementia and a lower percentage of total life expectancy without dementia; and higher education levels are associated with longer life expectancy with and without dementia and greater percentage of total life expectancy without dementia. Our finding of an increase in dementia incidence with each successive age group was consistent with most previous findings for the overall population3 and for men4,5 and women2–5,7 only. Similarly, our finding of an increase in Alzheimer’s disease incidence with each successive age group for the overall population3 and for men2,3,5,7 and women3–5,7 only was consistent with most previous findings. Our findings regarding life expectancy with and without dementia and percentage of total life expectancy without dementia in men and women by age were consistent with previous results.9–13 We found that greater education was related to increased life expectancy without dementia and a greater percentage of life expectancy without dementia. To our knowledge, no previous studies have estimated dementia-free life expectancy according to educational attainment. Consistent with our findings, previous studies found that people with lower educational attainment have a slightly longer life expectancy with cognitive impairment.21,22 More highly educated people may have greater cognitive reserve that delays reaching the threshold at which cognitive impairment is detectable and may have access to more resources that benefit health.21

This study had several limitations. Because the respondents in the sample were from a particular region of the United States and were relatively racially homogeneous, the results may not be representative of national trends. In addition, our piecewise homogeneous Markov model assumed that transition rates to dementia and death were constant within certain age intervals. This assumption was necessary to obtain estimates in age ranges for which few deaths or dementia cases were observed. Although we used the smallest practical age intervals, our results may be somewhat dependent on this choice of intervals.

The continued increase in dementia incidence into the 10th decade of life has major population health implications for an aging population with increased obesity prevalence at all stages of life. Between 2010 and 2050, the proportion of US adults aged 65 years or older is expected to increase from 13% to 20%, and the percentage aged 85 years or older is expected to double from 2% to 4%.23 In addition to an increase in the number of people at greater dementia risk because of demographic shifts, the obesity epidemic means that older adults as a group will be at greater risk for so-called mixed dementia, which is typically the coexistence of Alzheimer’s disease and vascular dementia.24,25 In 2009 to 2010, nearly 40% of adults aged 60 years or older were obese (BMI ≥ 30 kg/m²); nearly 20% of adolescents aged 12 to 19 years were obese.26 Thus, current age cohorts are experiencing a high lifetime exposure to obesity-related vascular risks, which will add to the vascular risks of dementia that are specific to older age. Although beyond the scope of this current study, investigation of BMI in relation to life expectancy with and without dementia would inform outcomes related to the obesity epidemic.

Life expectancy with dementia is a population health measure that accounts for risk of dementia incidence as well as survival following dementia onset. We used life expectancy with and without dementia to describe ACT participants, who were dementia-free at study enrollment. For the average ACT participant, the expectation is that life with dementia will be a few years and a small proportion of overall life expectancy. The apparent low life expectancy with dementia reflects, in part, a relatively modest dementia incidence rate until late in life. At older ages, the percentage of total life expectancy without dementia decreases. Thus, the burden of suffering reflected in this population measure affects a progressively greater proportion of remaining life in persons 80, 85, and 90 years old. The low percentage of overall life expectancy lived with dementia also reflects gains in life expectancy free of dementia. In industrialized countries, overall life expectancy has increased since the second half of the 20th century, in large part because of decreases in mortality in older adulthood.27 At the same time, a compression of cognitive impairment may have occurred such that the onset of dementia has been delayed and people are spending fewer years of life with dementia. Several studies reported slower cognitive decline and lower prevalence rates in age cohorts born later in the 20th century.28,29 Therefore, on the population level, dementia represents a relatively small proportion of expected life.

Nonetheless, we realize that the consequences of dementia to the people affected and their families are tremendous and a great fear in an aging society. In the ACT cohort, considering only participants with an Alzheimer’s disease diagnosis, the average survival for women was 5.7 years and for men was 4.2 years.20 Yearly US costs of dementia are estimated at $42 000 to $56 000 per person.30 This study provides reliable, up-to-date information about the effect of dementia on specific population groups. We hope our results will contribute to public health policies that delay dementia onset and improve the quality of life for patients with dementia and their caregivers.

Acknowledgments

Human Participant Protection

References