Tracking the Societal Impact of Dementia

Using the Health and Retirement Study

Kenneth M. Langa, MD, PhD

Division of General Medicine
Institute for Social Research
VA Ann Arbor Healthcare System
Institute for Healthcare Policy and Innovation
Institute of Gerontology
University of Michigan

National Press Foundation

Washington, DC
April 29, 2019
HRS
No Conflicts of Interest
 Overview

• Dementia / Alzheimer’s Disease, and their

Societal Impact
• Population Trends Important to “Brain Health”
• The Health and Retirement Study
– Measures of cognition and dementia risk factors
– Trends in dementia prevalence in the US
– Harmonized Cognitive Assessment Protocol (HCAP)
 Dementia (NIA-AA Criteria, 2011)

• Impairment in at least two of: memory; reasoning;

visuospatial abilities; language; or changes in
personality / behavior; AND
• The impairments “interfere with the ability to
function at work or at usual activities”; are a
decline from previous levels of functioning; AND
• Not explained by delirium or psychiatric disorder

**Clinical and research diagnoses are NOT straightforward; biomarkers (blood,

CSF, imaging), it is hoped, will help to standardize and clarify diagnosis.
 Causes of Dementia
• Alzheimer’s Disease (~ 60 - 70% of cases)
• Vascular Dementia (~ 20 - 30% of cases)
• Other (~10% of cases)
– Parkinson’s Disease
– Frontotemporal Dementia
– Dementia with Lewy Bodies
– Reversible Causes

• Mixed Dementia
– Overlapping AD and vascular pathology is likely most
common, especially in oldest-old

Sources: MRC CFAS, Lancet, 2001; Langa et al, JAMA, 2004; Schneider et al, Neurology, 2007.
 Overlap / Interaction of
Cardiovascular disease and AD
• Anatomy / Physiology
– Brain is 2% of body mass, but gets 20% of blood flow and
uses 20% of the body’s oxygen

• Risk Factors
– Hypertension, Diabetes, Hyperlipidemia, Obesity, Physical
inactivity, Smoking, Alcohol, Inflammation, Apo E e4 genotype
– More vascular risk factors in mid-life is associated with more
amyloid protein in the brain in later life

• Clinical
– For a given level of AD pathology, the more cerebrovascular
lesions, the greater the likelihood of CI / dementia
Sources: Snowdon et al, JAMA, 1997; Langa et al, JAMA, 2004; Schneider and Bennett, Stroke, 2010;
Zlokovic, Nature Rev Neuro, 2011; Snyder, Alz and Dem, 2015; Gottesman et al, JAMA, 2017
The Social and Economic Impact
of Dementia
 Demographic Imperative
World Population, Age 65+ and Age <5

Source: Population Reference Bureau

 HRS-ADAMS Estimates of
2012 US Population Prevalence, Age 65+

MCI
19%
Dementia
9% US Population, Age 65+:
Dementia 4.0 – 4.5 million
Normal MCI 8.0 – 8.7 million
72%

Sources : Plassman, et al, Neuroepidemiology, 2007; Annals of Internal Medicine, 2008; Langa et al, JAMA IM, 2017.
 Dementia and the Family
Informal Caregiving Time and Cost, USA

Dementia Stroke
30% 10%

Cancer
Depression 3%
15%
COPD
3%
Incontinence
8%
Other
Diabetes
21%
10% Sources: Langa et al,
2000 - 2004.
 Living Arrangements of US Adults 65+
800,000
with Dementia

600,000

400,000 Men
Women

200,000

0
With spouse Living w/other Living alone Nursing Home

Source: Langa, Kabeto, Weir, HRS 2012, unpublished data.

 Dementia Annual Cost Per Case
United States, 2010

Total: $50,000 per case;

Source: Hurd et al, NEJM, 2013. $200 billion nationwide
 Projected Global Growth in Dementia Cases

Source: Prince et al, World Alzheimer Report 2015.

Recent Trends Important to “Brain Health”
 Recent Trends Important to Brain Health

• Cardiovascular risk factors and treatment:

– Increasing prevalence of obesity, diabetes, and,
hypertension

BUT:

– More wide-spread and intensive treatment of diabetes,

hypertension, and high cholesterol
 Trends Important to Brain Health:
Diabetes Complications, US, 1990 to 2010

Source: Gregg et al, NEJM, 2014.

 Is 60 the New 50?:
Change in “Biological Age”, 1990 to 2010
US NHANES Data
(n=21,575)

Biological Age:
HgbA1c
Cholesterol
BP
FEV1
Creatinine
Alk Phos
Albumin
CRP
Source: Levine and Crimmins, Demography, 2018. ~73% of decline attributable to increased use of
hypertension and cholesterol medications in oldest group
 SES Disparities in CV Risk in the US

Source: Odutayo et al, JAMA Cardiology, 2017.

 Trends Important to Brain Health:
The Worldwide Education Boom

Fraction of 30-34 Year-olds with

College Education

Source: Becker et al, J. of Human Capital, 2010.

 Education and Brain Health
• Adults are now reaching older age having had
significantly more formal schooling earlier in life:
– 53% of age 65+ in US finished HS in 1990; 80% in 2010
– 11% of age 65+ finished college in 1990; 23% in 2010

• Education is protective against dementia, likely through

multiple pathways, including cognitive reserve, health
behaviors, occupation, leisure activities, social network,
and wealth

Sources: Older Americans 2012: Key Indicators of Well-Being; Meng and D’Arcy, PLoS One, 2012.
 Risk Factor Reduction and Declining
Dementia Prevalence
• Norton et al (2014) estimate that up to 30% of AD
cases worldwide are attributable to modifiable risks,
including low levels of education, cardiovascular risk
factors, and depression
• Even modest improvements in population levels of
these risk factors could lead to the prevention of a
significant number of AD and dementia cases

Source: Norton, Lancet Neurology, 2014.

Education, the Brain, and Life Expectancy
 Did Dementia Prevalence in the United
States Change Between 2000 and 2012?:

The Health and Retirement Study

HRS
 Health and Retirement Study (HRS)
• Ongoing, nationally representative, longitudinal,
biennial survey of ~ 20,000 Americans aged > 50
• Performed at the UM Institute for Social Research,
funded by the NIA and SSA
• Largest project / grant at UM: ~$23 million / year
• Data collection started in 1992
• Face-to-face and telephone interviews (50 / 50)
• > 3,500 HRS publications by > 2,000 authors;
>20,000 registered data users

Source: Sonnega et al, International J of Epidemiology, 2014.

 HRS Survey Content
• Demographic characteristics
• Physical and functional
health
• Performance-based cognitive
testing
• Family structure and
transfers
• Employment status, job
history, and disability
• Retirement plans and
perspectives
• Assets, income, and net
worth
• Housing and services use
• Health insurance and
pension plans
• Out-of-pocket health costs
• Links to data from
employers, Medicare, NDI,
VA, and SSA
• Biomarkers (2006)
– Cholesterol, HgbA1c, CRP,
Cystatin C, BP, Pulse, Peak
flow, Balance, Gait
• Genetics (2012)
– 2.5 M SNPs on 20,000 people
AGE HRS Longitudinal Cohort Sample Design
90

85

80
AHEAD <1924

75 CODA 1924-30
HRS 1931-41
70 WB 1942-47
EBB 1948-53
65
MBB 1954-59
LBB 1960-65
60

55

50

YEAR
HRS INTERNATIONAL PARTNER
STUDIES AROUND THE WORLD

Harmonization tool at USC Gateway to Global Aging

g2aging.org
 Tracking Brain Health in the HRS
• Modified Telephone Interview for Cognitive Status
– orientation to day, date, month, year
– immediate and delayed recall of 10 nouns
– serial 7 subtraction
– counting backwards
– object naming
– naming of the president and vice-president
• Verbal fluency
• Number Series
• Numeracy questions
• Speed of processing

• Self report of:

– Memory function; ADL / IADL limitations; prior diagnosis of AD or
dementia; medications for AD or dementia
 Tracking Brain Health in the HRS (2)
• Protective and risk factors for brain health
– CV disease risks (measured BP, obesity, health behaviors)
– Acute medical events (stroke, sepsis, CABG => Medicare)
– Genetics (ApoE, 2.5 million SNPs)
– Education and leisure time activities (reading, puzzles, etc.)
– Employment history, wealth
– Social ties and extent of social interactions
– Early-life factors (childhood health, parents’ education)

• Respondents represented by proxy:

– IQCODE
– Memory function, judgment, and change over the last 2 years
– Doctor diagnosed AD or dementia
– ADL / IADL limitations

• “Exit” Interview for those who’ve died

 Aging, Demographics, and Memory Study
(ADAMS)
• Supplemental study to the HRS funded by the NIA
• First US national, population-based study of dementia
to include subjects from all regions of the country
• Initial field period: 2001 – 2005, follow-up through 2010
• 856 HRS subjects, evaluated in their homes
– 3-4 hour neuropsychological assessment, neuro exam,
informant interview, ApoE genotype
• Consensus panel diagnosis of CIND or dementia, with
differential diagnosis of cause (AD, Vascular, Other)
Source: Langa et al, Neuroepidemiology, 2005..
 Defining Dementia in the HRS
• Equipercentile Equating
– Define cut-points on HRS cognitive (and other)
measures that result in similar dementia prevalence
estimates as the “gold-standard” ADAMS estimates
– Self-respondents: 27-point HRS cognitive scale
• 0-6 => Dementia; 7-11 => CIND

– Proxy-respondents: 1) proxy assessment of memory;

2) proxy assessment of IADL limitations; and 3)
interviewer assessment of cognitive impairment
• 11-point combined scale

Source: Crimmins et al, J. of Gerontology, 2011.

 HRS: Dementia Prevalence, Age 65+

15 P<.001
11.6
8.8
2000
% 2012

Dementia Prevalence
Source: Langa et al, JAMA Internal Medicine, 2017..
 Studies of Population Trends
• Declining Prevalence / Incidence of Cog impairment / Dementia:
– US NLTCS (Manton et al, 2005)
– US HRS (Langa et al, 2008, 2017; Hudomiet et al. 2018)
– UK CFAS / ELSA (Llewellyn and Matthews, 2009)
– US Mayo Clinic Study on Aging (Rocca, 2011)
– Rotterdam Study (Schrivjers et al, 2012)
– Swedish Kungsholmen Project (Qiu et al, 2013)
– Danish Cohorts Study (Christensen et al, 2013)
– UK Cognitive Function and Ageing Study (Matthews et al, 2013, 2016)
– US MoVIES Cohort (Dodge et al, 2014, 2016)
– US Framingham Heart Study (Satizabal et al, 2016)
– Indianapolis-Ibadan Project (Gao et al, 2016; Hendrie et al, 2018)
– US NLTCS (Stallard and Yashin, 2016)
– US Americans’ Changing Lives Study (Leggett et al, 2017)
– US Einstein Aging Study (Derby et al, 2017)
– US NHATS (Freedman et al, 2018)
HRS Harmonized Cognitive Assessment
Protocol (HCAP)
 From ADAMS to HCAP
• ADAMS showed embedding a valid dementia diagnosis in a
representative longitudinal study provides valuable information on
the impact of cognitive decline on patients, families, and public
programs
• ADAMS was expensive due to traveling teams of nurses
neuropsychology technicians
• NIA requested that the HRS team develop a less expensive
method of identifying dementia in the HRS to:
– Allow a larger sample size of in-depth cognitive assessments
– Increase opportunities to study trends in dementia incidence / prevalence in
the US
– Increase opportunities for international comparisons of dementia
prevalence and trends across the HRS family of studies
 HCAP Development and Design
• Development
– Analyses of ADAMS, ROS / MAP data for respondent and
informant tests most valuable for diagnosis
– Consultation with US and international researchers to try to
maximize overlap in tests / cognitive domains (10/66;
CFAS; ROS/MAP) and be useable in low-income countries
– Pilot Tests in Cog-USA; ACT (Seattle); and UM ADC
• General Design
– One hour of in-home cognitive testing
– 20-minute informant report
– Conducted by interviewers, not specialists
– For HRS, a random sample of half of those aged 65+
HCAP INSTRUMENT AND HARMONIZATION ACROSS STUDIES
Order Section/test name ELSA HAALSI CHARLS LASI MHAS
1 MMSE ✖ ✗ ✖ ✗ ✗
2 HRS TICS – 3 items ✖ ✖ ✖ ✖ ✖
3 Word Recall – Immediate ✖ ✖ ✖ ✖ ✖
4 Retrieval Fluency (animals) ✖ ✖ ✖ ✖ ✖ ✖ = Same
5 Letter cancellation ✖ ✗ ✗ ✗
6 Backward Count ✖ ✖ ✗ ✗ = Similar /
7 Brief CSI-D – 4 items ✖ ✖ ✖ ✖ ✖ Harmonizable
8 Word Recall – Delayed ✖ ✖ ✖ ✖ ✖
9 Logical Memory – Immediate ✖ ✖ ✖ ✖
10 Word List Recognition ✖ ✖ ✖ ✖ ✖
11 Constructional Praxis ✖ ✖ ✖ ✖
12 Symbol Digit (SDMT) ✖ ✖ ✗
13 Con Praxis Recall ✖ ✖ ✖ ✖
14 Logical Memory – Recall ✖ ✖ ✖ ✖
15 Logical Mem -Recognition ✖ ✖ ✖
16 Number Series ✖ ✗ ✗
17 Raven’s Matrices ✖ ✖ ✖
18 Trail Making Test ✖ ✗
I-1 Blessed Dementia Scale ✖ ✖ ✗ ✗
I-2 Jorm IQCODE ✖ ✖ ✖ ✖ ✖
I-3 CSI-D Informant ✖ ✖ ✖ ✖ ✖
HCAP STATUS OF HRS INTERNATIONAL PARTNER STUDIES
Country Study Applied Funded Begun Completed

USA HRS ✔ ✔ ✔ ✔
Mexico MHAS ✔ ✔ ✔ ✔
England ELSA ✔ ✔ ✔ ✔
S. Africa HAALSI ✔ ✔ ✔ ✔
China CHARLS ✔ ✔ ✔ ✔
India LASI ✔ ✔ ✔
EU SHARE ✔ ✔
S. Korea KLOSA ✔ ✔
Ireland TILDA ✔
N. Ireland NICOLA ✔
Brazil ELSI
 Conclusions
• Dementia prevalence appeared to decline in the US
between 2000 and 2012, in line with recent findings
from Framingham, England, and other HIC.
• Rising levels of education (and “cognitive reserve”?) in
more recently-born cohorts may be contributing to a
decline in age-specific dementia risk in older adults
• Better control of cardiovascular risk factors also may be
contributing to the decline in dementia risk
• Primary prevention of dementia through social and
behavioral interventions appears possible and valuable
 Key Unanswered Questions

• How will population levels of cardiovascular risk factors

change in the coming decades?
– Will rising rates of diabetes and obesity reverse the apparent
positive trend in dementia prevalence?
– Will trends be different in high-income vs. LMIC?

• What are the “active ingredients” of both early-life and

later-life education on brain health, and how can they be
efficiently and widely delivered in low-, middle-, and high-
income countries?
 Key Unanswered Questions

• Why is brain health (and life expectancy) becoming

increasingly unequal across race / ethnicity / SES in the
United States?

• Who will care for the growing number of older adults with
(and without) dementia, given the relative decline in the
number of younger adults in the decades ahead?
Citation: Langa KM, Larson EB, Crimmins EM, Faul
JF, Levine DA, Kabeto MU, Weir DR. JAMA Internal
Medicine, 2017.

Funding: National Institute on Aging: U01 AG009740

(Health and Retirement Study), K23 AG040278, P30
AG053760, and P30 AG024824; and the Social Security
Administration.
THANK YOU