Declining Executive Control in Normal Aging Predicts Change
in Functional Status: The Freedom House Study
Donald R. Royall, MD, wz Raymond Palmer, PhD,§ Laura K. Chiodo, MD,w
and Marsha J. Polk, MMEdw
OBJECTIVES: To assess the contribution of executive
control function (ECF) to functional status.
DESIGN: Three-year longitudinal cohort study.
SETTING: A comprehensive-care retirement community.
PARTICIPANTS: Five hundred forty-seven noninstitution-
alized septuagenarians.
MEASUREMENTS: The Mini-Mental State Examination
(MMSE) and Executive Interview (EXIT25). Functional
status was assessed using instrumental activities of daily
living (IADLs). Latent growth curves of MMSE, EXIT25,
and IADL were modeled. The rate of change in IADLs
(DIADL), adjusted for baseline IADLs and cognition, was
regressed on the rate of change in each cognitive measure
(adjusted for baseline cognition). Models were also
adjusted for baseline age, level of care, and comorbid
illnesses.
RESULTS: Baseline test scores were within normal ranges,
but mean EXIT25 scores reached the impaired range by the
second follow-up. There was significant variability around
the baseline means and slopes for all variables. The rate of
change in EXIT25 was strongly correlated with DIADL
(r 5
0.57, Po.001). This remained significant after
adjusting for baseline EXIT25 scores, IADLs, age, comor-
bid disease, and level of care. The effect of the EXIT25 on
DIADL was stronger than those of age, baseline IADLs,
comorbid disease, or level of care. The rate of change in
MMSE scores was not significantly associated with DIADL.
CONCLUSION: ECF is a significant and independent
correlate of functional status in normal aging. Traditional
dementia case finding is likely to underestimate cognition-
related disability. Neither a normal baseline MMSE score
nor stable MMSE scores over time preclude functionally
significant changes in ECF. J Am Geriatr Soc 52:346–352,
2004.
From the Departments of Psychiatry, wMedicine, zPharmacology, and
§
Family Practice, South Texas Veterans’ Health System Audie L. Murphy
Division GRECC, and the University of Texas Health Science Center,
San Antonio, Texas.
This study has been supported by a grant from the Alzheimer’s Research
Foundation.
Address correspondence to Dr. Donald Royall, Department of Psychiatry,
The University of Texas Health Science Center at San Antonio, 7703 Floyd
Curl Drive, San Antonio, TX 78284. E-mail: royall@uthscsa.edu
JAGS 52:346–352, 2004
r 2004 by the American Geriatrics Society
Key words: aging; disability; assessment; longitudinal;
executive function
A ge-related decline in executive control functions
(ECFs) is one plausible explanation for the decline in
functional status that accompanies old age. Executive
functions are cognitive processes that orchestrate complex,
goal-directed activities.1 The latter include functional
activities such as cooking, dressing, and housework.
Traditionally, ECF has been associated with the
prefrontal cortex, but it is also dependent on the integrity
of frontal-subcortical systems.2 Thus, frontal lesions are
sufficient, but not necessary, for ECF impairment. This
greatly broadens the differential of conditions that affect
executive control. ECF impairment has been associated
with conditions as diverse as major depression, subcortical
vascular disease, adult-onset diabetes mellitus, Alzheimer’s
disease (AD), and even normal aging.3
The natural history of ECF during normal aging cannot
be easily derived from the existing literature because
familiar screening measures such as the Mini-Mental State
Examination (MMSE)4,5 or its derivatives do not detect it
well. The American Psychiatric Association added ECF
impairment to its definition of dementia in 1994.6 None-
theless, epidemiological studies seldom incorporate ECF
measures into their case definitions. Moreover, many
epidemiological studies are cross-sectional in nature. The
conclusions about age-related cognitive decline that can be
validly drawn from cross-sectional models may be severely
limited.7
The authors have been studying longitudinal changes in
ECF in residents of continuing-care retirement communities
(CCRCs). CCRCs provide longitudinal access to large,
stable elderly populations receiving care across a wide range
of services. The Freedom House Study (FHS) examines the
incidence of ECF impairment within a single CCRC, the Air
Force Villages (AFV). This FHS report investigates the
longitudinal effects of emergent cognitive impairment on
functional status using two measures: the MMSE, a familiar
measure of ‘‘global’’ cognitive ability, and the Executive
0002-8614/04/$15.00
JAGS MARCH 2004–VOL. 52, NO. 3 DECLINING EXECUTIVE CONTROL PREDICTS CHANGE IN FUNCTIONAL STATUS
Interview (EXIT25),8 a measure specifically designed to
assess ECF.
METHODS
Participants
Five hundred forty-seven elderly retirees were recruited
from a randomly ordered list of AFV residents aged 70 and
older living at noninstitutionalized levels of care; these
included garden apartments and congregate high-rises.
Garden apartments were considered the lowest level of
care. Each is a standard duplex apartment with a full
kitchen/laundry, standard bath, and attached parking
space. There is no special handicap access. In contrast,
high-rise residents eat in common. There is handicap access
and handicap-adapted bathroom facilities. Light house-
cleaning and laundry services are provided. Limited out-
patient medical care, including medication supervision, is
available on site. All AFV residents benefit from on-campus
security and restricted public access.
The FHS subjects represented 57.3% of potentially
eligible participants and did not differ significantly from
nonrecruited residents with regard to age, sex, or baseline
level of care. Informed consent was obtained before their
evaluations. The University of Texas Health Science Center
at San Antonio’s institutional review board approved this
protocol.
Participants were evaluated at three separate points
over 3 years (mean interval  standard deviation 5 18.2 
3.6 months). As functional status declined over time, frail
subjects were interviewed in their own residences. Collat-
eral informants were used when available.
The attrition rate across the first two waves was 10.2%
cases/y (4.0% to death, 5.5% to refusal, and 0.4%
untestable at follow-up), but unforeseen budget restrictions
limited Wave 3 data collection to a subset of the surviving
cohort (n 5 137 (25.0%)). These represented the surviving
fraction (71.0%) of 193 FHS subjects who were randomly
selected at baseline to receive a detailed formal neuropsy-
chological test battery. Thus, the attrition was unrelated to
clinical condition and should satisfy the ‘‘missing at
random’’ constraint on the analyses (see below).
Clinical Variables
Functional status and comorbid medical conditions were
assessed using the Older Adults Resources Scale (OARS).9
The OARS is a structured clinical interview that provides
self-reported information on activities of daily living
(ADLs), instrumental activities of daily living (IADLs),
health history, healthcare usage, and current medications.
The OARS IADL scale is scored inversely relative to
disability, with lower scores indicating higher levels of
disability. ADL and IADL are often highly skewed.
Summing them into a single disability index following the
method of a reported study can ameliorate this,10 but this
did not appreciably affect the results, because ADLs had
little variance in this sample.11 Moreover, IADL perfor-
mance is more transparent than a combined disability
index. Thus, only IADL data are reported here. The number
of self-reported medical conditions was summed to create a
proxy for comorbid disease.
347
Depressive symptoms were assessed using the short
Geriatric Depression Scale (GDS).12,13 GDS scores range
from 0 to 15. Higher scores are worse. A cutpoint of 6 to 7
best discriminates clinically depressed from nondepressed
elderly.
Predictor Variables
The Executive Interview
The Executive Interview (EXIT25)8 provides a standard-
ized clinical ECF assessment. Items assess verbal fluency,
design fluency, frontal release signs, motor/impulse control,
imitation behavior, and other clinical signs associated with
frontal system dysfunction. It takes 15 minutes and can be
administered by nonmedical personnel. Interrater reliability
is high (r 5 0.90). It correlates well with other ECF
measures including the Wisconsin Card Sorting Task
(r 5
0.54), Trail Making Part B (r 5 0.64), Lezak’s Tinker
Toy test (r 5 0.57), and the Test of Sustained Attention
(Time, r 5 0.82; Errors, r 5 0.83). EXIT25 scores range
from 0 to 50. High scores indicate impairment. A score of
10 reflects the 5th percentile for young adults. Scores of 15
or higher suggest clinically significant ECF impairment.
The Mini-Mental State Examination
The Mini-Mental State Examination (MMSE)6,14 is a well-
known and widely used test for screening cognitive
impairment. Scores range from 0 to 30. A score of 28 is
the median for normal octogenarians with more than 12
years of education.14,15 Scores below 24 reflect cognitive
impairment.14 The MMSE has no items that are specifically
addressed to ECF5 and may underestimate cognitive
impairment in frontal system disorders.16 The MMSE was
chosen for this study not only because it is a familiar and
widely used measure, but also because it has been the
starting point for many other widely used global cognitive
measures, including those employed as dementia screening
measures in epidemiological studies.
Analysis
Data in the third wave were available for approximately
30% of the eligible subjects. Rather than discarding
information from the first two waves by limiting the
analysis to cases with complete data, Full Information
Maximum Likelihood (FIML) methods were used to
address this problem.17,18 FIML uses the entire observed
data matrix to estimate parameters with missing data. In
contrast to list-wise or pair-wise deletion, FIML yields
unbiased parameter estimates and preserves the overall
power of the analysis.18,19 This analysis was performed
using AMOS software (Small Waters Corp., Chicago, IL).
FIML can accommodate significant data loss, provided
the data are ‘‘missing at random.’’20 Because the vast
majority of missing data was from Wave 3, and because the
subjects observed at Wave 3 represent the survivors from a
random subset of the original sample, the missing data
points should satisfy this constraint. As a test of this
assumption, the baseline cross-sectional age, education,
sex, EXIT25, GDS, and MMSE scores were examined for
those with complete data at Wave 3 versus those lost to
follow-up. Attrition had no significant effects on these
variables (by analysis of variance: df (1, 546), all P 5 ns).
348 ROYALL ET AL.
The survival rates (mean observation time 5 60.9  15.3
months) to the outcomes of death, next transition in level of
care, and nursing home placement were also retrospectively
tested for those with complete data versus those lost to
follow-up at Wave 3. Attrition did not affect survival to any
of these outcomes (by log rank test, all P 5 ns).
Latent-Growth-Curve Analysis
Data were submitted to latent-growth-curve (LGC) model-
ing. In contrast to multiwave auto-regressive models, which
estimate interindividual rates of change across measure-
ments, LGC models estimate the full trajectory of change
across each individual’s measurement points.21 This is a
well-documented strategy for assessing longitudinal rates of
change in the elderly.22 The first and second factor loadings
on the latent-growth parameter were fixed to 0 and 1,
respectively. The last time-point loading was freely esti-
mated from the data.
LGC Variables of Interest
To test the hypothesis that the rate of change in cognition is
associated with the rate of change in disability, associative
models were assembled in which two fully adjusted LGCs
were estimated simultaneously. The latent growth variables
(e.g., rate of change in IADLs (DIADL) and rate of change in
the predictor measure) reflect the mean rates of change in
functional status and cognition, respectively. The variances
about these means are also of interest because they indicate
whether there is heterogeneity among the individual growth
curves within the sample. Significant variation in the rate of
change in a latent growth variable indicates that there are
statistically significant interindividual differences in the
rates of change. Nonsignificant variation would indicate
that all subjects were following the same change trajectory.
DIADL was regressed on the rate of change in each
cognitive measure separately, while adjusting for baseline
IADLs and baseline cognition. This association was further
adjusted for baseline age, level of care, and comorbid
illnesses. The covariances between the latent variables and
covariates were also estimated.
Figure 1 depicts the final, fully adjusted model using the
rate of change in EXIT25 scores (DEXIT25) to predict the
rate of change in functional status. (The MMSE was tested
separately in an identical model.) To the left are latent
variables (IADL and EXIT25) representing the mean and
variance of the observed data at baseline. To the right are
the latent growth variables (DIADL and DEXIT25). A
unidirectional arrow represents the association between
DIADL and DEXIT25. This implies that the rate of change
in ECF causes the rate of change in functional status, rather
than vice versa.
Goodness of Fit
The fit of each individual growth model was compared with
a competing model of ‘‘no growth.’’ The validity of
structural models was assessed using three common test
statistics. A nonsignificant chi-square signifies that the data
are consistent with the model.23 A root mean square error
of approximation (RMSEA) of 0.05 or less indicates a close
fit to the data, with models up to 0.10 viewed as acceptable
fit.24 The comparative fit index (CFI) compares the specified
MARCH 2004–VOL. 52, NO. 3 JAGS
Baseline 0.26
IADL ∆IADL
−0.19
−0.35
0.14
Age
−0.18
−0.16
−0.57
−0.48 0.15 Level of care
0.31
−0.09
0.44 Disease
-0.21
Base line 0.32
∆EXIT25
EX IT 25
Figure 1. Latent growth curve model. Significant standardized
coefficients are shown. Bold associations stronger than r 5 0.30.
Dashed associations not significant at Po.05. EXIT25 5 execu-
tive interview; DEXIT25 5 rate of change in EXIT25; IADL 5
instrumental activities of daily living; DIADL 5 rate of change in
IADLs.
model with a model of no change.25 With values ranging
between 0 and 1, values below 0.95 would suggest model
misspecification. Values of 0.95 or greater indicate ade-
quate to excellent fit. All three fit statistics should be
simultaneously considered to assess the adequacy of the
models to the data.
RESULTS
Cross-Sectional Analysis
Baseline EXIT25, MMSE, and GDS scores were within
their normal ranges (Tables 1 and 2). The EXIT25 and
MMSE were moderately inversely correlated (r 5 –0.36,
Po.001), but the EXIT25 was more sensitive to the
cognitive impairments in this sample. One hundred
seventy-five participants (32.0%) failed the EXIT25 with
a score of 15 or higher. In contrast, only n 5 36 (6.6% of the
sample) failed the MMSE with a score less than 24. One
hundred forty-nine (85.1%) of those who failed the
EXIT25 passed the MMSE. The mean GDS score was well
below its clinical threshold. Only 12 individuals scored
higher than 6 out of 15 on the GDS. Four hundred fifty-five
(83.2%) subjects estimated their health as good or
excellent.
Baseline EXIT25 and MMSE scores were significantly
correlated with baseline functional status (IADL scores)
(EXIT25IADL: r 5
0.48, Po.001; MMSEIADL: r 5
0.50, Po.01). Both associations remained significant after
adjusting for age, comorbid illness, and level of care (both
Po.01).
Longitudinal Analysis
Table 2 shows the descriptive statistics for the EXIT25 and
MMSE. The observed mean values for the EXIT25
increased (deteriorated). The means for the MMSE and
IADLs showed small decreases (deterioration). Clinically
significant ECF impairment (mean EXIT25 415/50) ap-
peared by the second follow-up. In contrast, clinically
JAGS MARCH 2004–VOL. 52, NO. 3 DECLINING EXECUTIVE CONTROL PREDICTS CHANGE IN FUNCTIONAL STATUS 349

Table 1. Baseline Clinical Features of Cases Table 2. Longitudinal Descriptive Statistics for the Study
Variables
Variable Value
Variable Mean  Standard Deviation Range
Age 77.9  4.9
Female, % 58.3 Executive Interview
Education, years, mean  SD 15.1  2.4 1 12.50  4.64 1–32
At lowest level of care, % 46.3 2 13.70  5.29 1–31
Living alone, % 27.8 3 15.15  6.77 0–32
Doctors’ visits in previous 3.8  4.1 Mini-Mental State
6 months, mean  SD Examination (MMSE)
Reporting, % 1 27.92  2.06 19–30
Glaucoma 18.3 2 27.77  2.28 13–30
Arthritis 61.2 3 27.16  2.79 15–30
Hypertension 41.9 Instrumental activities
Coronary heart disease 22.5 of daily living (IADLs)
Adult-onset diabetes mellitus 5.6 1 13.33  1.60 2–14
Stroke 6.3 2 12.95  2.18 0–14
Thyroid illness 17.7 3 12.70  2.33 3–14
Good or excellent self-rated health 82.3
Geriatric Depression Scale score, 1.6  1.9 Note: Higher scores on EXIT25 indicate impairment; lower scores on MMSE
short version and IADL indicate impairment. Five hundred forty-four cases had complete data
at baseline, versus 468 at Time 2 and 137 at Time 3.
Instrumental activities of daily living 13.3  1.7
score, mean  SD (best 5 14.0)
Needing assistance, %
Telephone use 3.4
trend in MMSE scores, and an approximate log or leveling
Transportation 12.7 trend in IADL ratings over time.
Shopping 10.6 The mean baseline and rate-of-change estimates are
Cooking 7.7 shown at the bottom of the table. The average baseline
Housework 18.8 scores estimated by the models were practically identical to
Medications 4.8 the observed baseline values shown in Table 2. There was
Finances 6.1 also significant variability around these means. Similarly,
Using prostheses 12.0 there was significant rate of change in all three variables and
significant variation around their respective rates of change.
 DEXIT25 averaged 0.89 per year. This can be
From the Older Adults Resources Scale.
SD 5 standard deviation. interpreted as an average increase of 0.89 EXIT25 points
per year of follow-up. The MMSE decreased 0.13 points per
year. IADLs decreased (deteriorated) by 0.50 points per
significant MMSE impairment (mean o24/30) never appea- year.
red during the 3 years of observation. Nonetheless, the Next, LGC models of change in the cognitive measures
sample as a whole suffered a slight decrease (deterioration) were regressed onto the LGC model of change in IADLs.
in mean IADL scores over the course of the study. Both models show acceptable to excellent fit (EXIT25:
Tables 3 and 4 present adjusted LGC models of w2 5 22.3, df 5 14, P 5.07, CFI 5 0.999, RMSEA 5 0.03;
DEXIT25, the rate of change in MMSE scores (DMMSE), MMSE: w2 5 11.9, df 5 14, P 5.62, CFI 5 0.999, RMSEA
and DIADL. The first row of Table 3 presents the fit o0.01). DEXIT25 scores were significantly associated with
estimates, and the second row presents the corresponding DIADL (b 5
0.36, standard error (SE) 5 0.16, Po.001).
parameter estimates for three separate longitudinal growth This association remained significant after adjusting for
curve models. Each model demonstrated acceptable to age, baseline comorbid disease, level of care, baseline
excellent fit to the data and was significantly different from disability, and baseline EXIT25 (b 5
0.37, SE 5 0.18,
a corresponding model of ‘‘no change.’’ Model loadings Po.001). The relative effects of DEXIT25 and its covariates
suggested a linear trend in EXIT25 scores, an accelerated on the rate of change in functional status can be estimated

Table 3. Latent Growth Curve Models of the Rate of Change in Executive Interview (EXIT25), Mini-Mental State
Examination (MMSE), and Instrumental Activities of Daily Living (IADL)
EXIT25 MMSE IADL

Model Fit Chi-Square (Degrees of Freedom) P-value [Comparative Fit Index] Root Mean Square Error of Approximation

Change models 3.2 (1) .07 [.999] .046 3.7 (1) .06 [.999] .060 2.7 (1) .10 [.999] .056
No change models 85 (4) .00 [.969] .192 33 (4) .00 [.995] .116 114 (4) .00 [.978] .222
350 ROYALL ET AL.
Table 4. Parameter Estimates for Latent Growth Curve Models of the Rate of Change in Executive Interview (DEXIT25),
Mini-Mental State Examination (DMMSE), and Instrumental Activities of Daily Living (DIADL)
DEXIT25
Parameter Estimate of
Change Models Mean  Standard Error Variance (Standard Error)
Baseline score 12.48  0.20w 12.03 (1.23)w
Annual rate of change 0.89  0.16w 2.91 (0.83)w

Po.05.
w
Po.001.
by examining their standardized coefficients in Figure 1.
The effect of DEXIT25 on DIADL was stronger than that of
age and more important than that of comorbid disease,
baseline physical function, or level of care. In contrast,
DMMSE was significantly associated with DIADL only in
an unadjusted model (b 5 2.33, SE 5 1.05, Po.05).
To test the assumption of causality, DIADL was
regressed on baseline EXIT25 (b 5
0.04, SE 5 0.006,
P 5.001) and DEXIT25 on baseline IADL (b 5
0.24,
SE 5 0.15, P 5.09) in separate post hoc models. The fact
that the former model is significant, whereas the latter is
not, is consistent with the hypothesized direction of
causality (that DEXIT25 causes observed changes in
disability).
DISCUSSION
A high prevalence of ECF impairment has previously been
reported in affluent, well-educated, noninstitutionalized
elderly retirees with normal MMSE scores.26 Moreover, in
the AFV, the EXIT25 is a significant independent correlate
level of care.27 The current analysis goes beyond previous
work to demonstrate that the deterioration in ECF during
normal aging is also independently associated with lon-
gitudinal declines in functional status. In this study, the
effect of EXIT25 on DIADL was the strongest of the
variables modeled.
There are several potential limitations to the results.
First, there was considerable loss of data in the third
observation. This was due to administrative reasons, not
simple attrition, and satisfies the ‘‘missing at random’’
constraint for valid FIML applications. As a check to these
findings, DEXIT25 and DMMSE were also modeled as
predictors of change in functional status under a variety of
alternative conditions, including as a simple change score
(IADL at T3
IADL at T1) and cross-lag regression (IADL
at T14IADL at T24IADL at T3). Each of these models
was performed with FIML and case-wise deletion (data not
shown). Under all four conditions, the DEXIT25 was a
significant independent predictor of change in functional
status; DMMSE was not. Although these results are
qualitatively similar to those presented in this paper, the
current model takes full advantage of the information
available in the data set. This allowed for the modeling of
changes in cognition and disability as LGCs. The resulting
model provides a better goodness of fit and explains more
variance in DIADL than any of the four models described
above. Nonetheless, the findings should be considered
MARCH 2004–VOL. 52, NO. 3 JAGS
DMMSE DIADL
27.94  0.09w 2.51 (0.24)w 13.33  0.07w 1.94 (0.36)w

0.13  0.07 0.17 (0.08)
0.50  0.08)w 0.97 (0.23)w
provisional until they can be replicated in future lon-
gitudinal samples.
Second, the MMSE is known to exhibit ceiling effects in
cross-sectional studies, and the IADL and MMSE exhibited
skewed distributions in this data set. Log and square root
transformations of these variables were tried, but this had
no appreciable effects on the results and would be much
more difficult to interpret than the untransformed models
reported here. In addition, it is not known that DMMSE
suffers from ceiling effects. Several studies (reviewed below)
find significant group effects on DMMSE in longitudinal
studies of community-dwelling older adults. The rates of
3-year change in MMSE scores in this sample ranged
from
23 to 17 points.
Third, the IADL variable is a convenient but meager
substitute for the direct measurement of functional status.
The clinical significance associated with the mean rate of
change in IADL ratings observed (
0.5 points/y) is not
clear. It would roughly translate into the loss of one self-
reported IADL every 2 years. Although self-reported
clinical symptoms might be considered suspect when
obtained from executively impaired subjects, the EXIT25
has also been shown to be a significant cross-sectional
correlate of objective and performance-based functional
measures.26–28
Nevertheless, it is important to distinguish between
fixed and random effects when interpreting LGC models.
Traditional regression examines only the fixed effects on
group means. Nonetheless, there is significant variability
about the mean baseline IADL in this sample and DIADL
(Table 3). These random effects represent interindividual
differences within the FHS sample with regard to baseline
IADL performance and individual DIADLs. Table 2
demonstrates a broad range of observed IADL scores at
baseline, despite the high mean performance. The strong
significant associations between EXIT25, IADL, DEXIT25,
and DIADL in an adjusted random effects model suggests
that the associations between these variables should be
valid across a broad range of individual functional
capacities and rates of change in these capacities.
The rate of change in ECF observed in this study is
likely to have substantial clinical implications. For example,
ECF as measured with the EXIT25 is strongly associated
with forensic capacities, such as the ability to execute an
advance directive29 or give informed consent for medical
care.30,31 These abilities may also suffer as a consequence of
the decline in ECF observed.
Similarly, the EXIT25 has previously been shown to dis-
tinguish each level of care in comparable CCRCs.8,26,27,32
JAGS MARCH 2004–VOL. 52, NO. 3 DECLINING EXECUTIVE CONTROL PREDICTS CHANGE IN FUNCTIONAL STATUS
The average separation between levels of care is about 7.0
EXIT25 points. By the third observation, mean EXIT25
scores had increased by almost 3.0 points and were already
in the impaired range. The linear deterioration in ECF
observed suggests that these subjects will eventually
succumb to increases in their levels of care.
Nevertheless, the significant variability about the
DEXIT25 parameter in Table 3 suggests that there is
heterogeneity across the individual DEXIT25. Subsets of
AFV residents may be deteriorating at faster or slower rates.
For example, after age 80, DEXIT25 is faster than reported
here for the entire FHS sample.26 Moreover, Figure 1
suggests that ECF may substantially mediate the association
between age and functional status. Baseline IADL and
DIADL are more strongly associated with baseline EXIT25
and DEXIT25, respectively, than with age. Moreover, age
itself is more strongly related to baseline EXIT25 and
DEXIT25 than to baseline IADL or DIADL. Age is
moderately associated with baseline IADL, but neither
age nor baseline IADL is strongly associated with DIADL
independently of its association with the EXIT25. Thus, it
appears that ECF mediates a significant fraction of age’s
total effect on change in IADL.
A growing literature suggests that the cognitive deficits
associated with normal aging are related to frontal systems
impairment.33 Aging has been associated with dispropor-
tionate frontal atrophy,34 frontal hypometabolism by single
photon emission computed tomography,35 and a dysex-
ecutive pattern of cognitive test performance.36–38 The
frontal atrophy associated with normal aging may be
regionally specific. Mesiofrontal regions are affected more
than dorsofrontal. Orbitofrontal regions seem relatively
spared. The EXIT25 has been associated with mesiofrontal
system pathology, particularly left-sided lesions.39,40 Simi-
lar lesions are associated with impaired performance on
verbal fluency measures, apathy, and depression, all of
which are increasingly common in advanced old age.
These data do not address the mechanisms by which
increasing age leads to a decline in ECF and hence
functional status, but two recent longitudinal studies have
applied LGC models to cognitive test scores in aging
twins.41,42 As in this study, cognitive measures showed
significant variability with regard to baseline distributions
and interindividual rates of change, but the variability in
cross-sectional baseline cognitive test performance ap-
peared to be much more heritable than the rates of change
in those measures. This suggests that the significant
variability about the rates of interindividual change in
EXIT25 scores in this study may be environmentally
determined and thus more amenable to intervention or
prevention. In this regard, it is interesting to note that
physical activity in old age appears to be specifically
associated with measures of executive control43 and that
aerobic exercise has a beneficial effect on age-related frontal
cortical volume by functional magnetic neuroimaging.44
In summary, a high prevalence of ECF impairment was
observed in this sample of initially noninstitutionalized
elderly retirees. ECF deficits progress over time, even in the
absence of an abnormal MMSE score, and are significantly
related to changes in disability. Current dementia case
finding is insensitive to ECF and may underreport the
incidence and prevalence of cognitive-related functional
351
decline. Thus, ECF represents a relatively unexplored
domain for investigation into and intervention on the
functional declines associated with normal aging.
ACKNOWLEDGMENTS
The authors wish to acknowledge the important coopera-
tion and support they received from the Air Force Villages.
REFERENCES
1. Royall DR, Lauterbach EC, Cummings JL et al. and the Committee on
Research of the American Neuropsychiatric Association Executive control
function: A review of its promise and challenges to clinical research. J Neuro-
psychiatry Clin Neurosci 2002;14:377–405.
2. Cummings JL. Frontal-subcortical circuits and human behavior. Arch Neurol
1993;50:873–880.
3. Royall DR. Executive cognitive impairment. A novel perspective on dementia.
Neuroepidemiology 2000;19:293–299.
4. Folstein M. Mini-Mental and son. Int J Geriatr Psychiatry 1998;13:285–294.
5. Folstein M, Folstein S, McHugh PR. ‘Mini-Mental State’. A practical method
for grading the cognitive state of patients for the clinician. Psychiatr Res
1975;12:189–198.
6. American Psychiatric Association. Diagnostic and Statistical Manual of
Mental Disorders, 4th Ed. Washington, DC: American Psychiatric Press, 1994.
7. Kraemer HC, Yesavage JA, Taylor JL et al. How can we learn about develop-
mental processes from cross-sectional studies, or can we? Am J Psychia-
try 2000;157:163–171.
8. Royall DR, Mahurin RK, Gray K. Bedside assessment of executive dyscontrol:
The Executive Interview (EXIT25). J Am Geriatr Soc 1992;40:1221–1226.
9. Fillenbaum GG. Validity and Reliability of the Multidimensional Functional
Assessment Questionnaire. In: Duke University Center for the Study of Aging
and Human Development. The OARS Methodology. Durham, NC: Duke
University, 1978.
10. Spector WD, Fleishman JA. Combining activities of daily living with instru-
mental activities of daily living to measure functional disability. J Gerontol
B Psychol Sci Soc Sci 1998;53B:S46–S57.
11. Stone RI, Murtaugh CM. The elderly population with chronic functional
disability: Implication for home care eligibility. Gerontologist 1990;30:491–
496.
12. Sheikh JI, Yesavage JA. Geriatric Depression Scale (GDS). Recent evidence and
development of a shorter version. Clin Gerontologist 1986;5:165–173.
13. Maixner SM, Burke WJ, Roccaforte WH et al. A comparison of two
depression scales in a geriatric assessment clinic. Am J Geriatr Psychiatry 1995;
3:60–67.
14. Folstein MF, Folstein SE, McHugh PR et al. Mini-Mental State Examination
User’s Guide. Odessa, FL: Psychological Assessment Resources, 2001.
15. Crum RM, Anthony JC, Bassett SS et al. Population based norms for the Mini-
Mental State Examination by age and education level. JAMA 1993;269:2386–
2391.
16. Royall DR, Polk M. Dementias that present with and without posterior
cortical features: An important clinical distinction. J Am Geriatr Soc 1998;46:
98–105.
17. Arbuckle JL. Full information estimation in the presence of incomplete data.
In: Marcoulides GA, Schumaker RE, eds. Advanced Structural Equation
Modeling: Issues and Techniques. Mahwah, NJ: Lawrence Erlbaum Associ-
ates, 1996, pp 243–277.
18. Wothke W. Longitudinal and multi-group modeling with missing data. In:
Little TD, Schnabel KU, Baumert J, eds. Modeling Longitudinal and Multiple
Group Data: Practical issues, Applied Approaches, and Specific Examples.
Mahwah, NJ: Lawrence Erlbaum Associates, 1999, pp 219–240.
19. Schafer J. Analysis of Incomplete Multivariate Data. New York, NY:
Chapman & Hall, 1997.
20. McArdle J, Hamagami F. Modeling incomplete longitudinal and cross
sectional data using latent growth structural models. Exp Aging Res
1992;18:145–166.
21. Willet J, Sayer A. Using covariance structure analysis to detect correlates and
predictors of individual change over time. Psychol Bull 1994;116:363–381.
22. Jacqmin-Gadda H, Fabrigoule C, Commenges D et al. A 5-year longitudinal
study of the Mini-Mental State Examination in normal aging. Am J Epidemiol
1997;145:498–506.
23. Bollen KA, Long JS. Testing Structural Equation Models. Thousand Oaks, CA:
Sage Publications, Inc., 1993.
24. Browne M, Cudeck R. Alternative ways of assessing model fit. In: Bollen KA,
Long JS, eds. Testing Structural Equation Models. Thousand Oaks, CA: Sage
Publications, Inc, 1993, pp 136–162.
352 ROYALL ET AL.
25. Bentler PM. Comparative fit indexes in structural models. Psychol Bull
1990;107:238–246.
26. Royall DR, Cabello M, Polk MJ. Executive dyscontrol. An important factor
affecting the level of care received by elderly retirees. J Am Geriatr Soc
1998;46:1519–1524.
27. Royall DR, Chiodo LK, Polk MJ. Correlates of disability among elderly
retirees with ‘sub-clinical’ cognitive impairment. J Gerontol A Biol Sci Med Sci
2000;55A:M541–M546.
28. Palmer R, Royall DR, Chiodo LK et al. Growth curve models of longitudinal
change in ECF. Relationship to functional status. [Abstract] Gerontology
2001;7(Suppl. 1):50.
29. Royall DR, Cordes J, Polk MJ. Executive control and the comprehension of
medical information by elderly retirees. Exp Aging Res 1997;23:301–313.
30. Holzer JC, Gansler DA, Moczynski NP et al. Cognitive functions in the in-
formed consent evaluation process. J Acad Psychiatry Law 1997;25:531–540.
31. Dymek MP, Atchison P, Harrell L et al. Competency to consent to medical
treatment in cognitively impaired patients with Parkinson’s. Neurology
2001;56:17–24.
32. Royall DR, Mahurin RK, True J et al. Executive impairment among the
functionally dependent. Comparisons between schizophrenic and elderly
subjects. Am J Psychiatry 1993;150:1813–1819.
33. Royall DR, Vicioso B. Aging overview. In: Aminoff M, Daroff RB, eds.
Encyclopedia of the Neurological Sciences, Vol. 1. San Diego, CA: Academic
Press, 2003, pp 53–57.
34. Coffey CE, Wilkinson WE, Parashos IA et al. Quantitative cerebral anatomy of
the aging human brain: A cross-sectional study using magnetic resonance
imaging. Neurology 1992;42:527–536.
MARCH 2004–VOL. 52, NO. 3 JAGS
35. Kuhl DE. The effects of normal aging on patterns of local cerebral glucose
utilization. Ann Neurol 1984;15:S133–S137.
36. Rinn WE. Mental decline in normal aging: A review. J Geriatr Psychiatry
Neurol 1988;1:144–158.
37. Hinkin C, Cummings JL, Van Gorp WG et al. Frontal/subcortical features of
normal aging: An empirical analysis. Can J Aging 1990;9:104–111.
38. Van Gorp WG, Mahler ME. Subcortical features of normal aging. In:
Cummings JL, ed. Subcortical Dementia. Oxford, UK: Oxford University
Press, 1990, pp 231–250.
39. Jobe T, Blend M, Sychra J et al. The Executive Interview (EXIT25) and
cerebral perfusion SPECT imaging in the assessment of the frontal lobes. Paper
presented at the Annual Meeting of the American Psychiatric Association,
New York, NY, May 1996.
40. Royall DR, Rauch R, Román GC et al. MRI findings associated with
impairment on the Executive Interview (EXIT25). Exp Aging Res 2001;27:
293–308.
41. Reynolds CA, Finkle D, Gatz M et al. Sources of influence on rate of cognitive
change over time in Swedish twins: An application of latent growth models and
rate of change in cognitive functioning in Danish twins aged 70 years and
older. Exp Aging Res 2002;28:407–433.
42. McGue M, Christensen K. Heritability of level of and rate of change in
cognitive functioning in Danish twins aged 70 years and older. Exp Aging Res
2002;28:435–451.
43. Colcombe SJ, Kramer EF. Fitness effects on cognitive function in older adults:
A meta-analytic study. Psychol Sci 2003;14:125–130.
44. Colcombe SJ, Erickson KI, Raz N et al. Aerobic fitness reduces brain tissue loss
in aging humans. J Gerontol A Biol Sci Med Sci 2003;58A:M176–M180.