Exploring Effects of Type 2 Diabetes on Cognitive Functioning in Older Adults

Neuropsychology
© 2009 American Psychological Association
2009, Vol. 23, No. 1, 1–9
0894-4105/09/$12.00
DOI:10.1037/a0013849
Exploring Effects of Type 2 Diabetes on Cognitive Functioning
in Older Adults
Sophie E. Yeung, Ashley L. Fischer, and Roger A. Dixon
University of Alberta
Type 2 diabetes may be associated with exacerbated aging-related declines in cognitive neuropsycho-
logical performance. The authors examined whether such effects are systematic (i.e., broadly distributed
across domains or domain-speci?c) or moderated by age (i.e., varying across age within older adults).
The authors assembled recent cross-sectional data from the Victoria Longitudinal Study (VLS) Sample 3
(Wave 1; initial n
570; initial age
53–90 years). Using a comprehensive, multidimensional spectrum
of cognitive neuropsychological tests, the authors examined performance differences by diabetes status
(diabetes group vs. healthy controls) and age (young-old vs. old-old). Our results showed that healthy
controls signi?cantly outperformed the diabetes group only on markers of executive functioning and
speed. Notably, the diabetes-related effects were robust across the two late-life age groups. Future
research examining longitudinal changes is recommended.
Keywords: Type 2 diabetes, cognitive aging, executive function, speed
Type 2 diabetes is a chronic metabolic condition characterized
these issues with a relatively healthy and generally cognitively
by abnormally high blood glucose levels as a result of insuf?cient
intact sample of 53–90 year-old adults tested on multiple domains
usage of insulin. Formerly known as Non-Insulin Dependent Di-
of cognitive neuropsychological performance. Speci?cally, our
abetes Mellitus or adult-onset diabetes, its prevalence signi?cantly
database includes multiple indicators of the key domains repre-
increases across adulthood, typically affecting individuals over the
sented (often separately) in the literature: episodic and semantic
age of 40 years (Votey & Peters, 2005). Recent estimates on the
memory, neurocognitive speed, executive functioning, ?uency,
prevalence of diabetes (Type 1 and Type 2) have indicated diag-
and global cognitive competence. This comprehensive approach is
nosis rates for adults over age 60 at about 12% in Canada (Health
designed to contribute to resolving some of the mixed patterns of
Canada, 2002) and 20% in the United States (National Institute of
results across studies varying in cognitive domains, measures, and
Health, 2005). Approximately 90% of these cases are Type 2.
age groups represented.
Associated with Type 2 diabetes are increased risk of hyperten-
Although the general trend is for diabetes-related de?cits in
sion, stroke, and cerebrovascular disease (e.g., Awad, Gagnon, &
performance, discrepant results are common (Nilsson, 2006). First,
Messier, 2004; Messier, 2005; Reunanen, Kangas, Martikainen, &
verbal episodic memory is typically (but not uniformly) affected in
Klaukka, 2000). These potential comorbidities have been shown to
diabetes patients (Arvanitakis, Wilson, Bienias, Evans, & Bennett,
affect neural integrity and cognition, especially when coexistent
2004; Messier, 2005; Ryan & Geckle, 2000; Wahlin, Nilsson, &
with diabetes (Hassing, Hofer, et al., 2004). Recent literature has
Fastbom, 2002), extending well-known patterns of normal aging-
reported a relationship between diabetes and an earlier or acceler-
related decline (e.g., Dixon et al., 2004). These de?cits have been
ated decline in cognition (e.g., Awad et al., 2004; Hassing, Grant,
seen predominantly in adults over the age of 70 (Messier, 2005),
et al., 2004; Hassing et al., 2003), including a twofold increase in
and for measures of immediate verbal memory (e.g., word list
the risk of dementia (Nilsson, 2006).
recall; Awad et al., 2004). Notably, Hassing, Grant, et al. (2004)
Few studies have examined whether adverse cognitive effects of
found no diabetes-related cognitive performance de?cits at base-
diabetes are broad or selective across domains, or whether such
line but observed accelerated longitudinal decline in episodic
effects differ across a broad age band of older adults. We explore
memory (and speed) for the diabetes group. Second, diabetes-
related slowing has been observed with a variety of speeded tasks,
especially those measuring basic reaction time or perceptual speed
Sophie E. Yeung, Ashley L. Fischer, and Roger A. Dixon, Department
(e.g., Arvanitakis, Wilson, & Bennett, 2006; Awad et al., 2004;
of Psychology, University of Alberta.
Fontbonne, Berr, Ducimetie`re, & Alpe´rovitch, 2001; Messier,
This research is supported by a grant (R37 AG008235) from the Na-
2005). However, Messier’s (2005) review indicated that less than
tional Institutes of Health (National Institute on Aging) to Roger Dixon,
half the included studies actually reported diabetes-related slow-
who is also supported by the Canada Research Chairs program. Sophie
ing. We include indicators of three main domains of neurocogni-
Yeung is now at Department of Psychology, Simon Fraser University,
tive speed: reaction time, perceptual speed, and a unique set of
Burnaby, Canada. The authors express gratitude to Jill Jenkins and Terry
semantic speed measures. Third, selected measures of executive
Perkins for technical support, to Åke Wahlin for suggestions on a previous
functioning have produced diabetes-related performance de?cits in
draft of the manuscript, and to the staff and participants of the VLS.
some (but not all) studies (Awad et al., 2004; Messier, 2005; Ryan
Correspondence concerning this article should be addressed to Roger A.
& Geckle, 2000; Stewart & Liolitsa, 1999). Because executive
Dixon, Department of Psychology, P-217 Biological Sciences Building,
University of Alberta, Edmonton, Alberta Canada. E-mail: rdixon@
functioning may involve multiple underlying processes or dimen-
ualberta.ca
sions (de Frias, Dixon, & Strauss, 2006; Miyake, Freidman,
1

---

2
YEUNG, FISCHER, AND DIXON
Emerson, Witzki, & Howerter, 2000), it may be especially sus-
Two additional methodological features should be noted. First,
ceptible to task-related selection effects in special population re-
our diabetes patients report that their cases are relatively mild or
search (Nilsson, 2006). Our database taps multiple aspects of
moderate (97.56%), and that their conditions are controlled by oral
executive functioning (i.e., inhibition, shifting, speed). Fourth,
medication (39.02%), insulin (7.32%), diet and exercise (24.39%),
some studies have focused on general measures of global cognition
or a variety of other combinations (19.53%). Accordingly, this
with diabetes-related de?cits both observed (e.g., Hassing et al.,
diabetes group supplements those in the literature comprised of
2003) and not found (e.g., Arvanitakis, Wilson, & Bennett, 2006;
relatively severe patients from nursing homes or health clinics
Fontbonne et al., 2001). We use and report global cognition results
(e.g., Arvanitakis et al., 2004). By examining community-dwelling
descriptively only.
volunteers to a large-scale project, the present sample may be more
The exact neuroanatomical or neurochemical effects of Type 2
representative of current early or well-controlled Type 2 diabetes
diabetes on cognitive performance are relatively unknown. One
populations in North America. Arguably, for more severe cases,
study suggests that frontal structures may be affected by diabetes
cognitive de?cits may be attributed to disease severity, neurolog-
sequelae and may therefore be associated with occasionally ob-
ical sequelae, or multiple comorbid conditions (Nilsson, 2006).
served de?cits in episodic memory recall, verbal ?uency, and
Second, because previous studies are each characterized by rela-
executive functioning (Wahlin et al., 2002). Additionally, reduced
tively few (and nonoverlapping) cognitive measures, our analyses
volumes of the amygdala and the hippocampus in diabetes patients
are conducted at the level of each measure, but clustered within
may underlie de?cits associated with learning and memory (den
cognitive domains. Regarding the ?rst research question, we ex-
Heijer et al., 2003). A recent report on MRI abnormalities and
pected signi?cant diabetes-related differences in performance on
cognitive changes found substantial white matter lesions and sub-
episodic memory, verbal ?uency, and neurocognitive speed, but
cortical atrophies in Type 2 diabetes patients, suggesting an accel-
not semantic memory. The mixed results in research on executive
erated rate of age-related structural changes (Manschot et al.,
functioning in diabetes and normal aging do not support a strong
2006). In addition to possible diabetes-related neurological mech-
hypothesis. Regarding age differences, we hypothesized (in the
anisms, several comorbid health conditions may confound (or
absence of previous research) that the group differences in cogni-
exacerbate) cognitive sequelae of diabetes. Prominent among these
tive performance will be more pronounced in the old-old adults.
comorbidities are neurological and psychiatric conditions, hyper-
tension, cardiovascular and cerebrovascular disease, and drug use
Method
(e.g., Arvanitakis, Wilson, Li, Aggarwal, & Bennett, 2006; Jacob-
son et al., 2007; Robertson-Tchabo, Arenberg, Tobin, & Plotz,
The Victoria Longitudinal Study (VLS) is a multicohort epide-
1986; Vanhanen et al., 1998; van Harten et al., 2007; Xu, Qiu,
miological study of biomedical, health, cognitive, and neurocog-
Wahlin, Winblad, & Fratiglioni, 2004). Accordingly, we applied a
nitive aspects of aging. Three independent samples of initially
series of cognitive and physical health-related exclusionary criteria
healthy older adults are followed at 3-year intervals (see Dixon &
to our group selections. Furthermore, because hypertension is
de Frias, 2004).
associated with brain and cognition changes in older adults (Elias,
Elias, Robbins, & Budge, 2004; Raz, Rodrigue, Kennedy, &
Participants
Acker, 2007) and especially among diabetes patients (Hassing,
Hofer, et al., 2004), systolic blood pressure was considered as a
The base participants were from the ?rst wave (2002– 03) of
covariate.
VLS Sample 3 (n
570, age range
53–90 years; M age
68.29
Given the considerable extent of con?icting results across most
years, SD
8.60). Data from an in-progress second wave are
neuropsychological domains, the general trends and precise effects
unavailable. A strict sequential procedure for selection and exclu-
of Type 2 diabetes on cognitive aging are not ?rmly delineated.
sion of participants in two groups (Type 2 diabetes and control)
The cross-study variability in results could stem from several
was adopted. Inclusion into the diabetes group was based on a
methodological differences, including (a) few and selective cog-
three-step diagnosis ?owchart, including required con?rmatory
nitive measures and (b) restricted or uncontrolled age-related sam-
information from all three sequential sources. First, all diabetes
pling of older adults. As recommended elsewhere (Brands et al.,
patients self-reported (a) a formal diabetes diagnosis, (b) an adult
2007), this study presents a comprehensive range of neuropsycho-
onset age (over 31), and (c) treatment or control practices (i.e.,
logical indicators, thus promoting both cross-study and within-
diet, exercise, oral medication, insulin, or a combination). Second,
sample comparisons. Speci?cally, we use recent cross-sectional
the actual objective medications of the diabetes patients were
data derived from the Victoria Longitudinal Study (VLS), a Ca-
checked concurrently for the presence of relevant drugs (e.g.,
nadian sample comprised of relatively healthy and non-demented
metformin, glyburide, tolbutamide, gliclazide, and pioglitazone
individuals. We explore these issues in two conventional age
hydrochloride). Third, all surviving diabetes patients were con-
groups, young-old adults (YO, 53–70 years old) and old-old adults
tacted three years after the present testing for con?rmation of
(OO, 71–90 years old), in order to test whether diabetes exerts its
self-reported diabetes diagnoses. Although the VLS does not have
cognitive effects in earlier or later late life (Wahlin et al., 2002).
access to additional diabetes-con?rming medical information (i.e.,
After accounting for recommended health confounds, we examine
blood glucose level above 6.0mmol/L at baseline, or elevated
two research questions. First, are there group differences between
HbA1c level), our three-step diagnostic procedure goes beyond the
those with Type 2 diabetes and healthy controls in performance
frequently used and validated self-report classi?cations (e.g.,
across a range of cognitive neuropsychological measures? Second,
Arvanitakis, Wilson, Li, et al., 2006; Connolly, Unwin, Sherriff,
are observed group differences affected by age differences within
Bilous, & Kelly, 2000; Kriegsman, Penninx, van Eijk, Boeke, &
this older adult sample?
Deeg, 1996; McNeely & Boyko, 2004; Midthjell, Holmen, Bjørndal,

---

DIABETES AND COGNITIVE FUNCTIONING
3
& Lund-Larsen, 1992; Reunanen et al., 2000), and it is consistent
poorer health relative to a perfect state, F(1, 461)
34.13, p
with diagnostic criteria currently used in the literature (Arvani-
.000, partial
2
0.069 (M
2.39, SD
0.86; M
1.69,
D
D
C
takis, Wilson, Li, et al., 2006; Gregg et al., 2000; Luchsinger,
SD
0.71), and relative to others their own age, F(1,
C
Tang, Stern, Shea, & Mayeux, 2001). Of the original 570
461)
17.33, p
.000, partial
2
0.036 (M
2.00,
D
adults, 48 were identi?ed as potential Type 2 diabetes patients.
SD
0.78; M
1.50, SD
0.68). Given the chronic illness
D
C
C
As a result of diagnosis procedures, we selected a provisional
for which they were selected into this study, these perceptions
diabetes group (n
44; M
69.33 years, SD
7.64) and a pool
accurately re?ect their different overall health status.
of nondiabetes control participants (n
522; M
68.13 years,
We further characterized the groups using VLS physiological
SD
8.68).
tasks (see MacDonald, Dixon, Cohen, & Hazlitt, 2004). Body mass
Next, we implemented four sets of standard exclusionary crite-
index (BMI; kg/m2) was signi?cantly higher in diabetes partici-
ria. First, we con?rmed that no participants had been previously
pants, F(1, 459)
29.99, p
.000, partial
2
0.061
diagnosed with Alzheimer’s disease or vascular dementia. Second,
(M
30.20, SD
4.65; M
26.46, SD
3.97). Eight
participants scoring less than 26 on the Mini-Mental Status Exam-
D
D
C
C
readings of blood pressure (mmHg) were averaged across four
ination (MMSE; Folstein, Folstein, & McHugh, 1975) were re-
testing sessions. Whereas no group differences were observed for
moved (diabetes n
0, control n
12). Third, based on the VLS
diastolic blood pressure (M
77.35, SD
9.20; M
75.06,
intake health inventory, we inspected the following clusters of
D
D
C
SD
8.89), mean systolic blood pressure in the diabetes group
potential comorbid diseases: (a) neurological conditions (i.e.,
C
stroke, Parkinson’s disease, epilepsy, and head injury), (b) cardio-
was signi?cantly higher than controls, F(1, 444)
14.38, p
vascular disease (i.e., heart trouble, hypertension, hypotension, and
.000, partial
2
0.031 (M
134.14, SD
15.27; M
D
D
C
atherosclerosis), (c) other related health conditions (i.e., spinal
125.21, SD
13.69). Potential diabetes-related visual acuity
C
condition and thyroid complications), and (d) psychiatric condi-
complications were assessed using the Close Vision task (Snellen
tions (i.e., depression, alcohol dependence, drug dependence, use
fractions), but no signi?cant differences were observed. Compar-
of antidepressant or antipsychotic medication). Participants from
isons of audition (using a test of audio acuity, dB) also showed no
the diabetes group (n
18) who self-rated that they had moder-
signi?cant group differences. As noted earlier, we assessed global
ately serious or very serious cases of any of these four clusters of
cognitive competence using the standard 18-item MMSE (Folstein
conditions were selected for follow-up and examined individually
et al., 1975). Scores were generally high and clinically insigni?-
for potential adverse cognitive effects. We reasoned that comor-
cant for both groups. Overall, the diabetes participants were aware
bidities in aging are common, so we ensured that the present
of their chronic condition and calibrated their personal health
diabetes patients would be removed only if they had both a
evaluation accordingly, but they were not substantially inferior in
potentially cognitively impairing health condition and a demon-
other health, sensory, and physiological characteristics. Final de-
strated cognitive de?cit. If any of the 18 selected diabetes patients
mographic and physiological characteristics of the age (YO, 53–70
scored 1 SD (or more) below the diabetes group mean on any of
years; old-old, OO, 71–90 years) X diabetes status (diabetes and
the cognitive reference tests (word recall, story memory, vocabu-
controls) groups are reported in Table 1.
lary, simple reaction time) they were removed from the sample.
Accordingly, an additional n
3 diabetes participants (two with
Measures
severe heart trouble and one with a severe spinal condition, and all
with cognitive impairment) were excluded. One moderate Parkin-
Episodic memory.
First, the VLS word list recall task con-
son’s disease participant performed outside the acceptable range
sisted of the immediate free recall of two standardized lists of 30
only on reaction time, so we excluded this individual for the
words, each including six words from each of ?ve taxonomic
speeded tasks. Fourth, because of relatively plentiful control group
categories (e.g., Dixon et al., 2004; Hultsch, Hertzog, Dixon, &
members, our procedures were simpler. Control participants (n
Small, 1998). The score was the average of the number of words
86) were excluded if they met at least one of the following criteria:
recalled from the two lists. Second, the Rey Auditory Verbal
(a) indication of very serious case on any of the above exclusion
Learning Test (RAVLT; see Vakil & Blachstein, 1993) required
health conditions, or (b) indication of a moderately serious history
that a word list (15 nouns; List A) was read to the participant,
of stroke or Parkinson’s disease, whether or not their behavioral
followed by free recall, in each of ?ve trials (Trial A1-A5). Next,
performances were adversely affected.
for Trial B, an interference list of 15 different nouns was followed
For the ?nal sample (n
465), the diabetes group (n
41; 23
by free recall. Next, the words from List A (Trial A6) were
women, 18 men) ranged from 55 to 81 years old (M
68.59 years,
recalled. Raw scores from Trial B (acquisition) and Trial A6
SD
7.16) and the control group (n
424; 294 women, 130 men)
(retention) were used. Third, we administered six standardized struc-
ranged from 53 to 90 years old (M
67.84 years, SD
8.50). The
turally equivalent story memory tests (with approximately 300 words
two groups were very similar in general age proportions and
and 60 propositions within 24 sentences), used in the VLS (e.g.,
speci?cally so in the oldest decade (80 –90 years, with similar
Dixon, Hertzog, Friesen, & Hultsch, 1993; Dixon et al., 2004). Av-
proportions of diabetes (4.95%) and control (4.87%) participants).
erage gist recall was computed and converted to proportions.
An ANOVA (n.s.) on years of education showed that the diabetes
Semantic memory.
First, the vocabulary test consisted of 54
group (M
15.12 years, SD
3.44) and the control group
multiple-choice questions from the Educational Testing Service kit
(M
15.33 years, SD
2.92) were high and similar. Participants’
of factor-referenced cognitive tests (Ekstrom, French, Harman, &
ratings of their health on 5-point scales (1
very good health, 5
Dermen, 1976). The score was the total number of correct items.
very poor health) were generally within the very good to fair range.
Second, fact recall was measured with two different 40-item tests
However, as compared to controls, diabetes participants perceived
of general information (e.g., from history, arts, sports) derived

---

4
YEUNG, FISCHER, AND DIXON
Table 1
Characteristics of Participants in Diabetes and Control Groups According to Age Group
Diabetes
Controls
Variables
Young-old n
24
Old-old n
17
Young-old n
273
Old-old n
151
Age
63.64 (4.57)
75.59 (3.07)
62.45 (4.46)
77.59 (4.36)
Range
55 – 69
71 – 80
53 – 70
71 – 90
Gender (% female)
66.7
41.2
72.5
63.6
Age at diagnosis
55.26 (10.09)
66.75 (5.88)
Duration of condition
8.12 (8.13)
8.54 (6.07)
Education
14.63 (3.55)
15.82 (3.24)
15.60 (2.84)
14.83 (3.00)
Perfect health
2.29 (0.91)
2.53 (0.80)
1.62 (0.70)
1.81 (0.70)
Relative health
2.08 (0.78)
1.88 (0.78)
1.50 (0.70)
1.50 (0.66)
BMI (kg/m2)
30.72 (4.05)
29.51 (5.41)
26.61 (4.10)
26.19 (3.73)
Systolic blood pressure (mmHg)
131.33 (15.19)
137.95 (14.98)
123.24 (13.55)
128.70 (13.27)
Diastolic blood pressure (mmHg)
77.76 (8.92)
76.80 (9.82)
75.97 (8.84)
73.44 (8.78)
Close vision (right eye; Snellen)
11.95 (0.21)
11.19 (2.99)
11.77 (1.63)
11.55 (2.11)
Close vision (left eye; Snellen)
10.96 (0.21)
10.25 (2.75)
10.80 (1.34)
10.61 (1.98)
Smoking status (%)
Never
41.7
41.2
47.6
32.5
Previously
50.0
58.8
49.1
59.6
Currently
8.3
0.0
3.3
7.9
Alcohol use (%)
54.2
70.6
89.0
91.4
Antihypertensive drugs (% use)
54.2
47.1
12.8
25.8
Note.
Means and standard deviations are presented as M (SD). Age, age at diagnosis, duration of condition and education are presented in years. Age is
calculated from the participant’s initial testing session. Perfect health represents participant self-rating relative to a perfect state, whereas relative health
re?ects self-rating relative to peers of a similar age based on a scale from 1–5 (1
very good, 5
very poor).
from a normed battery (Nelson & Narens, 1980). The two scores
(read the color name when printed in incongruent ink), divided by
were averaged and converted to a percentage of correct (see
the initial response latency in Part A alone ([Part C – Part A]/Part
Hultsch et al., 1998).
A). Fourth, the Color Trails 2 (CT-2) task measured response
Verbal ?uency.
The VLS ?uency tests have three standard
inhibition without the in?uence of language. Numbers from 1–25
parts (Hultsch et al., 1998), including subtests of opposites, ?gures
were randomly arranged twice on a page, once in pink-colored and
of speech, and similarities. Participants used a limited time to write
once in yellow-colored circles. Participants were instructed to
as many correct words as possible. Raw scores for each subtest
connect the numbers from 1–25 in proper sequence, during which
were recorded.
they must alternate from one color to the next. The time to
Executive functioning.
Four tests of executive functioning
complete the task was measured in seconds.
have been validated, normed, and analyzed in the VLS and else-
Neurocognitive speed.
Five standard speed tests have been
where (see Bielak, Mansueti, Strauss, & Dixon, 2006; de Frias et
used in previous VLS research in aging and special populations
al., 2006). First, for the Hayling sentence completion test, initiation
(e.g., Dixon et al., 2007; Hultsch et al., 1998). Four were comput-
speed and response inhibition were tested (Burgess & Shallice,
erized tests presented using a 386 IBM-compatible computer that
1997). In two sections requiring speeded responses, participants
controlled stimulus timing and presentation. Participants re-
were read 15 sentences, each with the last word missing. Whereas
sponded by pressing designated keys on a response console, and
the goal of the ?rst section was to respond with a word that swiftly
performance was recorded in milliseconds (ms). Two of these were
completed the sentence, the goal of the second section was to
semantic speed tests (lexical decision, sentence veri?cation) and
suppress an initial response by providing a word disconnected to
two were reaction time tests. The ?fth test measured perceptual
the sentence. Recorded were response latencies (ms) and overall
speed (digit symbol substitution). First, for lexical decision, par-
standard scores based on correct responses from the two sections.
ticipants read a string of ?ve to seven letters and indicated whether
Second, the Brixton spatial anticipation test measured abstraction
the letters produced an English word (e.g., island vs. nabion). The
of logical rules (Andre´s & Van der Linden, 2000). Each page of a
scores were the mean latencies across the 60 trials (composed
56-page booklet contained two rows of ?ve circles, with one of
of 30 words and 30 nonwords). Second, for sentence veri?cation,
the 10 circles ?lled in blue. Participants selected which circle they
participants read 50 individually presented sentences and indicated
anticipated would be ?lled on the next page based on the pattern
whether each sentence was plausible or nonsensical (e.g., “the tree
they deduced. A standard score out of 10 (ranging from 1
fell to the ground with a loud crash” vs. “the pig gave birth to a
impaired to 10
very superior) was derived. Third, for the
litter of kittens this morning”). Two outcome measures were used:
three-part Stroop test, participants were required to inhibit their
the average latency (correct responses) of the 50 trials and the
automatic verbal responses in reading printed words and instead
percentage of errors. Outlier latency values greater or less than
name the color in which each word was printed. A standard
three standard deviations from the mean were removed. Third, for
interference index calculated the difference in response latency
the simple reaction time (SRT) test, a warning stimulus was
between Part A (name the color of the printed dots) and Part C
presented in the middle of a screen, followed by a signal stimulus

---

DIABETES AND COGNITIVE FUNCTIONING
5
to which participants pressed a key. Ten practice trials were
tion has been previously documented (Dixon & de Frias, 2004;
followed by 50 test trials. Ten trials were presented at a time, with
Hultsch et al., 1998). In order to optimize comparison with previ-
randomly alternating intervals separating the warning and signal
ous reports of speci?c cognitive effects, a series of two-way
stimuli (500, 625, 750, 875, and 1000 ms). Each interval was
univariate analyses of covariance (ANCOVA) was conducted. The
presented ?ve times across the trials. The score was the average
covariate was systolic blood pressure, given (a) observed group
latency of the 50 trials. Fourth, for the choice reaction time
differences as noted above, and (b) recent research showing effects
(CRT4) test, a 2
2 grid corresponding with the key arrangement
of hypertension on diabetes-cognition relationships (Hassing,
on a response console was presented. Each block had 10 trials,
Hofer, et al., 2004). Diabetes status (diabetes or control) and age
wherein the participant attended to four plus signs, one of which
group (YO or OO) were the ?xed factors, with each cognitive
transformed into a square, to which the matching key was pressed.
measure as the dependent variable. Regarding statistical signi?-
Following 10 practice trials, the average latency across 20 test
cance, to partially adjust for multiple ANCOVAs, we focused on
trials was calculated. Fifth, the Wechsler Adult Intelligence Scale-
alpha levels of p
.01. For limited archival, exploratory, and
Revised Digit Symbol Substitution task (DSS; Wechsler, 1991) has
comparative purposes, we also note statistical trends up to p
.05
been used widely to measure perceptual speed (Hassing, Grant, et
(Nilsson, 2006). Analyses were conducted using SPSS Ver-
al., 2004; MacDonald, Hultsch, Strauss, & Dixon, 2003). In a 90-s
sion 15.0 statistical software.
period, participants used a coding key of nine numbers paired with
speci?c symbols to ?ll rows of empty numbered test boxes. The
Results
score was the number of correctly transcribed items.
See Table 2 for basic age and diabetes status group results. No
signi?cant interaction effects were found, indicating that the cog-
Procedure and Analyses
nitive effects of Type 2 diabetes are not moderated by late-life age
VLS protocol requires that all measures are administered in the
differences. Signi?cant main effects are described below, empha-
same sequence to all participants. For each wave, actual testing
sizing the diabetes status factor.
occurs across four sessions about one week apart (for participant
Episodic memory.
In contrast to several previous ?ndings, no
comfort and testing ef?ciency). Detailed VLS procedural informa-
signi?cant differences were observed between diabetes and control
Table 2
Mean Cognitive Performance for Main Effects of Diabetes Status and Age Group
Diabetes status group
Age group
Cognitive measure
Diabetes n
41
Controls n
424
p
Young-old n
297
Old-old n
168
p
Global cognition
MMSE
28.75 (1.08)
28.87 (1.04)
.701
29.06 (0.91)
28.52 (1.18)
.002
Episodic memory
Word recall
16.01 (5.37)
17.63 (4.32)
.055
18.50 (3.79)
15.72 (4.92)
.000
RAVLT acquisition
5.70 (2.26)
6.31 (4.46)
.531
6.77 (5.19)
5.38 (1.76)
.063
RAVLT retention
9.93 (3.42)
10.25 (4.92)
.871
10.94 (5.38)
8.98 (3.23)
.031
Story memory
35.27 (9.72)
39.10 (10.21)
.071
41.07 (9.61)
34.75 (10.01)
.001
Semantic memory
Vocabulary
41.70 (6.48)
42.45 (6.62)
.500
42.12 (6.15)
42.82 (7.33)
.365
Fact recall
50.25 (16.52)
51.43 (15.36)
.456
52.31 (15.32)
49.63 (15.51)
.273
Verbal ?uency
Opposites
12.98 (5.20)
13.66 (4.49)
.390
13.96 (4.44)
12.97 (4.70)
.143
Figures of speech
9.33 (3.12)
9.81 (3.16)
.271
10.18 (3.16)
9.05 (3.02)
.060
Similarities
14.70 (6.22)
15.79 (6.04)
.302
16.25 (6.07)
14.73 (5.92)
.036
Executive functioning
Hayling
5.16 (1.44)
5.84 (1.21)
.003
6.03 (1.05)
5.34 (1.42)
.002
Brixton
4.43 (1.88)
5.11 (2.12)
.110
5.49 (1.95)
4.29 (2.16)
.005
Color Trails 2 (s)
106.09 (36.05)
91.08 (29.10)
.013
83.05 (22.61)
108.65 (34.21)
.000
Stroop Test
1.39 (0.88)
1.17 (0.61)
.072
1.10 (0.60)
1.36 (0.69)
.030
Semantic speed
Lexical dec. (ms)
1550.26 (606.68)
1315.34 (494.66)
.015
1257.10 (397.21)
1474.46 (639.08)
.029
S.V. (% error)
5.25 (4.48)
3.99 (3.25)
.038
3.76 (3.16)
4.70 (3.70)
.047
S.V. (ms)
4074.65 (1071.75)
3451.27 (1207.54)
.008
3290.98 (1021.15)
3883.52 (1404.35)
.040
Reaction time
SRT (ms)
370.59 (76.80)
343.90 (79.06)
.097
328.83 (68.18)
376.36 (87.54)
.001
CRT4 (ms)
951.34 (168.56)
886.14 (171.53)
.114
825.90 (123.94)
1004.48 (184.38)
.000
Perceptual speed
DSS raw score
46.25 (12.42)
50.43 (10.84)
.093
53.69 (9.98)
43.77 (9.92)
.000
Note.
Means and standard deviations are presented as M (SD). SRT, CRT4, Lexical dec., S.V., and DSS represent simple reaction time, choice reaction
time, lexical decision, sentence veri?cation, and Digit Symbol Substitution, respectively. Actual ns vary across cognitive tasks due to missing data. All ps
are for main group effects.

---

6
YEUNG, FISCHER, AND DIXON
groups on any outcome measure of episodic memory. However, in
relatively well-educated volunteer sample and reported MMSE
concordance with normal aging patterns, YO groups performed
results for descriptive purposes only. Second, previous studies
signi?cantly better than OO groups on two episodic measures,
have employed older adults from different and mixed age bands
including word list recall, F(1, 443)
18.32, p
.000, partial
(e.g., under age 70, Fontebonne et al., 2001; over age 70, Hassing,
2
0.040, and story memory recall, F(1, 443)
11.49, p
.001,
Grant, et al., 2004). Examining whether systematically sampling
partial
2
0.025, with a trend for RAVLT retention, F(1,
along the age index would affect diabetes-related cognitive effects,
443)
4.69, p
.031, partial
2
0.010.
we found commonly observed late-life cross-sectional age effects
Semantic memory.
No signi?cant group differences were ob-
but no interactions of age group with diabetes status. Diabetes-
served on either measure.
related cognitive effects may be generally constant across age, at
Verbal ?uency.
No signi?cant diabetes-related group differ-
least for the current ranges of duration and severity.
ences were observed.
Regarding the executive functioning results, two measures pro-
Executive functioning.
The control group performed signi?-
duced signi?cant performance differences in favor of the controls:
cantly better than the diabetes group on the Hayling task, F(1,
that is, the Hayling (involving speed and inhibition) and Color
434)
8.86, p
.003, partial 2
0.020, with a trend for Color
Trails 2 (involving speed and shifting). In addition, the group
Trails 2, F(1, 440)
6.26, p
.013, partial
2
0.014. No
means were nonsigni?cant but in the same direction for the other
diabetes-related group differences were evident on the Brixton or
two executive tests: that is, the Brixton (involving rule attainment
the Stroop test. Signi?cant age group differences were observed
and planning) and Stroop interference index (involving inhibition
for all four measures of executive functioning, including the Hay-
of responses). The fact that executive functioning tests are asso-
ling, F(1, 434)
9.32, p
.002, partial
2
0.021; Color
ciated with inconsistent patterns of diabetes-related de?cits (Mess-
Trails 2, F(1, 440)
21.73, p
.000, partial 2
0.047; and the
ier, 2005) is not unexpected given the broad and sometimes mul-
Brixton, F(1, 438)
7.98, p
.005, partial
2
0.018.
tidimensional nature of the construct and the variable tasks used to
Semantic speed.
Healthy controls performed signi?cantly
measure associated processes in different populations (e.g., de
faster than participants with diabetes on the sentence veri?cation
Frias et al., 2006; Zhang, Han, Verhaeghen, & Nilsson, 2007).
task, F(1, 442)
7.17, p
.008, partial 2
0.016, with a trend
Because studies on diabetes and aging may produce executive
for a lower error rate, F(1, 442)
4.32, p
.038, partial
functioning results that are in part a function of the tests used, a
2
0.010. Healthy controls also displayed a trend for faster
theoretically convergent pattern has been elusive. The present
performance on the lexical-decision task, F(1, 442)
5.96, p
results contribute to potential consolidation of the executive func-
.015, partial
2
0.013. Regarding age, the YO group tended to
tioning de?cit associated with milder forms of diabetes in that the
outperform the OO group on the lexical-decision task, F(1,
tasks for which we found group differences required a contribution
442)
4.83, p
.029, partial
2
0.011, and the percent error,
of speed. Conceivably, simpler tasks measuring less speed-inten-
F(1, 442)
3.98, p
.047, partial
2
0.009, and latency
sive aspects may be less sensitive to milder diabetes-related ef-
measure, F(1, 442)
4.25, p
.040, partial
2
0.010, of the
fects. Some of the inconsistency in the literature may be due to the
sentence veri?cation task.
fact that multiple aspects of executive functioning are differen-
Reaction time.
There were no observed diabetes group differ-
tially represented in neuropsychological test batteries and perhaps
ences on SRT or CRT4. As expected, YO participants were faster
even applied unsystematically across age and disease severity
than OO participants on both the SRT, F(1, 440)
12.11, p
continua. The notion that earlier effects may be observed in tasks
.001, partial 2
0.027, and CRT4, F(1, 440)
47.25, p
.000,
requiring rapid performance of executive-demanding processes
partial
2
0.097.
may be tested in future research using samples of broader clinical
Perceptual speed.
No signi?cant diabetes-related effect was
severity and with longitudinal follow-ups. Such de?cits may cas-
observed between the diabetes group and the healthy controls on
cade throughout the executive functioning domain as diabetes
the DSS task. YO participants performed signi?cantly faster than
progresses and the rates of aging- and disease-related structural
OO participants, F(1, 442)
30.42, p
.000, partial 2
0.064.
changes in the brain accelerate (Manschot et al., 2006).
The results were selective also within the neurocognitive speed
domain. Of the three sets of speed measures, no group perfor-
Discussion
mance differences were found for either traditional reaction time
A growing literature has examined the extent and depth of
(SRT and CRT4) or perceptual speed tasks (e.g., Cosway, Stra-
potential cognitive effects of Type 2 diabetes in older adults. We
chan, Dougall, Frier, & Dreary, 2001; Fontbonne et al., 2001;
contribute to this literature by examining simultaneously a broad
Hassing, Grant, et al., 2004). Instead, the diabetes group performed
range of cognitive neuropsychological measures as performed by
most prominently and signi?cantly slower than the control group
both YO and OO healthy controls and relatively mild diabetes
on the sentence veri?cation task, with a consistent trend for lexical
patients. Our ?ndings revealed signi?cant group differences within
decision. Perhaps tasks requiring quick and precise processing of
select domains, most consistently in speed-intensive measures of
new verbal information may be sensitive markers for detecting
executive functioning and semantic speed. As suggested by Nils-
cognitive de?cits in relatively milder diabetes patients (Arvani-
son (2006), not all aspects of cognition may be equally or coinci-
takis, Wilson, & Bennett, 2006; Nilsson, Fastbom, & Wahlin,
dentally affected by Type 2 diabetes, at least in relatively mild-
2002), a conclusion not available without the presence of semantic
to-moderate cases.
speed tasks (as well as the Hayling). Normal aging-related slowing
Two results are brie?y noted and not further discussed. First,
of performance is well known, and accelerated (or inconsistent)
performance on global cognition (i.e., MMSE) for both groups was
slowing may signal early cognitive impairment or Alzheimer’s
high and virtually identical. We excluded low performers from a
disease (Dixon et al., 2007; Rapp & Reischies, 2005). Future

---

DIABETES AND COGNITIVE FUNCTIONING
7
clinical and longitudinal research may test the possibility that
ments of the three themes. Among the strengths of this study is the
speed-intensive tasks involving semantic operations may be dif-
unusually broad age range of older adults, with which we con-
ferentially sensitive markers in older diabetes patients. However,
?rmed that aging-related diabetes effects may be invariant across
given the novelty of semantic speed assessments in diabetes liter-
young-old and old-old age groups. Second, we accessed VLS
ature, replication studies would be useful.
Sample 3 archives for self-report and objective diagnostic infor-
Two conspicuous null ?ndings require brief comment, as they
mation, background and comorbid health indicators, and an exten-
are relevant to previous literature and complement the two ob-
sive battery of cognitive neuropsychological domains. Third, with
served de?cits described above. First, although verbal episodic
the broad cognitive neuropsychological battery, we were able to
memory tends to be more frequently impaired in both healthy older
detect a pro?le of robust effects in select domains, most notably
adults (e.g., Dixon et al., 2004) and older diabetes patients (see
executive functioning and speed. Given the current comprehensive
Nilsson, 2006), we observed no signi?cant diabetes-related differ-
cross-sectional baseline, future longitudinal research—with the
ences for any of our three verbal episodic tasks. The provisional
VLS and other studies— can examine potentially differential de-
importance of this null result is weakened by the informal obser-
cline patterns. Fourth, we covaried systolic blood pressure in our
vation that the group means are generally in the expected direction
analyses, as there is evidence to suggest that elevated blood pres-
(see Table 2). This implies both methodological (e.g., role of
sure increases cognitive decline independent of diabetes (Elias,
covariates, statistical power) and clinical directions for future
D’Agostino, Elias, & Wolf, 1995; Waldstein, 2003). The impor-
research. We were able to check one of these issues, namely, the
tance of considering hypertensive effects is highlighted, as it may
role of a comorbidity covariate. Post hoc ANOVAs (without
differentially contribute to cognitive decline and in?ate the differ-
systolic BP as covariate) revealed a tendency for two memory
ences attributed to diabetes.
tasks (and the two additional speeded tasks) to produce trends
A ?rst limitation re?ects several unmodi?able characteristics of
( p
.05) in the expected direction. Therefore, active diabetes-
our sample: It is volunteer-based, from a smaller urban population
related comorbidities (e.g., hypertension) may be a contributing
in Canada, predominantly Caucasian, relatively well-educated,
factor to whether domain-speci?c cognitive de?cits are observed
initially selected on global cognitive intactness, and with generally
(Saxby, Harrington, McKeith, Wesnes, & Ford, 2003; Hassing,
available health care. Thus, results are not necessarily representa-
Hofer, et al., 2004; Waldstein, 1995). Future clinical research may
tive of the entire Canadian or western population, but they may
examine whether episodic memory de?cits are not leading indica-
generalize to a large and growing population of relatively healthy
tors of early diabetes-related cognitive effects, but markers of
aging preboom and boomer populations. Future studies of greater
further progression of the disease and expanded neurological
diversity are encouraged. Second, although provisional diagnoses
involvement.
of Type 2 diabetes were based on a combination of commonly
A second set of null ?ndings merits brief comment. In contrast
used self-report, follow-ups, and objective medication data, more
to some previous research (see Nilsson, 2006), we did not ?nd
precise biological information (e.g., HbA1c levels, Fasting Blood
signi?cant group differences in performance on any measure of
Glucose levels) is currently unavailable in the VLS. Conceivably,
verbal ?uency. As with executive functioning, the ?uency tests
some nondiagnosed cases may be present in the control group, but
used in various studies differ considerably in procedure and cog-
their unlikely presence would have rendered a more conservative
nitive demands. For example, whereas our measures required
test of the hypotheses. Third, although well characterized, a larger
relatively abstract thinking in ?nding opposites, ?gures of speech,
diabetes group would have been preferable. We noted earlier,
and similarities between words, the ?uency tasks used by Wahlin
however, that our diabetes group (n
41) is well within the range
et al. (2002) required more basic letter-word ?uency. Future re-
of comparable neuropsychological studies (with ns of 20 – 41).
search comparing more levels of complexity and demand in ?u-
Moreover, as a proportion of the VLS parent sample, it is similar
ency could be helpful in delineating the extent of the de?cit, its
to Canadian population expectations.
relationship to severity of the disease, and the possibility of some
Overall, this study contributes to the literature with a compre-
early preserved ?uency skill. Similarly, expected null ?ndings for
hensive neuropsychological battery and a broad age range with
semantic memory were observed for the typical vocabulary mea-
which to explore de?cits associated with relatively mild Type 2
sure and this pattern was extended to include the previously
diabetes in older adults The results both qualify and extend those
untested fact memory task. Broadening the range of diabetes
of previous reports, particularly (but differentially) in the domains
severity, exploring further comorbidities, and conducting longitu-
of speed, executive functioning, and episodic memory. Given the
dinal follow-ups will begin to clarify the timing of and extent to
modern western lifestyle, associated health risks, and growing
which ?uency and semantic memory may be affected by diabetes
populations of older adults, Type 2 diabetes will likely increase as
(Arvanitakis et al., 2004; Hassing, Grant, et al., 2004).
a common aging-related challenge to neurobiological and cogni-
Overall, our interpretation has emphasized three important
tive health. Future studies examining longitudinal trends in neuro-
themes: (a) classi?cation and diagnosis clarity (e.g., disease iden-
psychological sequelae of diabetes will help determine whether
ti?cation and severity, comorbidities, exclusionary criteria), (b)
different patterns of cognitive decline occur across both health con-
possible temporal ordering of diabetes-related cognitive neuropsy-
dition (diabetes group vs. healthy controls) and neuropsychological
chological outcomes as the disease progresses, and (c) potential
domain (executive functioning, cognitive speed, episodic memory).
theoretical and clinical value of comprehensive cross-sectional and
follow-up longitudinal assessments. If executive resources and
References
speed may be most prominently compromised relatively early in
Andre´s, P., & Van der Linden, M. (2000). Age-related differences in
Type 2 diabetes, con?rmatory and complementary evidence should
supervisory attentional system functions. Journal of Gerontology: Psy-
be observed across studies with various combinations of the ele-
chological Sciences, 55B, 373–380.

---

8
YEUNG, FISCHER, AND DIXON
Arvanitakis, Z., Wilson, R. S., & Bennett, D. A. (2006). Diabetes mellitus,
Fontbonne, A., Berr, C., Ducime`tiere, P., & Alpe´rovitch, A. (2001).
dementia, and cognitive function in older persons. The Journal of
Changes in cognitive abilities over a 4-year period are unfavorably
Nutrition, Health, and Aging, 10, 287–291.
affected in elderly diabetic subjects. Diabetes Care, 24, 366 –370.
Arvanitakis, Z., Wilson, R. S., Bienias, J. L., Evans, D. A., & Bennett,
Gregg, E. W., Yaffe, K., Cauley, J. A., Rolka, D. B., Blackwell, T. L.,
D. A. (2004). Diabetes mellitus and risk of Alzheimer disease and
Narayan, K. M., et al. (2000). Is diabetes associated with cognitive
decline in cognitive function. Archives of Neurology, 61, 661– 666.
impairment and cognitive decline among older women? Archives of
Arvanitakis, Z., Wilson, R. S., Li, Y., Aggarwal, N. T., & Bennett, D. A.
Internal Medicine, 160, 174 –180.
(2006). Diabetes and function in different cognitive systems in older
Hassing, L. B., Grant, M. D., Hofer, S. M., Pedersen, N. L., Nilsson, S. E.,
individuals without dementia. Diabetes Care, 29, 560 –565.
Berg, S., et al. (2004). Type 2 diabetes mellitus contributes to cognitive
Awad, N., Gagnon, M., & Messier, C. (2004). The relationship between
decline in old age: A longitudinal population-based study. Journal of the
impaired glucose tolerance, type 2 diabetes, and cognitive function.
International Neuropsychological Society, 10, 599 – 607.
Journal of Clinical and Experimental Neuropsychology, 26, 1044 –1080.
Hassing, L. B., Hofer, S. M., Nilsson, S. E., Berg, S., Pederson, N. L.,
Bielak, A. A. M., Mansueti, L., Strauss, E., & Dixon, R. A. (2006).
McClearn, G., et al. (2004). Comorbid type 2 diabetes mellitus and
Performance on the Hayling and Brixton tests in older adults: Norms and
hypertension exacerbates cognitive decline: Evidence from a longitudi-
correlates. Archives of Clinical Neuropsychology, 21, 141–149.
nal study. Age and Ageing, 33, 355–361.
Brands, A. M. A., van der Berg, E., Manschot, S. M., Biessels, G. J.,
Hassing, L. B., Johansson, B., Pedersen, N. L., Nilsson, S. E., Berg, S., &
Kappelle, L. J., De Haan, E. H., et al. (2007). A detailed pro?le of
McClearn, G. (2003). Type 2 diabetes mellitus and cognitive perfor-
cognitive dysfunction and its relation to psychological distress in pa-
mance in a population-based sample of the oldest old: Impact of comor-
tients with type 2 diabetes mellitus. Journal of the International Neu-
bid dementia. Aging, Neuropsychology, and Cognition, 10, 99 –107.
ropsychological Society, 13, 288 –297.
Health Canada. (2002). Diabetes in Canada (2nd ed.). Ottawa, Ontario,
Burgess, P. W., & Shallice, T. (1997). The Hayling and Brixton tests.
Canada: Health Canada.
Thurston, Suffolk, England: Thames Valley Test Company.
Hultsch, D. F., Hertzog, C., Dixon, R. A., & Small, B. J. (1998). Memory
Connolly, V., Unwin, N., Sherriff, P., Bilous, R., & Kelly, W. (2000).
change in the aged. New York: Cambridge University Press.
Diabetes prevalence and socioeconomic status: A population based
Jacobson, A. M., Musen, G., Ryan, C. M., Silvers, R. N., Cleary, P.,
study showing increased prevalence of type 2 diabetes mellitus in
Waberski, B., et al. (2007). Long-term effect of diabetes and its treat-
deprived areas. Journal of Epidemiology and Community Health, 54,
ment on cognitive function. The New England Journal of Medicine, 356,
173–177.
1842–1852.
Cosway, R., Strachan, M. W. J., Dougall, A., Frier, B. M., & Deary, I. J.
Kriegsman, D. M., Penninx, B. W., van Eijk, J., Boeke, J. P., & Deeg, D. J.
(2001). Cognitive function and information processing in type 2 diabe-
(1996). Self- reports and general practitioner information on the pres-
tes. Diabetic Medicine, 18, 803– 810.
ence of chronic diseases in community dwelling elderly: A study on the
de Frias, C. M., Dixon, R. A., & Strauss, E. (2006). Structure of four
accuracy of patients’ self-reports and on determinants of inaccuracy.
executive functioning tests in healthy older adults. Neuropsychology, 20,
Journal of Clinical Epidemiology, 49, 1407–1417.
206 –214.
den Heijer, T., Vermeer, S. E., van Dijk, E. J., Prins, N. D., Koudstaal, P. J.,
Luchsinger, J. A., Tang, M-X., Stern, Y., Shea, S., & Mayeux, R. (2001).
Hofman, A., et al. (2003). Type 2 diabetes and atrophy of medial
Diabetes mellitus and risk of Alzheimer’s disease and dementia with
temporal lobe structures on brain MRI. Diabetologia, 46, 1604 –1610.
stroke in a multiethnic cohort. American Journal of Epidemiology, 154,
Dixon, R. A., & de Frias, C. M. (2004). The Victoria Longitudinal Study:
635– 641.
From characterizing cognitive aging to illustrating changes in memory
MacDonald, S. W. S., Dixon, R. A., Cohen, A., & Hazlitt, J. E. (2004).
compensation. Aging, Neuropsychology, and Cognition, 11, 346 –376.
Biological age and 12-year cognitive change in older adults: Findings
Dixon, R. A., Garrett, D. D., Lentz, T. L., MacDonald, S. W. S., Strauss,
from the Victoria Longitudinal Study. Gerontology, 50, 64 – 81.
E., & Hultsch, D. F. (2007). Neurocognitive markers of cognitive
MacDonald, S. W. S., Hultsch, D. F., Strauss, E., & Dixon, R. A. (2003).
impairment: Exploring the roles of speed and inconsistency. Neuropsy-
Age-related slowing of digit symbol substitution revisited: What do
chology, 21, 381–399.
longitudinal age changes re?ect? Journal of Gerontology: Psychological
Dixon, R. A., Hertzog, C., Friesen, I. C., & Hultsch, D. F. (1993).
Sciences, 58, 187–194.
Assessment of intraindividual change in text recall of elderly adults. In
Manschot, S. M., Brands, A. M. A., van der Grond, J., Kessels, R. P. C.,
H. H. Brownwell & Y. Joanette (Eds.), Narrative discourse in neuro-
Algra, A., Kappelle, L. J., et al. (2006). Brain magnetic resonance
logically impaired and normal aging adults (pp. 77–101). San Diego:
imaging correlates of impaired cognition in patients with type 2 diabetes.
Singular.
Diabetes, 55, 1106 –1113.
Dixon, R. A., Wahlin, Å., Maitland, S. B., Hultsch, D. F., Hertzog, C., &
McNeely, M. J., & Boyko, E. J. (2004). Type 2 diabetes prevalence in
Ba¨ckman, L. (2004). Episodic memory change in late adulthood: Gen-
Asian Americans. Diabetes Care, 27, 66 – 69.
eralizability across samples and performance indices. Memory & Cog-
Messier, C. (2005). Impact of impaired glucose tolerance and type 2
nition, 32, 768 –778.
diabetes on cognitive aging. Neurobiology of Aging, 26S, S26 –S30.
Ekstrom, R. B., French, J. E. W., Harman, H. H., & Dermen, D. (1976).
Midthjell, K., Holmen, J., Bjørndal, A., & Lund-Larsen, P. G. (1992). Is
Manual for the kit of factor-referenced cognitive tests. Princeton, NJ:
questionnaire information valid in the study of a chronic disease such as
Educational Testing Service.
diabetes? The Nord-Trondelag diabetes study. Journal of Epidemiology
Elias, P. K., D’Agostino, R. B., Elias, M. F., & Wolf, P. A. (1995). Blood
and Community Health, 46, 537–542.
pressure, hypertension, and age as risk factors for poor cognitive per-
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., & Howerter,
formance. Experimental Aging Research, 21, 393– 417.
A. (2000). The unity and diversity of executive functions and their
Elias, P. K., Elias, M. F., Robbins, M. A., & Budge, M. M. (2004). Blood
contributions to complex “frontal lobe” tasks: A latent variable analysis.
pressure-related cognitive decline: Does age make a difference? Hyper-
Cognitive Psychology, 41, 49 –100.
tension, 44, 631– 636.
National Institute of Health. (2005). National diabetes statistics fact sheet:
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). Mini-mental state.
General information and national estimates on diabetes in the United
A practical method for grading the cognitive state of patients for the
States (NIH Publication No. 06 –3892). Bethesda, MD: U.S. Department
clinician. Journal of Psychiatric Research, 12, 189 –198.
of Health and Human Services.

---

DIABETES AND COGNITIVE FUNCTIONING
9
Nelson, T. O., & Narens, L. (1980). Norms of 300 general-information
Vanhanen, M., Koivisto, K., Kuusisto, J., Mykka¨nen, L., Helkala, E.,
questions: Accuracy of recall, latency of recall, and feeling-of-knowing
Ha¨nninen, T., et al. (1998). Cognitive function in an elderly population
ratings. Journal of Verbal Learning and Verbal Behavior, 19, 338 –368.
with persistent impaired glucose tolerance. Diabetes Care, 21, 398 – 402.
Nilsson, E. (2006). Diabetes and cognitive functioning: The role of age and
van Harten, B., Oosterman, J., Muslimovic, D., Potter van Loon, B.,
comorbidity. Unpublished doctoral dissertation, Karolinska Institutet,
Scheltens, P., & Weinstein, H. C. (2007). Cognitive impairment and
Stockholm.
MRI correlates in the elderly patients with type 2 diabetes mellitus. Age
Nilsson, E., Fastbom, J., & Wahlin, Å. (2002). Cognitive functioning in a
and Ageing, 36, 164 –170.
population- based sample of very old non-demented and non-depressed
Votey, S. R., & Peters, A. L. (2005). Diabetes mellitus, type 2- A review.
persons: The impact of diabetes. Archives of Gerontology and Geriat-
Retrieved December 5, 2006, from http://www.emedicine.com/emerg/
rics, 35, 95–105.
topic134.htm
Rapp, M. A., & Reischies, F. M. (2005). Attention and executive control
Wahlin, Å., Nilsson, E., & Fastbom, J. (2002). Cognitive performance in
predict Alzheimer disease in late life. American Journal of Geriatric
very old diabetic persons: The impact of semantic structure, preclinical
Psychiatry, 13, 134 –141.
dementia, and impending death. Neuropsychology, 16, 208 –216.
Raz, N., Rodrigue, K. M., Kennedy, K. M., &Acker, J. D. (2007). Vascular
health and longitudinal changes in brain and cognition in middle-aged
Waldstein, S. R. (1995). Hypertension and neuropsychological function: A
and older adults. Neuropsychology, 21, 149 –157.
lifespan perspective. Experimental Aging Research, 21, 321–352.
Reunanen, A., Kangas, T., Martikainen, J., & Klaukka, T. (2000). Nation-
Waldstein, S. R. (2003). The relation of hypertension to cognitive function.
wide survey of comorbidity, use, and costs of all medications in Finnish
Current Directions In Psychological Science, 12, 9 –13.
diabetic individuals. Diabetes Care, 23, 1265–1271.
Wechsler, D. (1991). Manual for the Wechsler Adult Intelligence Scale-
Robertson-Tchabo, E. A., Arenberg, D., Tobin, J. D., & Plotz, J. B. (1986).
Revised. New York: Psychological Corporation.
A longitudinal study of cognitive performance in noninsulin dependent
Xu, W. L., Qiu, C. X., Wahlin, Å, Winblad, B., & Fratiglioni, L. (2004).
(type 2) diabetic men. Experimental Gerontology, 21, 459 – 467.
Diabetes mellitus and risk of dementia in the Kungsholmen project: A
Ryan, C. M., & Geckle, M. (2000). Why is learning and memory dysfunc-
6-year follow-up study. Neurology, 63, 1181–1186.
tion in Type 2 diabetes limited to older adults? Diabetes/Metabolism
Zhang, Y., Han, B., Verhaeghen, P., & Nilsson, L.-G. (2007). Executive
Research and Reviews, 16, 308 –315.
functioning in older adults with mild cognitive impairment: MCI has
Saxby, B. K., Harrington, F., McKeith, I. G., Wesnes, K., & Ford, G. A.
effects on planning, but not on inhibition. Aging, Neuropsychology, and
(2003). Effects of hypertension on attention, memory, and executive
Cognition, 14, 557–570.
function in older adults. Health Psychology, 22, 587–591.
Stewart, R., & Liolitsa, D. (1999). Type 2 diabetes mellitus, cognitive
impairment and dementia. Diabetic Medicine, 16, 93–112.
Received October 15, 2007
Vakil, E., & Blachstein, H. (1993). Rey Auditory-Verbal Learning Test:
Revision received J