Advanced Psychometrics Workshop 2009UC Davis Alzheimer’s Disease Center

Modeling Reserve as Residual Cognition

Evidence That Lifetime Cognitive Activities Mediate the Reserve-Building Effects of

Education

Maritza Dowling, Dan Mungas, Bruce Reed, Joshua Sonnen, Milton Strauss,

Sarah Tomaszewski Farias

Advanced Psychometrics Workshop 2009

Background• Brain pathology diminishes cognitive

performance; effects are highly variable across individuals

• Cognitive reserve is often used to explain the imperfect correlation between brain pathology and clinical status

• It is a hypothetical construct

• Proxy variables, especially education, are often used in studies of cognitive reserve

Advanced Psychometrics Workshop 2009

Approach

• Conceptually, reserve explains deviations from the level of cognitive performance that would be expected for a given amount of brain pathology High reserve = better than expected

performance

• We modeled reserve as residual cognition after accounting for neuropathology

Advanced Psychometrics Workshop 2009

Cognitive factor

Residual

Npath 1 Npath 3Npath 2

Test 1 Test 2 Test 3 Test 4 Test 5 Test 6

Conceptual Illustration of Approach

‘reserve’

Advanced Psychometrics Workshop 2009

Data Sources

• 701 Neuropathology cases Rush Memory Assessment Project (MAP) (n=272) Rush Religious Orders Study (ROS)(n=429)

• All cases had comparable annual neuropsychological assessments

• Potential Reserve Indicators: leisure time cognitive activity (MAP), education, lifetime SES

Advanced Psychometrics Workshop 2009

Model-building Steps paper #1, paper #2

• CFA of AD neuropathology

• CFA of cognitive measures

• Regression of cognitive domains (factors) on neuropathology (AD, VD, LBD)

• Capture residual cognition in latent variables

• Examine dimension structure of residuals across domains; create a second order reserve factor

• Examine independent predictors of reserve

A Model to Study the Determinants of Cognitive Reserve

Episodic Memory Semantic Memory Fluency

Story

Delay Recall

Word List

Word List Recall

East Boston Story

East Boston Delay

BostonNamingReading Animals Fruits

Symbol Digit

Number Comparison

Digits Forward

Digits Backwards

Digit Ordering

ProgrMatrices

Line Orientation

NeurofibrillaryTangle Cts I

NeurofibrillaryTangle Cts II

Brain Weight Microinfarcts Strokes Lewy Bodies

NT Cts Mid Frontal Brain

NT Cts Inferior Parietal Region

NT Cts Mid Temporal

NT Cts Entorhinal

Cortex

NT Cts CA1 Hippocampal

NP Cts Mid Frontal Brain

NP Cts Mid Temporal

NP Cts Inferior Parietal Rgion

NP Cts Entorhinal

Cortex

NP Cts CA 1 Hippocampal Brain

Weight

Sex

Height

Neuritic Plaque Cts

Working Memory Perceptual Speed

DiffusePlaque Cts

DP Cts Mid Frontal Brain

DP Cts Mid Temporal

DP Cts Inferior Parietal Rgion

DP Cts Entorhinal

Cortex

DP Cts CA 1 Hippocampal

Perceptual Organization

EpisR1

SemR1

FluenR1

WmR1

ResidualComposite

PorgR1 PspR1

0 0 0 0 0 0

CognitiveActivities Age 40

Education

LfSES

Advanced Psychometrics Workshop 2009

episR

semR

EducpspR

porgR

wmRResComp

flR

.11

Advanced Psychometrics Workshop 2009

episR

CogActBLsemR

EducpspR

porgR

CogAct40wmRResComp

LifeSES

.29

.35

-.20

.11

.28

flR

Advanced Psychometrics Workshop 2009

Summary

• It is possible to operationally define reserve as residual cognition using latent variable modeling

• The 6 domain-specific measures of reserve were represented well by one global reserve construct.

• Results consistent with previous studies from the Rush MAP showing that that have used cognitive activity as a proxy indicator of reserve

• Unaware of any studies that examine how including cognitive activities modifies the effect of education on incident dementia or cognitive decline

Advanced Psychometrics Workshop 2009

• Prior papers suggest cognitive activities are particularly important markers of reserve, e.g. Wilson 2002: Education effect becomes ns when CA are in model Verghese 2003; Education and occupation effects are much

weaker when CA are in model Valenzuela 2005 review: Effect sizes in studies of CA on incident

dementia are stronger than effects sizes of education or occupation.

• Results are novel in showing that education is not positively related to reserve independent of cognitive activity Unaware of any papers that have shown this reversal in the

directionality of effect

• Operationally defining reserve as residual cognition is a useful approach

Advanced Psychometrics Workshop 2009

Next steps

• More background investigation on reversal of education effect--would any further analyses clarify what is going on?

• Otherwise, no more analysis needed• Draft Paper Quickly