ARTICLE IN PRESS

Psychoneuroendocrinology (2007) 32, 1041–1051

Available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/psyneuen

Predictors of cognitive impairment in type 1 diabetes

Tom Brismara,b,, Liselotte Maurexb, Gerald Cooraya,b,
Lisa Juntti-Berggrenc,d, Per Lindströme, Karin Ekbergd,
Nils Adnerf, Sten Anderssonb

a
Department of Clinical Neurophysiology, Karolinska University Hospital (Solna), SE-17176 Stockholm, Sweden
b
Department of Clinical Neuroscience, Karolinska Institutet, SE-17177 Stockholm, Sweden
c
Department of Endocrinology and Diabetology, Karolinska University Hospital (Solna), SE-17176 Stockholm, Sweden
d
Department of Molecular Medicine and Surgery, Karolinska Institutet, SE-17177 Stockholm, Sweden
e
Department of Neurology, Karolinska University Hospital (Solna), SE-17176 Stockholm, Sweden
f
Department of Internal Medicine, South Hospital, SE-11883 Stockholm, Sweden

Received 21 February 2007; received in revised form 9 August 2007; accepted 9 August 2007

KEYWORDS Summary
Type 1 diabetes; A decline in cognitive function has been reported in type 1 diabetes, but its relation to
Cognitive function; different disease factors such as hypoglycemic events and peripheral neuropathy is
Hypoglycemia; controversial. The objective of the present study was to identify factors that are important
Neuropathy; for cognitive impairment in type 1 diabetes. A cross-sectional study was performed in adult
IGF; patients (N ¼ 150) with type 1 diabetes (duration 26.6711.4 years). Function in different
Human cognitive domains was evaluated by the same trained examiner, in order to eliminate
inter-rater variability. Peripheral nerve function was tested quantitatively. Predictors of
cognitive impairment were identified using multiple regression analysis. The major finding
was that long diabetes duration and young age of diabetes onset were the strongest
predictors of low scores in psychomotor speed, memory, processing speed, attention,
working memory, verbal ability, general intelligence, executive functions and a low global
score. The number of previous hypoglycemic events had no defined effect upon cognitive
functioning. Other significant predictors were low compound muscle action potential
(CMAP) (for visual perception–organization), old age (for visual–spatial ability), short
stature, high BMI and hypertension. Presence of retinopathy and long-term metabolic
control correlated with nerve conduction defects, but not with cognitive impairment.
Although a history of hypoglycemic events was not a predictor of cognitive impairment, we
cannot exclude the possibility that the influence of young age of diabetes onset depends on
the effect of hypoglycemic events early in life. The clinical relationships of cognitive

Corresponding author. Department of Clinical Neurophysiology, Karolinska University Hospital (Solna), SE-17176 Stockholm, Sweden.
Tel.: +46 8 51772030, Mobile: +46 707214123; fax: +46 8 51774415.
E-mail address: tom.brismar@ki.se (T. Brismar).

0306-4530/$ - see front matter & 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.psyneuen.2007.08.002
 ARTICLE IN PRESS
1042
impairment differ from those of peripheral neuropathy, indicating a different pathogen-
esis. The influence of diabetes duration, BMI, height, age and CMAP may suggest that loss
of the neuroprotective effects of insulin or insulin-like growth factors plays a role.
& 2007 Elsevier Ltd. All rights reserved.
1. Introduction
Several reports have indicated that diabetes may cause
cognitive dysfunction or an alteration in brain signals
related to cognitive function, and a recent meta-analysis
reported that patients with type 1 diabetes have a mild
decline in overall cognitive function (Brands et al., 2005).
Such cognitive impairment is heterogeneous with regard to
functional domains and considerably more difficult to assess
than complications in the peripheral nervous system. In
particular, the role of glycemic control is complex since
strict metabolic control is associated with more frequent
hypoglycemic events (Reichard et al., 1991; DCCT, 1996)
which are potentially harmful to the brain. Conversely, an
improved glucose control might prevent complications in the
central nervous system as demonstrated for other late
complications. This ‘dual effect’ of glycemic control was
addressed early on after the introduction of intensified
insulin therapy (Ryan et al., 1993).
The role of recurrent hypoglycemic events is controversial
which to some extent may be related to the difficulty in
evaluating this retrospectively and to different effects in
adults and children. Although several early cross-sectional
studies indicated an effect in adults (Wredling et al., 1990;
Langan et al., 1991; Sachon et al., 1992; Deary et al., 1993;
Lincoln et al., 1996) several large prospective studies
showed no effect of recurrent hypoglycemic events on
cognitive performance in adults (Reichard et al., 1991;
DCCT, 1996; Jacobson et al., 2007). Prospective studies in
children are conflicting showing both presence (Rovet and
Ehrlich, 1999; Northam et al., 2001) and absence (Schoenle
et al., 2002) of an effect. Another possible risk factor in
children is early disease onset (before age of 4 years) which
was related to impaired cognitive development (Rovet
et al., 1987; Northam et al., 1999; Schoenle et al., 2002),
although these results are conflicting as either a stronger
effect in girls (Rovet et al., 1987) or only in boys (Schoenle
et al., 2002) was noted. While cross-sectional studies have
not been consistent considering the effect of diabetes
duration (Brands et al., 2005), two longitudinal studies
indicated the presence of a time dependency (Schoenle
et al., 2002; Ryan et al., 2003). An effect on the cogni-
tion-related brain potential (P300) latency was demon-
strated in adult patients which correlated to diabetes
duration and not to a history of hypoglycemic episodes
(Kramer et al., 1998).
The purpose of the present study was to contribute to the
understanding of the origin of the cognitive defects in type 1
diabetes by studying the effect of demographic and medical
parameters on cognitive performance. The participants
were administered a 2-h-long battery of neurocognitive
tests focused on attention, memory and speed. The
rationale for this was earlier evidence of more pronounced
defects on these functions (Ryan et al., 1992, 2003; Northam
T. Brismar et al.
et al., 1999; Hershey et al., 2003) and our findings in
patients with type 1 diabetes of EEG abnormalities in the
frontal and temporal lobes (Brismar et al., 2002; Hyllien-
mark et al., 2005) which are brain regions known to be
important for attention and memory. Measures of body
length and BMI were included in the analysis since previous
studies in healthy populations have shown that these
parameters influence intellectual performance (Teasdale
et al., 1991, 1992; Tuvemo et al., 1999). Multiple regre-
ssion analysis was used to distinguish how cognitive
function was affected by discrete, but not independent,
disease variables such as disease duration, number of
hypoglycemic events, metabolic control and neuromuscular
function.
2. Methods
2.1. Patients
Patients with type 1 diabetes (N ¼ 150) from the Stockholm
City area were recruited between October 2003 and
February 2006. Most patients (N ¼ 142) were recruited
among those who had been screened for a clinical study of
C-peptide effects (Ekberg et al., 2007) and had either not
started or not been included in the drug trial due to lack of
sufficient sensory nerve conduction defects in the sural
nerves or lack of detectable action potentials in both sural
nerves. The study only included patients with diabetes
duration of more than 5 years and age between 22 and 56
years. After giving their informed consent, the patients
underwent physical examination and clinical chemistry
laboratory testing. All patients were C-peptide negative
(o0.2 nM). B-glucose was measured immediately before or
after the cognitive testing. Metabolic control was estimated
from HbA1c data in the medical record and data were
available over 5.072.6 years preceding the cognitive
testing. Data of recent insulin dose were obtained from
the medical record.
Patients were asked about number and time of all
previous hypoglycemic events (which had required help
from another person or hospital admission) and about
number and time of hypoglycemic comas. Patients were
also asked about premature birth and neonatal events, birth
weight, brain trauma and diseases of the brain and the
nervous system, and about other diseases and medication.
Occurrence of other diabetic complications was evaluated
from the medical records. Presence of retinopathy was
detected by ophtalmologists from retinal phothography
and scored as absent (0), non-proliferative (1) and pro-
liferative (2), and nephropathy was scored as absent (0),
with microalbuminuria (1) and with proteinuria (2). Pre-
sence of hypertension included all patients who were
treated for hypertension defined as having a systolic
 ARTICLE IN PRESS
Predictors of cognitive impairment in type 1 diabetes 1043

pressure 4140 mmHg or diastolic pressure 490 mmHg.

Table 1 Test grouping in cognitive domains.
Patients on angiotensin-converting enzyme (ACE) inhibitors
due to microalbuminuria only were not included in this Cognitive domain Cognitive test
category. The presence of peripheral neuropathy was
quantified by the use of a neurological impairment assess- Psychomotor speed Digit Symbol Coding
ment (NIA) score (Ekberg et al., 2007). All patients were Grooved Pegboard Test for
informed about the nature and purpose of the study before dominant hand Grooved
consenting to participate. The protocol was approved by the Pegboard Test for non-
institutional human ethics committee of the Karolinska dominant hand
Institutet, Stockholm. Trail-Making Test B
Psychomotor speed Digit Symbol Coding
excluding pegboard Trail-Making Test B
2.2. Neuropsychological assessment
Memory Claeson–Dahl’s learning and
A licensed neuropsychologist (LM) tested all patients. The retention
tests were performed in a fixed order and took 120 min. Digit Symbol Coding incidental
After completion, the patients were asked about their learning
subjective experience (yes/no) of any decline in cognitive RCFT Immediate and Delayed
function. Patients were also asked about their education recall
(number of years in schools and universities) in order to get RCFT Recognition Trial
a surrogate measure of premorbid intellectual ability. Visual perception and Block Design
The following tests were utilized: organization Digit Symbol Coding
Picture Completion
(1) Tests of Picture Completion, Digit Span, Vocabulary, RCFT Copy Trial
Block Design and Digit Symbol Coding from the Wechsler
Adult Intelligence Scale-Revised (WAIS-R) test battery Visual–spatial ability Block Design
(The Psychological Corporation, 1981) and the Block RCFT Copy Trial
Span test from WAIS-R as a neuropsychological instru- Speed of information Digit Symbol Coding
ment (WAIS-R NI) (The Psychological Corporation, 1991, processing PASAT
2002). Trail-Making Test B
(2) Rey Complex Figure Test (RCFT) (Psychological Assess-
ment Resources, 1995). This test is composed of a copy Attention Digit Span forwards
trial and an immediate and a delayed recall trial, as well Block Span forwards
as a recognition trial. PASAT
(3) Trail Making Test Part B (TMT B) (Lezak, 1995).
Working memory Block Span backwards
(4) Paced Auditory Serial Attention Test (PASAT), measuring
Digit Span backwards
attention, information processing speed and working
PASAT
memory is sensitive to deficits in information processing
(Lezak, 1995; Diehr et al., 1998). Verbal ability FAS
(5) Grooved Pegboard Test (Lezak, 1995; Lafayette Instru- Vocabulary
ment, 1997). The score was obtained for both dominant
General intelligence Block Design
and non-dominant hand.
Vocabulary
(6) Controlled Oral Word Association FAS (Fernaeus and
Almkvist, 1998; The Psychological Corporation, 2001). Executive functions FAS
(7) Zoo Map Test, from the behavioral assessment of the The Zoo Map Test
Dysexecutive Syndrome test battery (BADS) (Chamberlain, Trail-Making Test B
2003).
(8) Claeson—Dahl’s test of learning and memory (Psykolo- Global score All tests
giförlaget, 1998). The retention was tested with a The cognitive domain scores were the mean values of the
delayed recall trial after 30 min. scores in the different tests. Some tests are included in more
than one domain, since they test several cognitive functions.
The scores of the tests were adjusted for the effect of age
2.3. Cognitive domains and education as described in Section 2.6.

The performance in 10 cognitive domains and a global score

was assessed by taking an average of the scores of the
individual tests (Table 1). The psychomotor speed was
assessed in two ways, with and without the Grooved
Pegboard Test. The Grooved Pegboard Test was not available
at the beginning of the study and for this reason
psychomotor speed (which included Pegboard data) was
evaluated only in 97 patients.
2.4. Nerve conduction
Sensory nerve conduction velocity (SCV) and action poten-
tial amplitude (SNAP) were measured in the sural nerves
bilaterally. Motor nerve conduction velocity (MCV) and
compound muscle action potential amplitude (CMAP) were
 ARTICLE IN PRESS
1044 T. Brismar et al.

measured bilaterally in the peroneal nerve. Further details

Table 2 Demographic and clinical data of study
of the methods and their reproducibility have previously
population.
been described (Ekberg et al., 2007).
Factor Mean7S.D. Min–Max
2.5. Neurological examination and symptom
Age (years) 43.377.8 22.6–55.7
assessment
Education (years) 13.072.0 9–19
Height (cm) 173.379.6 153–199
The examination followed a fixed protocol and included Weight (kg) 75.4713.2 52–115
sensory screening for touch, pin prick, vibration and BMI (kg/m2) 25.073.0 20–36.7
temperature, assessed on the big toes, on the dorsum of Mean HbA1c (%) 7.371.2 4.8–13.1
the feet and the tibial regions. The examination also Blood glucose (mmol/l) 8.874.3 2.4–19.8
included reflex testing at two levels and joint propriocep- Insulin dose (units/kg) 0.6970.18 0.30–1.40
tion for the big toes. The different responses were graded as Age at diabetes onset 17.2711.2 1.0–46.0
normal, decreased or absent (0, 1 or 2 points, respectively) (years)
and the sum was the neurological impairment score (NIA). Diabetes duration (years) 26.6711.4 5.8–49.1
Hypoglycemic events 3.0712.1 0–100
2.6. Statistical analysis during the last 3 years
(total number)
Values are given as mean7S.D. with ranges in brackets. All hypoglycemic events 6.3714.7 0–100
Normal age-dependent effects on performance in the more than 3 years ago
neuropsychological test were compensated for by calculat- (total number)
ing z-scores defined as the observed value minus the mean Years since latest 6.578.2 0–40.4
of the reference value divided by its S.D. The obtained hypoglycemic event
scores were further adjusted (z-scored) with regard to the All comas (total number) 3.077.8 0–50
education of the patients (number of years in school and Years since latest coma 10.9710.1 0–40
higher education) in order to control for premorbid Weight at birth 34647586 1200–4580
intellectual ability. Significance was tested using t-tests as
indicated. Correlations were studied with Spearman’s rank Female gender 56%
correlation. A value of po0.05 was considered significant. Potential confounding factors 20%
Calculations were performed with Matlab 7.0 and SPSS 14.0. Psychotropic medication 10%
Retinopathy, non-proliferative 39%
Retinopathy, proliferative 32%
3. Results Nephropathy with microalbuminuria 8%
Nephropathy with proteinuria 0.8%
3.1. Demographic and clinical properties of study Hypertension 26%
population NIA score more than 7 70%
Subjective feeling of cognitive decline 39%
The demographic and clinical characteristics of the diabetic
patients are described in Table 2. The study population was
not randomized and it differed from a cross section of
patients since the sex ratio of women/men was 1.27–1.0. measured before or after the cognitive test was o3.0 mmol/
The mean age and diabetes duration were 3–4 years higher l in five patients which may have affected their cognitive
and longer than the average for a subject with type 1 tests results and these patients were excluded from the
diabetes in the Swedish population, whereas BMI and HbA1c analysis.
were the same as the average (Gudbjornsdottir et al.,
2003). Twenty-six patients had a diabetes onset before the
age of 6 years. Potential confounding factors were pre- 3.2. Correlation between cognitive function and
maturity (N ¼ 19), concussion of the brain (N ¼ 5), (possi- demographic and medical properties
ble) meningitis (N ¼ 3), and headache (N ¼ 2) as reported
by the patients. The medical records did not indicate that A primary analysis was performed on the scores of the
any of these patients had signs of previous brain damage and separate cognitive tests. The outcome of the pegboard test
none of these were excluded. Psychotropic medications such differed from the other tests, having higher correlation with
as serotonin reuptake inhibitors were used by 12, other anti- neuromuscular function and neuropathy. This suggested that
depressive agents (presynaptic alpha-2 antagonists) by two, poor sensory–motor function in the hands contributed to low
benzodiazepine derivatives by one, and tioxantene deriva- scores in the pegboard tests, and for this reason the
tives by one patient. This latter patient was excluded from cognitive domain ‘psychomotor function’ was evaluated
the analysis since tioxantene medication may compromise both with and without the pegboard test scores.
cognitive performance. Many patients had never experi- In the present analysis, the separate cognitive scores were
enced a hypoglycemic event (35 patients) or a hypoglycemic normalized to years of education. Before normalization
coma (66 patients), and in three patients these data were this parameter showed strongest correlation to vocabulary
uncertain and not included in the analysis. Blood glucose (0.48, po108) followed by PASAT B (0.36, po104).
 ARTICLE IN PRESS
Predictors of cognitive impairment in type 1 diabetes
The cognitive domain scores correlated significantly
with several of the demographic and medical parameters
(Figure 1). Except for gender, these correlations were in the
same direction (positive or negative) for each effector.
Female gender was positively correlated with psychomotor
speed (0.23), but negatively correlated with working
memory (0.16) and general intelligence (0.22). Old
age, short body height, high BMI, young age at diabetes
onset, and long diabetes duration all had negative influence
on cognitive performance. For these variables, the strongest
correlations were found between age and psychomotor
speed (0.41), between height and general intelligence
(0.29), between BMI and visual perception–organization
(0.19), between age at diabetes onset and processing
speed (0.30), and between long diabetes duration and
psychomotor speed w/wo pegboard (0.42 and 0.32,
respectively). High number of all previous hypoglycemic
comas correlated with low scores in visual perception–orga-
nization (0.21), visual–spatial ability (0.17) and general
intelligence (0.19). Poor metabolic control (defined by
mean HbA1c) and presence of nephropathy, respectively,
had only weak correlation to a single domain score.
Presence of retinopathy or hypertension correlated with
low scores of psychomotor speed (0.28 and 0.29), visual
perception–organization (0.23 and 0.25) and visual–spa-
tial ability (0.22 and 0.23, respectively). Similarly, poor
neuromuscular function correlated with low scores of
psychomotor speed, visual perception–organization, visual–
spatial ability, and a low global score, and of these
parameters CMAP and SNAP had the strongest correlations,
followed by SCV and MCV (not illustrated). In particular,
CMAP and SNAP correlated with psychomotor speed
Figure 1 Correlation between demographic, medical and
neuromuscular function parameters and cognitive domain
scores. Cognitive domains on the x-axis: (1) psychomotor speed,
(2) psychomotor speed excl. pegboard, (3) memory, (4) visual
perception–organization, (5) visual–spatial ability, (6) proces-
sing speed, (7) attention, (8) working memory, (9) verbal ability,
(10) general intelligence, (11) executive functions, (12) global
score. Coma is total number of hypoglycemic coma during
disease duration. Hypoglycemia is number of hypoglycemic
events last 3 years. Symbols in order of size from smallest to
largest depict p-values of Spearman’s rho: po0.05, po0.01,
po0.001, and po0.0001. Open symbols for positive correlation
and filled symbols for negative correlation. Effect of gender
tested with t-test.
1045
(0.48 and 0.44), psychomotor speed without pegboard
(0.28 and 0.26) and visual perception–organization (0.28
and 0.23, respectively). A high NIA score correlated with low
scores in psychomotor speed, visual perception–organiza-
tion and visual–spatial ability (0.28 for all). Blood glucose,
insulin dose, potential confounding factors, psychotropic
medication or subjective feeling of cognitive decline did not
correlate with any measure of cognitive function (data not
included).
3.3. Bivariate analysis of predictors of cognitive
impairment
Many of the variables that correlated to the cognitive
performance were inter-correlated (Figure 2). SNAP and
CMAP had strong negative correlation with age, diabetes
duration, mean HbA1c, retinopathy and NIA score. Presence
and severity of retinopathy was highly correlated to young
age at diabetes onset, long diabetes duration, high HbA1c,
high NIA score, low SNAP, low CMAP and more weakly to
presence of nephropathy and low number of hypoglycemic
events. Similarly, nephropathy, although more weakly,
correlated to young age at diabetes onset and long diabetes
duration. Hypertension correlated to older age and more
weakly to high BMI, long diabetes duration, high NIA score
and low SNAP. Total number of comas and number of
hypoglycemic events during the last 3 years were strongly
correlated. In addition, there were inherent correlations
between gender and height, and between the time
dependent parameters age, age at diabetes onset, diabetes
duration and total number of comas. Education was
positively correlated with height and negatively correlated
with mean HbA1c. The cognitive scores had been normalized
for years of education and this parameter was therefore not
included in Figure 1 and the following analysis.
Figure 2 Inter-correlation between demographic, medical and
neuromuscular function parameters. Numbers on x-axis corre-
spond to the parameters listed on the y-axis. Coma is total
number of hypoglycemic coma during disease duration. Hypo-
glycemia is number of hypoglycemic events last 3 years.
Symbols in order of size from smallest to largest depict
p-values of Spearman’s rho: po0.05, po0.01, po0.001, and
po0.0001. Open symbols for positive correlation and filled
symbols for negative correlation.
 ARTICLE IN PRESS
1046 T. Brismar et al.

3.4. Step-wise forward multiple regression analysis
of cognitive impairment
A multiple forward linear regression analysis was performed
in order to separate the effects of different interrelated
disease variables (Table 3). Long diabetes duration and
early age of DM onset were the strongest predictors for
low performance in seven cognitive domains (psychomotor
speed w/wo pegboard, memory, processing speed, attention,
working memory, verbal ability and executive functions) and
for a low global score. Early age of DM onset was also a
predictor of low general intelligence. Old age was the only
predictor of low visual–spatial ability. Low CMAP was the
strongest predictor for a low score in visual perception–
organization and predicted low psychomotor speed. Short
body height was the strongest predictor for a low score in
general intelligence and a predictor for low performance in
working memory. If height was excluded from the model,

Table 3 Effect of demographic, medical and neuromuscular function parameters on cognitive performance.

Dependent F Predictor B S.E. Beta po R2

Psychomotor speed 13.0 (Constant) 0.185 0.175 0.289

DM duration 0.017 0.005 0.291 0.000
Female gender 0.342 0.099 0.262 0.001
CMAP 0.065 0.024 0.215 0.007 0.218

Psychomotor speed excluding pegboard 8.63 (Constant) 0.387 0.165 0.020

DM duration 0.020 0.006 0.280 0.001
Female gender 0.374 0.125 0.235 0.003
Nephropathy 0.460 0.216 0.167 0.035 0.156

Memory 5.74 (Constant) 0.753 0.205 0.000

DM duration 0.017 0.005 0.278 0.002
SNAP 0.032 0.011 0.247 0.006
Hypertension 0.277 0.140 0.161 0.050 0.110
Visual perception–organization 10.7 (Constant) 0.214 0.119 0.076
CMAP 0.090 0.026 0.273 0.001
Hypertension 0.344 0.140 0.196 0.015 0.131

Visual–spatial ability 11.5 (Constant) 1.473 0.443 0.001

Age 0.034 0.010 0.273 0.001 0.075

Processing speed 7.97 (Constant) 0.860 0.613 0.163

Age DM onset 0.023 0.006 0.291 0.000
BMI 0.050 0.024 0.165 0.042 0.102

Attention 8.07 (Constant) 0.241 0.111 0.032

Age DM onset 0.015 0.005 0.232 0.005 0.054

Working memory 5.50 (Constant) 2.458 1.099 0.027

Age DM onset 0.013 0.006 0.199 0.016
Height 0.013 0.006 0.166 0.043 0.072

Verbal ability 5.53 (Constant) 0.190 0.106 0.074

Age DM onset 0.012 0.005 0.194 0.020 0.037

General intelligence 7.90 (Constant) 2.554 1.090 0.021

Height 0.020 0.006 0.273 0.001 0.145
Age DM onset 0.013 0.005 0.197 0.014
BMI 0.046 0.019 0.191 0.017 0.145
Executive function 5.65 (Constant) 0.314 0.133 0.019
DM duration 0.017 0.005 0.305 0.001
Retinopathy 0.158 0.077 0.189 0.042 0.074

Global score 14.3 (Constant) 0.347 0.098 0.001

DM duration 0.013 0.003 0.302 0.000 0.091

Multiple linear forward regression analysis. The predictor variables in the model were female gender, age, height, BMI, age at diabetes
onset, diabetes duration, number of hypoglycemic events during the last 3 years, number of comas, mean HbA1c, retinopathy,
nephropathy, hypertension, NIA score, SNAP and CMAP. F is variance ratio, B is regression coefficient, Beta is standardized regression
coefficient, R2 is the proportion of variance accounted for by the independent variables. Missing variables were replaced with mean.
 ARTICLE IN PRESS
Predictors of cognitive impairment in type 1 diabetes
female gender became the best predictor of low score in
general intelligence (B ¼ 0.282, po0.016). Conversely,
female gender predicted higher psychomotor speed
(Table 3). High BMI was a predictor of low scores in two
cognitive domains (processing speed and general intelli-
gence). Other more weak predictors were hypertension (of
low scores in memory and visual perception–organization)
and nephropathy (of low psychomotor speed excl. pegboard)
and retinopathy (of high score in executive function).
If the demographic variables age, height and gender were
forced into the regression model, ‘age at diabetes onset’
replaced ‘diabetes duration’ as the strongest predictor of
poor performance in all cognitive domains for which
diabetes duration was a predictor. Other effects were that
CMAP entered as a weak predictor of visual–spatial ability
(B ¼ 0.084, po0.035), BMI entered as a weak predictor of
low global score (B ¼ 0.026, po0.044) and retinopathy
as a weak predictor of high verbal ability (B ¼ 0.165,
po0.049).
In contrast to the positive correlation between CMAP and
cognitive function, high SNAP was a weak predictor of a low
score in tests of memory (Table 3). When HbA1c was forced
into the model some weak predictors disappeared (nephro-
pathy for psychomotor speed, hypertension for memory, and
BMI for processing speed), whereas female gender entered
as weak predictor of high score in executive function. If
CMAP was removed from the model, old age entered as the
major predictor of low score in visual perception–organiza-
tion (B ¼ 0.090, po0.020) followed by retinopathy
(B ¼ 0.328, po0.027), whereas SNAP entered as a weak
predictor of psychomotor speed (B ¼ 0.021, po0.043). If all
neuromuscular parameters (NIA, SNAP and CMAP) were
removed from the model, age entered as a predictor of
psychomotor speed (B ¼ 0.015, po0.029) and visual–
spatial perception (B ¼ 0.018, po0.020) and hypertension
disappeared as a predictor of low-memory score.
The regression analysis was repeated after exclusion of
patients on any psychotropic medication (N ¼ 13) and
patients with blood glucose o4.0 mM (N ¼ 11), which
resulted in a study population of 120 subjects. The
variability in the scores of verbal ability and executive
function had no predictors in this smaller study population
and some other predictions disappeared (diabetes duration
for memory, young age of onset for general intelligence and
CMAP for psychomotor speed). Overall, the analysis showed
that the cognitive scores were predicted by the same main
factors and that all predictors had the same sign of the
regression coefficients as the corresponding coefficients
shown in Table 3.
4. Discussion
Long diabetes duration, young age at diabetes onset and
presence of peripheral nerve conduction defects were the
disease variables that had the strongest correlation to
cognitive impairment. Psychomotor speed and visual per-
ception–organization were those functions, which showed
the strongest dependency on any of the disease variables,
and the least affected was verbal ability. The present study
was designed to study effects within a large group of adult
patients and not to identify differences in cognitive function
1047
relative to non-diabetic subjects. However, the affected
domains were the same as those most commonly found to
show a decline in diabetes patients relative to healthy
controls (see Brands et al., 2005).
In order to better resolve the effect of disease related
variables on the cognitive function, the test scores were not
only normalized to age but also to years of education.
Regression models based on demographic data or on current
scores from individual tests (such as vocabulary, best
performance or reading tests) have been used to evaluate
premorbid functioning (for review and critique see Franzen
et al., 1997; Reynolds, 1997; Griffin et al., 2002). Assessing
premorbid intelligence is particularly difficult in patients
with a disease onset at early age, since it cannot be assumed
that a certain subset of functions are unaffected. Since the
median age of diabetes onset (14 years) was during the
school years, the educational level does not represent a
reliable measure of premorbid functioning in this popula-
tion, but this parameter did explain, e.g., 23% of the
variability in the scores of vocabulary (from WAIS-R). The
educational level was not influenced by an early age of
diabetes onset, but a higher educational level correlated
with better glycemic control. Although earlier studies have
shown that health literacy correlates to the educational
level, an association between literacy and glycemic control
has not been a consistent finding (Schillinger et al., 2002;
Morris et al., 2006).
Depression has previously been reported to be more
prevalent in patients with type 1 diabetes than in the
general population (Anderson et al., 2001). The psychologi-
cal well-being was not evaluated in the present study, but a
relatively high percentage (10%) of the patients were on
anti-depressive medication as compared to the overall use
(about 5% with adjustment for age and gender differences)
in the Swedish population (Apoteket, 2005). Depressive
symptoms (expressed as BDI-II scores) and cognitive pro-
blems were not found to be significantly related to cognitive
performance in older patients (Brands et al., 2006). Neither
did the present study show any correlation between self-
reported cognitive decline and the cognitive scores. The use
of serotonin reuptake inhibitors did not affect the cognitive
scores in the present study, and has been shown to have no
detrimental effect on psychomotor speed or attention, but
may cause memory impairment (Wadsworth et al., 2005).
Another possible confounding factor was low blood glucose
during the cognitive testing. Exclusion of patients on
psychotropic medication and blood glucose o4.0 mM did
not change the main outcome of the regression analysis.
Previous studies have been contradictory with regard to
the effect of diabetes duration. Performance in a delayed
recall task (Prescott et al., 1990), psychomotor speed (Ryan
et al., 2003) and visual–motor integration (Skenazy and
Bigler, 1984) were found to be inversely related to diabetes
duration, whereas other studies showed no effects on
delayed recall, vocabulary, digit symbol block design, and
performance (Lawson et al., 1984) or psycho-motor speed
(Skenazy and Bigler, 1984; Ryan et al., 1992). A recent study
of a large population of patients with type 1 diabetes gave
no evidence of a substantial cognitive decline over 18 years
(Jacobson et al., 2007). However, this is a 9 years shorter
period of time than the average disease duration in the
present study and the effect of lifetime duration was not
 ARTICLE IN PRESS
1048
analyzed. Kramer et al. (1998) found a correlation between
diabetes duration and cognition-related brain signals (P300
latency). They discerned no effect on the cognitive scores as
measured with mini-mental examination or trail making
test, which, however, may be explained by the insufficient
sensitivity of these tests.
Age was the strongest predictor of visuo-spatial ability, and
hypothetically normal aging might have contributed to a
decline in several cognitive domains, causing an apparent
dependence on diabetes duration. Although normal aging
effects were compensated for by scoring against age-
matched reference data, some unrecognized effect of aging
cannot be excluded. Forcing age into the regression model is
likely to have overcompensated for the normal aging effects,
but it is interesting to observe that this made early age at
diabetes onset the strongest predictor for a low score in 8 out
of 10 cognitive domains and for a low global score. Two
longitudinal studies in children have shown a specific decline
in verbal intelligence quotient (Rovet and Ehrlich, 1999;
Schoenle et al., 2002), and our results support this finding,
revealing young age at diabetes onset as a the only predictor
of poor verbal performance in the regression model.
Psychomotor speed in particular, but also visual percep-
tion–organization and visuo-spatial ability correlated to
CMAP and SNAP (or NIA score), which agrees with the
findings by Ryan et al. (1992, 1993) of an association
between psychomotor slowing and polyneuropathy in adults
with type 1 diabetes. In these studies and the present one,
the score of psychomotor speed included results of the
Grooved Pegboard Test where the subject was asked to
place the pegs in a board with holes as rapidly as possible.
The pegs are grooved and must be rotated in order to fit into
the hole (Lafayette Instrument, 1997), and both tactile and
visual information are important to perform the task rapidly.
The presence of peripheral diabetic neuropathy is therefore
likely to have contributed to a low pegboard score, and
accordingly, when psychomotor speed was scored without
the pegboard data the correlation diminished. There was a
strong correlation between neuropathy and diabetes dura-
tion, and the muscle electrical response CMAP was the only
neuromuscular parameter that predicted cognitive impair-
ment in the regression model.
A history of frequent or recent hypoglycemic events did
not predict cognitive dysfunction. Our finding is in agree-
ment with some previous studies on adult patients (Reichard
et al., 1991; Ryan et al., 1992; DCCT, 1996; Kramer et al.,
1998) and a recent large longitudinal study of adult patients
(Jacobson et al., 2007). The number of hypoglycemic events
as estimated retrospectively by the patient may be
uncertain and result in a negative finding, and the effect
of hypoglycemic events is more accurately evaluated in a
longitudinal study. In particular, it should be pointed out
that the present study, like several of the above cited, has
not ruled out the possibility that hypoglycemic events at a
young age may affect the cognitive function later in life. Our
finding that early age of diabetes onset was a major
predictor suggests that factors in early life contribute to
the cognitive decline, and we cannot exclude that early
hypoglycemic events is such a factor.
Performance in several cognitive domains had a positive
correlation with body length and a negative correlation with
BMI. In addition, it was noted that years of education as a
T. Brismar et al.
surrogate marker for intellectual ability was correlated to
height. A similar correlation between body length and
intellectual performance also exists in the healthy popula-
tion (r ¼ 0.24, Teasdale et al. (1991); r ¼ 0.14, Tuvemo
et al. (1999)). Due to the strong correlation between gender
and height, the gender differences would have been
exaggerated if the effect of height had been omitted in
the regression model. Similarly, BMI in a healthy population
has a negative correlation with intellectual performance
(r ¼ 0.073, Teasdale et al. (1992); r ¼ 0.06, Tuvemo
et al. (1999)), but this can only partially explain the stronger
correlations (about 0.19) presently found with visual
perception–organization and visual–spatial ability. Our find-
ings suggest that high BMI might have a more detrimental
effect on cognitive function in patients with type 1 diabetes
than in a healthy population.
A measure of the long-term metabolic control was
obtained from HbA1c levels for a period of about 5 years
prior to the cognitive testing. It is a limitation to the study
that we have not been able to measure the lifetime
metabolic control, or the degree of glucose instability.
However, our results on the effect of metabolic control
agree with previous studies showing that nerve conduction
defects as well as peripheral neuropathy are strongly
correlated to poor metabolic control and long disease
duration (DCCT, 1988, 1993; Hyllienmark et al., 2001). In
contrast, mean HbA1c was only weakly correlated to
cognitive impairment, and mean HbA1c had no effect in
the regression models. The possibility that the influence of
mean HbA1c was masked by the explanatory effect of CMAP
was tested by exclusion of the neuromuscular function
variables from the model, but this did not uncover any
effect of mean HbA1c. The difference between cognitive
impairment and neuropathy with regard to the effect of
mean HbA1c strengthens the argument that the pathogen-
esis of the cognitive impairment is not the same as for
peripheral diabetic neuropathy, although both are strongly
correlated to diabetes duration.
Although retinopathy correlated with some scores of
cognitive function, it had an even stronger correlation with
diabetes duration and young age of diabetes onset, and in
the regression model retinopathy did not predict cognitive
impairment. For this reason, the present analysis does not
support the conclusion that the pathogenesis of the
cognitive defects is related to microvascular complications
in the brain as revealed by the presence of retinopathy
(Ferguson et al., 2003; Ryan et al., 2003). Nephropathy and,
more strongly, hypertension correlated to low cognitive
function and they were both significant predictors in the
regression model. Cognitive decline and a higher risk for
dementia have been related to hypertension in middle aged
and elderly populations with or without diabetes (van
Swieten et al., 1991; Kilander et al., 1998; Tzourio et al.,
1999; Sierra et al., 2004; Verdelho et al., 2007) and it has
earlier been pointed out that hypertension for several
reasons may play a role in the pathophysiology of cerebral
complications in diabetes (Biessels, 1999). In the present
analysis, hypertension was only classified as present or
absent and our findings motivate a more detailed staging
including duration of hypertension.
Three of the predictors, BMI, body height and CMAP, may
be correlated to circulating levels of insulin-like growth
 ARTICLE IN PRESS
Predictors of cognitive impairment in type 1 diabetes
factors (IGFs) and IGF binding proteins (BPs). For men, high
BMI is a better predictor of low IGF-I levels than is low
stature, whereas for women low stature is more important,
and for both genders lower IGF-II levels are associated with
higher BMI (Chang et al., 2002; Gram et al., 2006). In poorly
controlled type 1 diabetes IGF-I, IGF-II and IGFBP-3
concentrations are decreased, whereas IGFBP-1 is increased
(Brismar and Lewitt, 2005). CMAP is an electrical measure of
the muscle response to suprathreshold nerve stimulation.
A significant part of the muscle anabolic and growth-
promoting effects of testosterone and muscle contraction
occurs via the induction of in situ IGF-I and the inhibition of
IGFBP-4 gene expression (Veldhuis et al., 2005). Several
studies have implicated a protective role of IGF-I against
age-dependent degenerative changes in the brain and
cognitive impairment (de la Monte and Wands, 2005; Carro
and Torres-Aleman, 2006; Okereke et al., 2006). In the
spontaneously type 1 diabetic BB/W rat correlations were
determined between neuronal loss, reduced mRNA expres-
sion of IGF-I, IGF-II IGF-IR and IR in hippocampus and
cognitive defects (Li et al., 2002). It is intriguing that the
influence of diabetes duration, BMI, height, old age and
CMAP might fit into a pathogenesis model in which loss of the
normal neuro-protective effects of insulin or IGF plays a
role. This hypothesis is readily tested by including measures
of IGFs and BPs of the patients in future studies.
In conclusion, long diabetes duration and young age of
diabetes onset were the strongest predictors of cognitive
impairment, with effects on psychomotor speed, memory,
processing speed, attention, working memory, verbal ability,
general intelligence, executive functions and the global
score. Other significant predictors were low CMAP (for visual
perception–organization), old age (for visual–spatial ability),
short stature, high BMI and hypertension. Although a history
of hypoglycemic events was not a predictor of cognitive
impairment, we cannot exclude that the influence of young
age of diabetes onset depends on the effect of hypoglycemic
events early in life. In contrast to the nerve conduction
defects, the cognitive impairment did not correlate to the
presence of retinopathy and the long-term metabolic
control, indicating that the pathogenesis of cognitive
impairment and peripheral neuropathy is different.
Role of the funding source
Funding for this study was provided by Juvenile Diabetes
Research Foundation International (JDRF) Grant 1-2004-685;
the JDRF had had no further role in study design; in the
collection, analysis and interpretation of data; in the writing
of the report; and in the decision to submit the paper for
publication.
Conflict of interest
All authors declare that they have no conflict of interest.
Acknowledgments
We thank Dr. Björn Jonsson Ph.D. for biostatistical advice
and Prof. Kerstin Brismar MD, Ph.D. for valuable comments
1049
on the manuscript. This research was supported by Juvenile
Diabetes Research Foundation International (JDRF) Grant
1-2004-685 to T.B. The JDRF had no further role in study
design; in the collection, analyses and interpretation of the
data; in the writing of the report; and in the decision to
submit the paper for publication.
References
Anderson, R.J., Freedland, K.E., Clouse, R.E., Lustman, P.J., 2001.
The prevalence of comorbid depression in adults with diabetes:
a meta-analysis. Diabetes Care 24, 1069–1078.
Apoteket, A.B., 2005. Försäljningsutveckling för läkemedel inom
ATC N—Nervsystemet. /http://www2.apoteket.se/NR/rdonlyres/
EE029B50-C038-4760-B7B0-0D911FF64FD3/0/2006atcNny.pdfS.
Biessels, G.J., 1999. Cerebral complications of diabetes: clinical
findings and pathogenetic mechanisms. Neth. J. Med. 54,
35–45.
Brands, A.M., Biessels, G.J., de Haan, E.H., Kappelle, L.J., Kessels,
R.P., 2005. The effects of type 1 diabetes on cognitive
performance: a meta-analysis. Diabetes Care 28, 726–735.
Brands, A.M., Kessels, R.P., Hoogma, R.P., Henselmans, J.M., van
der Beek Boter, J.W., Kappelle, L.J., de Haan, E.H., Biessels,
G.J., 2006. Cognitive performance, psychological well-being,
and brain magnetic resonance imaging in older patients with
type 1 diabetes. Diabetes 55, 1800–1806.
Brismar, K., Lewitt, M.S., 2005. In: Houston, M.S., Holly, J.M.,
Feldman, E.L. (Eds.), The IGF and IGFBP System in Insulin
Resistance and Diabetes Mellitus. Humana Press, Totowa, NJ,
pp. 251–270.
Brismar, T., Hyllienmark, L., Ekberg, K., Johansson, B.L., 2002. Loss
of temporal lobe beta power in young adults with type 1 diabetes
mellitus. Neuroreport 13, 2469–2473.
Carro, E., Torres-Aleman, I., 2006. Serum insulin-like growth factor I
in brain function. Keio J. Med. 55, 59–63.
Chamberlain, E., 2003. Test review of the behavioural assessment of
the dysexecutive syndrome. J. Occup. Psychol. Empl. Disabil. 5,
33–37.
Chang, S., Wu, X., Yu, H., Spitz, M.R., 2002. Plasma concentrations
of insulin-like growth factors among healthy adult men and
postmenopausal women: associations with body composition,
lifestyle, and reproductive factors. Cancer Epidemiol. Biomar-
kers Prev. 11, 758–766.
DCCT (The DCCT Research Group), 1988. Factors in development of
diabetic neuropathy. Baseline analysis of neuropathy in feasi-
bility phase of Diabetes Control and Complications Trial (DCCT).
Diabetes 37, 476–481.
DCCT (The Diabetes Control and Complications Trial Research
Group), 1993. The effect of intensive treatment of diabetes on
the development and progression of long-term complications in
insulin-dependent diabetes mellitus. N. Engl. J. Med. 329,
977–986.
DCCT, 1996. Effects of intensive diabetes therapy on neuropsycho-
logical function in adults in the Diabetes, Control and Complica-
tions Trial. Ann. Intern. Med. 124, 379–388.
Deary, I.J., Crawford, J.R., Hepburn, D.A., Langan, S.J., Blackmore,
L.M., Frier, B.M., 1993. Severe hypoglycemia and intelligence in
adult patients with insulin-treated diabetes. Diabetes 42,
341–344.
de la Monte, S.M., Wands, J.R., 2005. Review of insulin and insulin-
like growth factor expression, signaling, and malfunction in the
central nervous system: relevance to Alzheimer’s disease.
J. Alzheimers Dis. 7, 45–61.
Diehr, M.C., Heaton, R.K., Miller, W., Grant, I., 1998. The Paced
Auditory Serial Addition Task (PASAT): norms for age, education,
and ethnicity. Assessment 5, 375–387.
 ARTICLE IN PRESS
1050
Ekberg, K., Brismar, T., Johansson, B.-L., Lindström, P., Juntti-
Berggren, L., Norrby, A., Berne, C., Arnqvist, H.J., Bolinder, J.,
Wahren, J., 2007. C peptide replacement therapy improves
sensory nerve function in type 1 diabetes neuropathy. Diabetes
Care 30, 71–76.
Ferguson, S.C., Blane, A., Perros, P., McCrimmon, R.J., Best, J.J.,
Wardlaw, J., Deary, I.J., Frier, B.M., 2003. Cognitive ability and
brain structure in type 1 diabetes: relation to microangiopathy
and preceding severe hypoglycemia. Diabetes 52, 149–156.
Fernaeus, S.E., Almkvist, O., 1998. Word production: dissociation of
two retrieval modes of semantic memory across time. J. Clin.
Exp. Neuropsychol. 20, 137–143.
Franzen, M.D., Burgess, E.J., Smith-Seemiller, L., 1997. Methods of
estimating premorbid functioning. Arch. Clin. Neuropsychol. 12,
711–738.
Gram, I.T., Norat, T., Rinaldi, S., Dossus, L., Lukanova, A., Tehard,
B., Clavel-Chapelon, F., van Gils, C.H., van Noord, P.A., Peeters,
P.H., Bueno-de-Mesquita, H.B., Nagel, G., Linseisen, J., Lah-
mann, P.H., Boeing, H., Palli, D., Sacerdote, C., Panico, S.,
Tumino, R., Sieri, S., Dorronsoro, M., Quiros, J.R., Navarro, C.A.,
Barricarte, A., Tormo, M.J., Gonzalez, C.A., Overvad, K., Paaske
Johnsen, S., Olsen, A., Tjonneland, A., Travis, R., Allen, N.,
Bingham, S., Khaw, K.T., Stattin, P., Trichopoulou, A., Kala-
pothaki, V., Psaltopoulou, T., Casagrande, C., Riboli, E., Kaaks,
R., 2006. Body mass index, waist circumference and waist–hip
ratio and serum levels of IGF-I and IGFBP-3 in European women.
Int. J. Obes. (London) 30, 1623–1631.
Griffin, S.L., Mindt, M.R., Rankin, E.J., Ritchie, A.J., Scott, J.G.,
2002. Estimating premorbid intelligence: comparison of tradi-
tional and contemporary methods across the intelligence
continuum. Arch. Clin. Neuropsychol. 17, 497–507.
Gudbjornsdottir, S., Cederholm, J., Nilsson, P.M., Eliasson, B., 2003.
The National Diabetes Register in Sweden: an implementation of
the St. Vincent Declaration for Quality Improvement in Diabetes
Care. Diabetes Care 26, 1270–1276.
Hershey, T., Lillie, R., Sadler, M., White, N.H., 2003. Severe
hypoglycemia and long-term spatial memory in children with
type 1 diabetes mellitus: a retrospective study. J. Int.
Neuropsychol. Soc. 9, 740–750.
Hyllienmark, L., Golster, H., Samuelsson, U., Ludvigsson, J., 2001.
Nerve conduction defects are retarded by tight metabolic
control in type I diabetes. Muscle Nerve 24, 240–246.
Hyllienmark, L., Maltez, J., Dandenell, A., Ludvigsson, J., Brismar,
T., 2005. EEG abnormalities with and without relation to severe
hypoglycaemia in adolescents with type 1 diabetes. Diabetologia
48, 412–419.
Jacobson, A.M., Musen, G., Ryan, C.M., Silvers, N., Cleary, P.,
Waberski, B., Burwood, A., Weinger, K., Bayless, M., Dahms, W.,
Harth, J., 2007. Long-term effect of diabetes and its treatment
on cognitive function. N. Engl. J. Med. 356, 1842–1852.
Kilander, L., Nyman, H., Boberg, M., Hansson, L., Lithell, H., 1998.
Hypertension is related to cognitive impairment: a 20-year
follow-up of 999 men. Hypertension 31, 780–786.
Kramer, L., Fasching, P., Madl, C., Schneider, B., Damjancic, P.,
Waldhausl, W., Irsigler, K., Grimm, G., 1998. Previous episodes of
hypoglycemic coma are not associated with permanent cognitive
brain dysfunction in IDDM patients on intensive insulin treat-
ment. Diabetes 47, 1909–1914.
Lafayette Instrument, 1997. Instruction/Owner0 s Manual for the
32025 Grooved Pegboard Test. Lafayette Instrument, Lafayette.
Langan, S.J., Deary, I.J., Hepburn, D.A., Frier, B.M., 1991.
Cumulative cognitive impairment following recurrent severe
hypoglycaemia in adult patients with insulin-treated diabetes
mellitus. Diabetologia 34, 337–344.
Lawson, J.S., Erdahl, D.L., Monga, T.N., Bird, C.E., Donald, M.W.,
Surridge, D.H., Letemendia, F.J., 1984. Neuropsychological
function in diabetic patients with neuropathy. Br. J. Psychiatry
145, 263–268.
T. Brismar et al.
Lezak, M., 1995. Neuropsychological Assessment, third ed. Oxford
University Press, New York.
Li, Z.G., Zhang, W., Grunberger, G., Sima, A.A., 2002. Hippocampal
neuronal apoptosis in type 1 diabetes. Brain Res. 946, 221–231.
Lincoln, N.B., Faleiro, R.M., Kelly, C., Kirk, B.A., Jeffcoate, W.J.,
1996. Effect of long-term glycemic control on cognitive function.
Diabetes Care 19, 656–658.
Morris, N.S., MacLean, C.D., Littenberg, B., 2006. Literacy and
health outcomes: a cross-sectional study in 1002 adults with
diabetes. BMC Fam. Pract. 7, 49.
Northam, E.A., Anderson, P.J., Werther, G.A., Warne, G.L.,
Andrewes, D., 1999. Predictors of change in the neuropsycho-
logical profiles of children with type 1 diabetes 2 years after
disease onset. Diabetes Care 22, 1438–1444.
Northam, E.A., Anderson, P.J., Jacobs, R., Hughes, M., Warne, G.L.,
Werther, G.A., 2001. Neuropsychological profiles of children with
type 1 diabetes 6 years after disease onset. Diabetes Care 24,
1541–1546.
Okereke, O.I., Kang, J.H., Ma, J., Gaziano, J.M., Grodstein, F.,
2006. Midlife plasma insulin-like growth factor I and cognitive
function in older men. J. Clin. Endocrinol. Metab. 91,
4306–4312.
Prescott, J.H., Richardson, J.T., Gillespie, C.R., 1990. Cognitive
function in diabetes mellitus: the effects of duration of illness
and glycaemic control. Br. J. Clin. Psychol. 29 (Pt. 2), 167–175.
Psychological Assessment Resources, 1995. Rey Complex Figure Test
and Recognition Trial. Psychological Assessment Resources, Lutz.
Psykologiförlaget, A.B., 1998. Claeson–Dahls test för inlärning och
minne. Psykologiförlaget AB, Halmstad.
Reichard, P., Berglund, A., Britz, A., Levander, S., Rosenqvist, U.,
1991. Hypoglycaemic episodes during intensified insulin treat-
ment: increased frequency but no effect on cognitive function.
J. Intern. Med. 229, 9–16.
Reynolds, C.R., 1997. Postscripts on premorbid ability estimation:
conceptual addenda and a few words on alternative and
conditional approaches. Arch. Clin. Neuropsychol. 12, 769–778.
Rovet, J.F., Ehrlich, R.M., 1999. The effect of hypoglycemic seizures
on cognitive function in children with diabetes: a 7-year
prospective study. J. Pediatr. 134, 503–506.
Rovet, J.F., Ehrlich, R.M., Hoppe, M., 1987. Intellectual deficits
associated with early onset of insulin-dependent diabetes
mellitus in children. Diabetes Care 10, 510–515.
Ryan, C.M., Williams, T.M., Orchard, T.J., Finegold, D.N., 1992.
Psychomotor slowing is associated with distal symmetrical
polyneuropathy in adults with diabetes mellitus. Diabetes 41,
107–113.
Ryan, C.M., Williams, T.M., Finegold, D.N., Orchard, T.J., 1993.
Cognitive dysfunction in adults with type 1 (insulin-dependent)
diabetes mellitus of long duration: effects of recurrent
hypoglycaemia and other chronic complications. Diabetologia
36, 329–334.
Ryan, C.M., Geckle, M.O., Orchard, T.J., 2003. Cognitive efficiency
declines over time in adults with Type 1 diabetes: effects of
micro- and macrovascular complications. Diabetologia 46,
940–948.
Sachon, C., Grimaldi, A., Digy, J.P., Pillon, B., Dubois, B., Thervet,
F., 1992. Cognitive function, insulin-dependent diabetes and
hypoglycaemia. J. Intern. Med. 231, 471–475.
Schillinger, D., Grumbach, K., Piette, J., Wang, F., Osmond, D.,
Daher, C., Palacios, J., Sullivan, G.D., Bindman, A.B., 2002.
Association of health literacy with diabetes outcomes. JAMA 288,
475–482.
Schoenle, E.J., Schoenle, D., Molinari, L., Largo, R.H., 2002.
Impaired intellectual development in children with Type I
diabetes: association with HbA(1c), age at diagnosis and sex.
Diabetologia 45, 108–114.
Sierra, C., De La Sierra, A., Salamero, M., Sobrino, J., Gomez-
Angelats, E., Coca, A., 2004. Silent cerebral white matter lesions
 ARTICLE IN PRESS
Predictors of cognitive impairment in type 1 diabetes
and cognitive function in middle-aged essential hypertensive
patients. Am. J. Hypertens. 17, 529–534.
Skenazy, J.A., Bigler, E.D., 1984. Neuropsychological findings in
diabetes mellitus. J. Clin. Psychol. 40, 246–258.
Teasdale, T.W., Owen, D.R., Sorensen, T.I., 1991. Intelligence and
educational level in adult males at the extremes of stature.
Hum. Biol. 63, 19–30.
Teasdale, T.W., Sorensen, T.I., Stunkard, A.J., 1992. Intelligence
and educational level in relation to body mass index of adult
males. Hum. Biol. 64, 99–106.
The Psychological Corporation, 1981. The Wechsler Adult Intelligence
Scale-Revised. The Psychological Corporation, San Antonio.
The Psychological Corporation, 1991. The Wechsler Adult Intelli-
gence Scale-Revised as a Neuropsychological Instrument. The
Psychological Corporation, San Antonio.
The Psychological Corporation, 2001. Delis–Kaplan Executive Func-
tion System, Technical Manual. The Psychological Corporation,
San Antonio.
The Psychological Corporation, 2002. WAIS-III and WMS-III, Technical
Manual. The Psychological Corporation, San Antonio.
Tuvemo, T., Jonsson, B., Persson, I., 1999. Intellectual and physical
performance and morbidity in relation to height in a cohort of
18-year-old Swedish conscripts. Horm. Res. 52, 186–191.
Tzourio, C., Dufouil, C., Ducimetiere, P., Alperovitch, A., 1999.
Cognitive decline in individuals with high blood pressure:
1051
a longitudinal study in the elderly. EVA Study Group. Epidemiol-
ogy of Vascular Aging. Neurology 53, 1948–1952.
van Swieten, J.C., Geyskes, G.G., Derix, M.M., Peeck, B.M., Ramos,
L.M., van Latum, J.C., van Gijn, J., 1991. Hypertension in the
elderly is associated with white matter lesions and cognitive
decline. Ann. Neurol. 30, 825–830.
Veldhuis, J.D., Roemmich, J.N., Richmond, E.J., Rogol, A.D.,
Lovejoy, J.C., Sheffield-Moore, M., Mauras, N., Bowers, C.Y.,
2005. Endocrine control of body composition in infancy, child-
hood, and puberty. Endocr. Rev. 26, 114–146.
Verdelho, A., Madureira, S., Ferro, J.M., Basile, A.M., Chabriat,
H.C., Erkinjuntti, T., Fazekas, F., Hennerici, M.M., O’Brien, J.,
Pantoni, L., Salvadori, E., Scheltens, P., Visser, M.C., Wahlund,
L.O., Gunhild Waldemar, G., Wallin, A., Inzitari, D., 2007.
Differential impact of cerebral white matter changes, diabetes,
hypertension and stroke on cognitive performance among non
disabled elderly. The LADIS study. J. Neurol. Neurosurg.
Psychiatry.
Wadsworth, E.J., Moss, S.C., Simpson, S.A., Smith, A.P., 2005. SSRIs
and cognitive performance in a working sample. Hum. Psycho-
pharmacol. 20, 561–572.
Wredling, R., Levander, S., Adamson, U., Lins, P.E., 1990.
Permanent neuropsychological impairment after recurrent epi-
sodes of severe hypoglycaemia in man. Diabetologia 33,
152–157.