From the Journals

Mortality burden of dementia may be greater than estimated



The mortality burden associated with dementia may be 2.7 times greater than estimated, according to an analysis of a prospective cohort study. This burden may be greatest among non-Hispanic black older adults, compared with Hispanic and non-Hispanic whites. This burden also is significantly greater among people with less than a high school education, compared with those with a college education.

Andrew C. Stokes, PhD, assistant professor of global health at Boston University School of Public Health.

Dr. Andrew C. Stokes

The study results underscore the importance of broadening access to population-based interventions that focus on dementia prevention and care, the investigators wrote. “Future research could examine the extent to which deaths attributable to dementia and underestimation of dementia as an underlying cause of death on death certificates might have changed over time,” wrote Andrew C. Stokes, PhD, assistant professor of global health at the Boston University School of Public Health, and colleagues.

The study was published online Aug. 24 in JAMA Neurology.

In 2019, approximately 5.6 million adults in the United States who were aged 65 years or older had Alzheimer’s disease, vascular dementia, or mixed-cause dementia. A further 18.8% of Americans in this age group had cognitive impairment without dementia (CIND). About one third of patients with CIND may develop Alzheimer’s disease or related dementias (ADRD) within 5 years.

Research suggests that medical examiners significantly underreport ADRD on death certificates. One community-based study, for example, found that only 25% of deaths in patients with dementia had Alzheimer’s disease listed on the death certificates. Other research found that deaths in patients with dementia were often coded using more proximate causes, such as cardiovascular disease, sepsis, and pneumonia.

Health and retirement study

Dr. Stokes and colleagues examined data from the Health and Retirement Study (HRS) to evaluate the association of dementia and CIND with all-cause mortality. The HRS is a longitudinal cohort study of adults older than 50 years who live in the community. Its sample is nationally representative. The HRS investigators also initiated the Aging, Demographics, and Memory study to develop a procedure for assessing cognitive status in the HRS sample.

In their study, Dr. Stokes and colleagues included adults who had been sampled in the 2000 wave of HRS. They focused on participants between ages 70 and 99 years at baseline, and their final sample included 7,342 older adults. To identify dementia status, the researchers used the Langa–Weir score cutoff, which is based on tests of immediate and delayed recall of 10 words, a serial 7-second task, and a backward counting task. They also classified dementia status using the Herzog–Wallace, Wu, Hurd, and modified Hurd algorithms.

At baseline, the researchers measured age, sex, race or ethnicity, educational attainment, smoking status, self-reported disease diagnoses, and U.S. Census division as covariates. The National Center for Health Statistics linked HRS data with National Death Index records. These linked records include underlying cause of death and any mention of a condition or cause of death on the death certificate. The researchers compared the percentage of deaths attributable to ADRD according to a population attributable fraction estimate with the proportion of dementia-related deaths according to underlying causes and with any mention of dementia on death certificates.

The sample of 7,342 older adults included 4,348 (60.3%) women. Data for 1,030 (13.4%) people were reported by proxy. At baseline, most participants (64.0%) were between ages 70 and 79 years, 31% were between ages 80 and 89, and 5% were between ages 90 and 99 years. The prevalence of dementia in the complete sample was 14.3%, and the prevalence of CIND was 24.7%. The prevalence of dementia (22.4%) and CIND (29.3%) was higher among decedents than among the full population.

The hazard ratio (HR) for mortality was 2.53 among participants with dementia and 1.53 among patients with CIND. Although 13.6% of deaths were attributable to dementia, the proportion of deaths assigned to dementia as an underlying cause on death certificates was 5.0%. This discrepancy suggests that dementia is underreported by more than a factor of 2.7.

The mortality burden of dementia was 24.7% in non-Hispanic black older adults, 20.7% in Hispanic white participants, and 12.2% in non-Hispanic white participants. In addition, the mortality burden of dementia was significantly greater among participants with less than a high school education (16.2%) than among participants with a college education (9.8%).

The degree to which the underlying cause of death underestimated the mortality burden of dementia varied by sociodemographic characteristics, health status, and geography. The burden was underestimated by a factor of 7.1 among non-Hispanic black participants, a factor of 4.1 among Hispanic participants, and a factor of 2.3 among non-Hispanic white participants. The burden was underestimated by a factor of 3.5 in men and a factor of 2.4 in women. In addition, the burden was underestimated by a factor of 3.0 among participants with less than a high school education, by a factor of 2.3 among participants with a high school education, by a factor of 1.9 in participants with some college, and by a factor of 2.5 among participants with a college or higher education.

One of the study’s strengths was its population attributable fraction analysis, which reduced the risk of overestimating the mortality burden of dementia, Dr. Stokes and colleagues wrote. Examining CIND is valuable because of its high prevalence and consequent influence on outcomes in the population, even though CIND is associated with a lower mortality risk, they added. Nevertheless, the investigators were unable to assess mortality for dementia subtypes, and the classifications of dementia status and CIND may be subject to measurement error.


Next Article: