The Association Between Midlife Cardiorespiratory
Fitness Levels and Later-Life Dementia
April 9th, 2013 Journal Club
University of Southern California José L González, MD
Introduction
Why this study? Prevalence
o 1/8 Americans > 65yoa o $200b/yr
Why this study?o Prevention of cognitive disabilityo Lifestyle modification = most cost-effectiveo Current evidence insufficiento No public health recommendation
Previous Studies Increased fitness protects against
o All-cause mortalityo Strokeo Diabeteso HTN
Other studies linked to dementiao Only associated dancingo Only vascular dementiao Only Alzheimers
Previous Studies Intermediate outcomes
o Brain atrophy – med. Temporal lobe vol.o MMSE
NIH consensus statement “physical activity may prevent dementia”o Self-reported physical activity
Canadian study of health and agingo 5-yr f/u, n= 4615
Study Objectives Assess association between objectively measured
fitness and all-cause dementia w/ long-duration of follow-up.
Hypothesized: pts w/ greater midlife fitness = lower risk for dementia later in lifeo Independent of antecedent cerebrovascular disease
Methods
Methods Study Design: Prospective, observational cohort
study
Cooper Longitudinal Studyo Non-profit, independent research organizationo Assessing lifestyle behavior on health outcomes
o Observational database of 28,968 community-dwelling participants
o Dallas, TX
Participants Generally healthy self-referred/employee referred
for preventative health (midlife) exam. Midlife exam:
o H&P (HTN, DM, smoker, level of education)o Physical Examo Fasting labs (blood glucose, lipids)o Anthropomorphic measurements (Ht, Wt, BMI)o ETT between 1971 - 2001
Cooper database: n = 28,968 and matched w/ indivdiuals w/ Medicare claims = 25,995
Participants w/ the following exclusions @ time of midlife
exam:o MI or strokeo Chronic illness leading to disabilityo On renal dialysiso >65yoao Prior dx of dementia before 1999
Final cohort, N = 19,458
Measurement of Cardiorespiratory Fitness
Fitness level = Max time on treadmill METs Adjusted for age and sex, classified into quintiles
o 1 = lowest levelo 5= highest level
No categorization or definition of fitness
Outcome Variables Diagnosis from Chronic Condition Data
Warehouseo Data from Medicare beneficiaries for research purposeso Used to identify chronic diseases
Primary Outcome of Interest: diagnosis of all-cause dementia defined by claim filed fromo SNF, home health, hospital outpatient or inpatient,
physician or supplier claim o 24 different ICD-9 codes for types of dementia:
• Alzheimers• Senile• Pre-senile• Vascular
Statistical Analysis
Statistical Analysis Hazard Ratios =
(chance of an event occuring)tx group (chance of an event occuring)control group
Resolution depicted on Kaplan-Meir curveo Proportion of each group where end-point has not been
reachedo End-point = dx of dementia
Cox-proportional hazards model: estimate of tx effect on survival after adjustment for other explanatory variables
Cox-Proportional Hazards Model
disease-free survival vs 5-level categorical covariate corresponding to age and sex-adjusted quintiles of fitness
Adjusted for demographic and study variableso Sex, exam age, exam year
Adjusted for clinical variableso HTN, fasting glucose level, current tobacco use, BMI,
total cholesterol, SBP, DM)
Repeated analysis w/ midlife fitness as a continuous variable (METs) rather than by category (quintile)
Results
Results Mean follow-up from CCLS data = 24 years Mean 7.2 years on Medicare data 1659 cases of all-cause dementia Prevalence of dementia increased w/ age
Age (years) 70 75 80 85
Dementia Prevalence
0.8% 2.9% 8.3% 14.8%
Table 1 Incidence of different variables amongst the 5
quintileso Raw numbers sorted by clinical variables (HTN, DM,
smoker, level of education, FLP, glucose level)o Sorted by quintiles (1 lowest, 5 highest)o Decreased incidence of all variables in higher quintiles
• Except etoh intake and education
Quintile
1 2 3 4 5 All
METs 8.1 9.4 10.4 11.3 13.3 10.6
Figure 1 Higher fitness levels = lower risk for incident
dementia Similar findings when fitness was modeled on a
continuous scale (i.e. by METs)
Figure 1: Kaplan-Meier curveo y-axis: probability of dementia-free survival (%)o x-axis: Age
Table 2 Derived hazard ratio for each quintile, reference
= 1o Lowest HR in quintile 5o Statistical significance reached in quintile 3 (CI and P-
value)
Adjusted for sex, age and listed RFso Statistical significance reached in quintile 3
Adjusted for individual RFso Only HTN was statistically significant
Results Association similar among pts w/ & w/o hx of
previous stroke o HR w/o stroke 0.74 [CI 0.61-0.90]o HR w/ stroke 0.74 [CI 0.53-1.04]
Discussion
Discussion Generally healthy community-dwelling pts +
association between o Midlife fitness levels (as measured by ETT)o Independent of other RFs
Association present w/ and w/o stroke suggesting a non-vascular MOA
No statistical significance between dementia and educationo Homogenous group (see table 1)
Discussion: MOA Previous studies confirm: ↑fitness = ↓risk DM,
HTNo Established RFs for dementia
Previous studieso Brain atrophyo ↑ # small caliber vessels, ↓ tortuosity = ? ↑ blood flowo ↓ prod. Neurotoxinso Enhanced neuroplasticity w/ exercise
Strengths & Weaknesses
Strengthso Large cohort study sizeo Long duration of f/u
Weaknesseso Not randomized: unmeasured cofounder, such as
lifestyle factors could lead to ↑ exercise & ↓ dementiao Based on Medicare claims data
• 85% sens, 89% spec
Limitations
Homogenous population (Medicare, non-Hispanic, mid to upper-mid class)
Initial exclusion criteria limits applicability Can’t give specific recommendations about
activity level due to breakdown into quintiles Future studies should focus on dose-specific
relationship to give recs
Sources: Defina LF, Willis BL, Radford NB, Gao, A,
Leonard, D, Haskell, WL et al. The Association Between Midlife Cardiorespiratory Fitness Levels and Later Life Dementia: A Cohort Study. Ann Intern Med. 2013;158:162-168