INDIVIDUAL DIFFERENCES IN MEMORY FUNCTION
Richard N. Jones, John N. Morris, Adrienne Rosenberg, Paul Malloy HRCA Research and Training Institute, Boston Ken Kleinman Harvard Medical School, Boston MA
Jason Allaire
Wayne State University, Institute of Gerontology, Detroit MI
Michael Marsiske
University of Florida, Institute on Aging, Gainesville
ACTIVE
Advanced Cognitive Training for Vital and Independent Elderly
funded as a cooperative agreement with the National Institute on Aging and the National Institute of Nursing Research.
Participating Institutions & Principal Investigators: Hebrew Rehabilitation Center for Aged John Morris, Ph.D. Indiana University David Smith, M.D. Johns Hopkins University George Rebok, Ph.D. Pennsylvania State University Sherry Willis, Ph.D. University of Alabama at Birmingham Karolin Ball, Ph.D. Wayne State University Michael Marsiske, Ph.D.
Coordinating Center New England Research Institutes
Sharon Tennstedt, Ph.D.
Objectives
• Characterize performance on memory task in cognitively normal older adults • Use Latent Growth Curve Modeling (LGM) techniques • Separately model initial recall and learning • Explore association of background variables and recall, learning
Rey AVLT
Auditory Verbal Learning Test
(Lezak, 1983; Rey, 1964)
• 15 unrelated word list, read at 2 sec. intervals • Participants asked to recall words • List re-read four times • Repeated recall trials • Intrusion list • Delayed Recall and Recognition task
Latent Growth Modeling
• Psychometric approach to repeated measures data • Individual differences in intercept and slope over time captured in latent variables (initial recall, learning) • Certain parameterizations equivalent to random coefficient models, mixed models, hierarchical linear models
Methods: Participants
•HRCA –elder housing •IU –community centers •UAB –drivers licenses, ID cards –eye clinic attendees •WSU –community organizations –urban residential settings •JHU –in and around urban geriatric center –urban senior high-rise housing –suburban retirement community –urban & rural medical centers •PSU
– PACE enrollees
Eligibility and Exclusion Criteria
• Eligibility Criteria
– age 65 – living in non-institutional setting – available for duration of trial
• Exclusion Criteria
– – – – – – – – – – vision impairment hearing loss impaired communication reported disability Alzheimer’s dx or MMSE<23 cancer dx w/ limited life expectancy current chemo/radiation CVA in previous 12 months in other cognitive training high level depressive sx (CESD>16)
Background Variables
• Demographics
– – – – Age Sex Ethnicity Education
• Cognitive Functioning
– General (MMSE) – Reasoning ability (STAMAT) – Vocabulary (Kit) – Digit symbol substitution (WAIS)
Table 1. Subject characteristics, ACTIVE pilot (n=169). mean (±s.d.) or valid frequency Characteristic n percent Total 169 100 Men 29 17 Women 140 83 Age group 65-74 years 96 58 75-84 59 35 85 and older 12 7 Ethnicity White, non-Hispanic 67 40 Black or African-American 94 56 Other 8 5 Level of education less than 12 years/GED 54 32 12 years/GED 54 32 more than 12/GED 61 36 SF-36 Physical SF-36 Mental Cognitive functioning Reasoning (Word Series) Digit symbol (DSST) Vocabulary (words) MMSE 149 149 167 169 169 169 43 (±10) 54 (±8) 7 (±5) 35 (±15) 6 (±3) 27 (±2)
15
12
words recalled
9
6
3
0 1 2 3 AV LT trial 4 5
LGM Parameterization
without Background Variables
yi = + i + i i = + i
0 0 0 0 0 =
1 1 1 1 1
* 0 0 1 2 VAR() = = 0 3 0 4 0
0 * 0 0 0
0 0 * 0 0
0 0 0 * 0
0 0 0 0 *
* *
VAR()
= =
* * *
Alternative Functional Forms of AVLT Growth
Linear
=
1 1 1 1 1
0 1 2 3 4
Quadratic
=
1 1 1 1 1
0 0 1 1 2 4 3 6 4 16
Logarithmic (ln(t+1)) 1 0.000 1 0.693 1 1.099 = 1 1.386 1 1.609
Logarithmic Growth Function
Estimated final time score
=
Transformed metric
=
1 1 1 1 1
0.000 0.693 1.099 1.386 *
42 = 1.494 (0.033) 1.609-1.494 0.033 = 3.48
1 1 1 1 1
0.000/1.494 0.693/1.494 1.099/1.494 1.386/1.494 * 1 1 1 1 1 0.000 0.464 0.736 0.928 *
or
=
Table 2. Fitted AVLT performance growth model, ACTIVE Pilot (n=169). parameter time scores 12 22 32 42 52 Growth factor means 1 (initial recall) 2 (learning) Growth factor variance/covariance 11 (initial recall) 22 (learning) 21 (covariance of growth factors) Observed variable (AVLT) residual variance 11 22 33 44 55 Model chi-square (df), P CFI RMSEA
1 2
estimate (se)1 0.00 2 0.46 0.74 0.93 1.00 (0.02) 5.53 (0.13) 5.21 (0.18) 2.58 (0.34) 4.43 (0.61)
0.00
0.33 (0.18) 1.33 (0.17) 0.86 (0.13) 0.93 (0.17) 1.99 (0.27) 9.73 (10), P=0.47 1.00 0.03
- standard error - entries in italic typeface are not estimated, but fixed equal to the shown value.
16 14 12 10 8 6 4 2 0 0
Model implied trajectories
1
2
3
4
5
6
LGM Parameterization with Background Variables
yi = + i + i i = + x + i
0 0 0 0 0 * * =
1 1 1 1 1
* 0.000 0 0.464 0.736 VAR() = = 0 0.928 0 0 *
0 * 0 0 0
0 0 * 0 0
0 0 0 * 0
0 0 0 0 *
* *
* * * * * * * * * VAR() = = 0 * * * * * * * * *
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Women Age group 75-84 85 and older Black or African-American less than 12 years/GED2 Mental/Physical Health SF-36 Physical† SF-36 Mental† Reasoning (Word Series) † Digit Symbol† Vocabulary (words) † General cognition (MMSE) † Model chi-square (df), P CFI RMSEA
est (se)1
est (se) 1
0.77 (0.34) -0.09 (0.29) -0.98 (0.55) -1.12 (0.30) 0.01 (0.31) 0.01 (0.01) 0.01 (0.02) 0.00 (0.04) 0.01 (0.01) 0.06 (0.05) -0.05 (0.07)
0.50* -0.06 -0.63 -0.73*** 0.01 0.06 0.04 -0.01 0.08 0.14 -0.07
0.67 (0.40) 0.35 -0.42 (0.34) -0.86 (0.66) 0.84 (0.35) 0.33 (0.37) 0.01 (0.02) -0.02 (0.02) 0.09 (0.05) 0.03 (0.01) 0.07 (0.06) 0.32 (0.08) -0.21 -0.44 0.42* 0.16 0.04 -0.08 0.20 0.21* 0.12 0.34***
57.14 (43), P=0.07 0.989 0.044
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Women
est (se)1
est (se) 1
0.77 (0.34)
0.50*
0.67 (0.40) 0.35
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Age group 75-84 85 and older
est (se)1
est (se) 1
-0.09 (0.29) -0.98 (0.55)
-0.06 -0.63
-0.42 (0.34) -0.21 -0.86 (0.66) -0.44
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Black or African-American
est (se)1
est (se) 1
-1.12 (0.30)
-0.73***
0.84 (0.35) 0.42*
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
15
12
AVLT recall (words)
9
6
3
0 0 1 2 3 trial Black/African-American, Not Hispanic White and other ethnicity groups 4 5 6
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
less than 12 years/GED2
est (se)1
est (se) 1
0.01 (0.31)
0.01
0.33 (0.37) 0.16
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Mental/Physical Health SF-36 Physical† SF-36 Mental†
est (se)1
est (se) 1
0.01 (0.01) 0.01 (0.02)
0.06 0.04
0.01 (0.02) 0.04 -0.02 (0.02) -0.08
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Reasoning (Word Series) †
est (se)1
est (se) 1
0.00 (0.04)
-0.01
0.09 (0.05) 0.20
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Digit Symbol†
est (se)1
est (se) 1
0.01 (0.01)
0.08
0.03 (0.01) 0.21*
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
Vocabulary (words) †
est (se)1
est (se) 1
0.06 (0.05)
0.14
0.07 (0.06) 0.12
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Table 3. Association of participant characteristics and AVLT growth model factors. ACTIVE Pilot (n=169). initial recall
standardized effect
learning
standardized effect
characteristic
est (se)1
est (se) 1
General cognition (MMSE) † -0.05 (0.07)
-0.07
0.32 (0.08) 0.34***
* P<0.05, ** P<0.01, *** P<0.001 † - effect size estimates shown standardized to respect to mean an variance of dependent and background variable; all other (dichotomous) covariates standardized with respect to dependent variable only. 1 - estimate (est) and standard error (se) 2 - general equivalency degree
Conclusion
• LGM provides a comprehensible model of initial recall and learning performance on the AVLT task • Alternative summary measures are suggested by considering effect of background variables on initial recall and rate learning
LGM Alternative Parameterization
without Background Variables
yi = + i + i i = + i
=* 1 2=1 2=1 2=1 = 2 1 =
1 1 1 1 1
* 0 0 1 2 VAR() = = 0 3 0 4 0
0 * 0 0 0
0 0 * 0 0
0 0 0 * 0
0 0 0 0 *
0 *
VAR()
= =
* * *
Word List Used in ACTIVE
• • • • • • • Drum Curtain Bell Coffee School Parent Moon • • • • • • • • Garden Hat Farmer Nose Turkey Color House River