Q J Med 2010; 103:495503

doi:10.1093/qjmed/hcq062 Advance Access Publication 28 April 2010

Abnormal glucose metabolism in non-diabetic patients

presenting with an acute stroke: prospective study and
systematic review
J.A. DAVE1, M.E. ENGEL2, R. FREERCKS1, J. PETER1, W. MAY1, M. BADRI2,
L. VAN NIEKERK1 and N.S. LEVITT1
From the 1Division of Diabetic Medicine and Endocrinology and 2Department of Medicine, Groote
Schuur Hospital, 7925 Cape Town, South Africa
Address correspondence to Dr J.A. Dave, Division of Diabetic Medicine and Endocrinology, Department of
Medicine, J-floor, Old Main Building, Groote Schuur Hospital, Anzio Road, Observatory, 7925, Cape Town,
South Africa. email: joel.dave@uct.ac.za
Received 21 August 2009 and in revised form 30 November 2009

Background: Non-diabetic patients presenting with

an acute stroke often have hyperglycaemia. In most
populations it is unknown whether the hyperglycaemia is transient and due to the acute stress
response or whether it represents undiagnosed
abnormal glucose metabolism.
Aim: To evaluate the prevalence and predictors of
persistent hyperglycaemia in non-diabetic patients
with an acute stroke.
Design: Prospective observational study.
Methods: Non-diabetic patients over 40 years old
with an acute stroke were enrolled over a 2-year
period. On admission patients were evaluated with
an HbA1c and a 75 g oral glucose tolerance test
(OGTT). The OGTT was repeated 3 months later.
A meta-analysis was performed to interpret our
results in the context of published data.
Results: One hundred and seven patients were analysed. On admission 26 (24%) patients had diabetes,

Introduction
There is a considerable global burden of diabetes. In
the year 2000, an estimated 171 million people
were affected by diabetes, whilst the excess global

39 (37%) had impaired glucose tolerance and

42 (39%) had normal glucose tolerance. Forty-four
(68%) patients with hyperglycaemia on admission
were re-investigated at least 3 months after
discharge. Of these, 6 (14%) had diabetes,
12 (27%) had impaired glucose tolerance and
26 (59%) had normal glucose tolerance. A 2-h
post-load glucose value 10 mmol/l predicted persistent hyperglycaemia with 72.2% sensitivity,
65.4% specificity and a positive predictive value
and negative predictive value of 59.1 and 77.3%,
respectively. A meta-analysis of prevalence data of
impaired glucose metabolism in non-diabetic individuals 3 months after having had an acute stroke
revealed a combined prevalence of 58% (95%
confidence interval 25.490.5%).
Conclusion: In this study hyperglycaemia in the
setting of an acute stroke was transient in the
majority of patients.

mortality attributable to diabetes was 2.9 million.

This accounted for 5.2% of all deaths.1 This excess
mortality was primarily due to cardiovascular disease (CVD), and is likely to rise as an estimated

! The Author 2010. Published by Oxford University Press on behalf of the Association of Physicians.
All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Downloaded from by guest on May 6, 2015

Summary

496

J.A. Dave et al.

Methods
Patients
Patients without known diabetes who were admitted
to two participating hospitals with a diagnosis of
acute stroke at specific times between July 2004

and 2006 were approached for participation in this

study. Exclusion criteria included: being <40 years
of age; having a haemaglobinopathy or chronic anaemia; being on chronic systemic glucocorticoid
therapy, or having received such therapy in the
last 1 month; inability to give consent. Acute
stroke was defined according to the World Health
Organisation (WHO) criteria i.e. rapidly developing
clinical symptoms or signs of focal disturbance of
cerebral function, lasting more than 24 h, with no
apparent cause other than vascular origin.20 The
diagnosis of stroke was established by history and
neurological examination. Confirmation by computed tomography was obtained in a minority of
patients due to problems of access. The research
protocol was approved by the University of Cape
Town Research and Ethics committee.

Protocol
Each patient fulfilling the selection criteria and
having signed informed consent was enrolled into
the study within 2 days of admission. After a 10-h
overnight fast a plasma glucose and HbA1c were
measured. Serum lipids were measured on Days 2
or 3 and a standard 75 g oral glucose tolerance test
(OGTT) was performed on Days 3 or 4 of admission.
During the acute admission the following sociodemographic and clinical data was collected: age,
gender, ethnicity, previous history of hypertension
and dyslipidaemia, family history of diabetes, smoking, systolic blood pressure (BP), diastolic BP,
height, weight and waist circumference. At least
3 months after discharge the OGTT was repeated
only in those patients diagnosed with hyperglycaemia during the acute admission. A fasting
serum insulin level was measured on the pre-test
fasting blood and the degree of insulin resistance
was estimated by the homeostasis model assessment
(HOMA) method.21
Definitions: diabetesfasting plasma glucose
(FPG) 7 mmol/l or 2-h plasma glucose during the
OGTT  11.1 mmol/l; pre-diabetes2-h plasma
glucose
during
the
OGTT  7.8 mmol/l
but <11.1 mmol/l [impaired glucose tolerance
(IGT)] or FPG  5.6 mmol/l but <7.0 mmol/l [impaired fasting glucose (IFG)]; dysglycaemiadiabetes plus pre-diabetes; normal glucose tolerance
(NGT)FPG < 5.6 mmol/l or a 2-h plasma glucose
during the OGTT <7.8 mmol/l.22,23

Assays
Plasma glucose was measured using the glucose
oxidase method on samples collected in fluoride
tubes. HbA1c analysis was done by the Diabetes
Control and Complications Trial (DCCT) method,

Downloaded from by guest on May 6, 2015

366 million people will have diabetes by the year

2030.2 Diabetes also increases the risk of ischaemic
stroke and is associated with a less favourable outcome than in people without diabetes. The importance of early initiation and maintenance of
glycaemic control on all-cause mortality and
myocardial infarction in people with diabetes has
recently been demonstrated, adding to the
well-recognized benefit of such control on microvascular complications.35 Given these data, ideally
diabetes should be diagnosed early, and treatment
instituted prior to presentation with complications.
Unfortunately this is often not the case. In a number
of studies 5060% of people who presented with an
acute myocardial infarction were found to have undiagnosed diabetes.6,7 In addition, many people
with newly diagnosed diabetes have pre-existing
CVD as demonstrated in the ADDITIONCambridge screening and intervention study, in
which 19% of screen positive people with diabetes
had pre-existing CVD.8
Patients without known diabetes commonly have
hyperglycaemia at presentation of an acute stroke,
making the diagnosis of diabetes difficult as the
hyperglycaemia may occur as an acute stress response,912 may represent previously undiagnosed
impaired glucose metabolism1316 or may be a
marker of infarct size.8,17,18 Although there is no
evidence that rendering these patients euglycaemic
is beneficial from the point of view of mortality and
morbidity from the stroke, there is evidence that preventing persistent hyperglycaemia in patients with
diabetes reduces microvascular disease and
CVD.3,4,19 It is therefore clinically important to recognize these patients on admission so that they may
benefit from long-term treatment with glucose lowering agents. However, there is a paucity of data on
predictors of persistent hyperglycaemia in patients
presenting with an acute stroke who are not
known to have diabetes. We therefore investigated
the prevalence and predictors of persistent hyperglycaemia in these patients. Furthermore, since there is
currently no systematic review of the published literature on the prevalence of persistent hyperglycaemia in non-diabetic patients who have had an
acute stroke, we analysed our results in the context
of published data through a meta-analysis.

Abnormal glucose metabolism in acute stroke

with an upper range of 6.0%. Insulin was measured
using a radioimmunoassay (Roche Modular E170).

Statistical analysis
Normality assumption was tested using Shapiro
Wilks test. Variables failing this assumption were
transformed when appropriate. These variables are
presented as median [interquartile range (IQR)] and
were analysed using the non-parametric Mann
Whitney U-test. Categorical variables are presented
as frequency (percentage) and were compared using
the 2 or Fischers exact test. The relationships between 3-month glycaemic status and biochemical
parameters as explanatory variables were assessed
using multiple and or logistic regression techniques.
A receiver operating characteristic (ROC) curve was
plotted to determine the cut-points for predictors of
dysglycaemia and their sensitivity, specificity, and
predictive values. Statistical analyses were performed using SPSS (version 16.0.1) for Windows
(SPSS Inc.) and STATA (version 10.0).

Systematic review and meta-analysis

Prospective cohort studies investigating the prevalence of dysglycaemia assessed according to
WHO and American Diabetes Association (ADA)
criteria in non-diabetic individuals with acute
stroke were eligible for inclusion.22,23 Studies with
missing admission data or having follow-up data
<3 months were excluded, as were those of chronic
stroke or acute stroke due to prosthetic valves.

Study identification and eligibility.

Two independent observers (J.A.D. and M.E.E.) identified studies from the MEDLINE database (from its earliest
date until December 2008) using a predetermined
search strategy incorporating the terms ("GLUCOSE
METABOLISM
DISORDERS"[MESH]
AND
"CEREBROVASCULAR DISORDERS"[MESH]) NOT
("MYOCARDIAL ISCHEMIA"[MESH] OR "HEART
DISEASES"[MESH]) limited to articles with abstracts
of studies conducted in adult participants. This process was complemented by reviewing citations and
searching in Google Scholar. Language of publication was restricted to English articles only. Titles and
abstracts were screened, and full text articles obtained from potentially eligible reports. We immediately excluded editorials and review articles.

Data extraction. From each study, we (J.A.D. and

M.E.E.) independently recorded year of publication,
study design, origin and demographics of participants, diagnostic criteria, disease information,

prevalence and standard error of the estimate.

Where not provided, confidence intervals (CIs)
were incorporated into the formula, SE = (upper
limit lower limit)/3.92.

Data synthesis.

Prevalence data from individual

studies were combined by random-effects
meta-analysis according to the MantelHaenszel
method. Heterogeneity was evaluated using the
2-based Q statistic (significant for P < 0.1).
We also used the I2 statistic, considering values
above 56% to be indicative of notable heterogeneity (Higgins). STATA software version 9.2 (STATA
Corporation, College Station, Texas) was used
to perform calculations and the meta-analysis
and to produce the forest plots using the metan
routine.

Results
One hundred and twenty-one patients were enrolled. Fourteen patients were excluded (one patient
tested HIV-positive, three patients died before completing the questionnaire or admission OGTT, nine
patients did not have an OGTT during the acute
admission and one patient withdrew from the
study) leaving 107 eligible patients who were
included in this report.

Acute admission
Baseline clinical and biochemical characteristics are
summarized in Table 1. Sixty-five patients (61%)
were dysglycaemic [26 (24%) had diabetes and 39
(37%) had pre-diabetes] and 42 (39%) had NGT.
The dysglycaemic patients were more likely to be
female (P < 0.01), had previous hypertension
(P = 0.02) and a family history of diabetes
(P < 0.01). In addition, they had higher systolic
(P < 0.001) and diastolic (P < 0.01) BPs, a higher
total cholesterol (P < 0.01) and a higher low-density
lipoprotein (LDL) cholesterol (P < 0.01) than the patients with NGT. Unexpectedly, the latter patients
were more likely to have ever smoked than the dysglycaemic patients (P < 0.01). This may be because
more women were dysglycaemic and there were
less women that ever smoked than men
(P < 0.0001).

Subsequent review 3 months after

discharge
Of the 65 patients with dysglycaemia on admission,
44 were re-investigated at 3 months after discharge; 41 with an OGTT and 3 patients with a
FPG and HbA1c because they were on oral

Downloaded from by guest on May 6, 2015

Review inclusion and exclusion criteria

497

498

J.A. Dave et al.

Table 1 Characteristics of the study population during the acute admission

Parameter

Dysglycaemia

59 (4969)
29 (69)

62 (5371)
28 (43)

0.23
<0.01

31
7
1
3
32 (76)

47
8
3
7
30 (46)

<0.01

19 (45)
1 (2)
5 (12)

44 (68)
3 (5)
23 (35)

0.02
0.49
<0.01

5.1
6.3
5.6
23
83
145
84
4.7
1.3
1.1
2.9

(4.65.2)
(5.56.8)
(5.55.8)
(2228)
(7592)
(132156)
(7592)
(3.85.4)
(1.11.7)
(0.91.3)
(2.23.5)

5.8
9.5
5.8
26
90
159
91
5.2
1.3
1.1
3.3

(5.36.8)
(8.312.2)
(5.46.3)
(2329)
(8098)
(145183)
(80104)
(4.56.0)
(1.01.7)
(0.91.3)
(2.83.9)

<0.01
<0.01
0.06
0.05
0.06
<0.001
<0.01
0.01
0.90
0.91
<0.01

Categorical variables presented as n (%) with P-values by the 2 test. Continuous variables presented as median (IQR) with
P-values by the MannWhitney U-test for non-parametric data. BP: blood pressure.

hypoglycaemic agents (Table 2). The remaining 21

patients were not re-assessed due to death after discharge [13 (20%) patients: eight patients with diabetes and five patients with pre-diabetes] and
inadequate contact details [8 (12%) patients].
Eighteen (41%) of the 44 patients re-investigated remained dysglycaemic [6 (14%) had diabetes, 12
(27%) had pre-diabetes] and 26 (59%) had NGT
(Table 2). The three patients on oral hypoglycaemic
agents were considered to have diabetes as supported by their elevated FPG and HbA1c despite
treatment (Patient A: FPG 6.6 mmol/l, HbA1c 6.7%;
patient B: FPG 6.8 mmol/l, HbA1c 6.6%; patient C:
FPG 6.8 mmol/l, HbA1c 6.3%). Of the 26 patients
classified with diabetes during the acute admission,
5 (19%) remained diabetic, 6 (23%) had
pre-diabetes and 6 (23%) had NGT (Table 3). In
the 39 patients with pre-diabetes during the acute
admission, 1 (3%) developed diabetes, 6 (15%)
remained with pre-diabetes and 20 (51%) reverted
to NGT (Table 3). Patients with dysglycaemia had a
higher median HbA1c (5.95%, IQR 5.56.4 vs.
5.5%, IQR 5.26.1; P = 0.04) and median homeostasis model assessment of insulin resistance

(HOMA-IR) (3.5, IQR 2.74.7 vs. 1.4, IQR 0.92.2;

P  0.001) than those with NGT, but age, gender,
past history of hypertension or dyslipidaemia,
family history of diabetes, current/previous smoking,
body mass index (BMI), waist circumference, BP,
triglycerides and high-density lipoprotein (HDL)
did not differ between those with NGT or
dysglycaemia.
On admission, patients that subsequently
died had a significantly higher FPG than
those alive at follow-up (10.4  3.9 mmol/l, 95%
CI: 8.8; 12 vs. 8.6  3.3 mmol/l, 95% CI: 7.9; 9.3,
P = 0.043) but there was no significant difference in
HbA1c.
Logistic regression analysis revealed the 2-h
post-load glucose value on admission to be the
only significant predictor of persistent dysglycaemia
3 months after discharge. A 2-h post-load glucose
value 10 mmol/l predicted dysglycaemia with
72.2% sensitivity, 65.4% specificity and gave a positive predictive value (PPV) and negative predictive
value (NPV) of 59.1 and 77.3%, respectively
(Table 4). The area under the ROC curve was 0.76
(95% CI: 0.610.90).

Downloaded from by guest on May 6, 2015

Socio-demographic characteristics
Age (years)
Male gender
Ethnicity
Mixed
Black
Asian
White
History of smoking
Clinical characteristics
Previous hypertension
Previous dyslipidaemia
Family history of diabetes
OGTT
FPG (mmol/l)
2-h serum glucose (mmol/l)
HbA1c (%)
BMI (kg/m2)
Waist circumference (cm)
Systolic BP (mmHg)
Diastolic BP (mmHg)
Total cholesterol (mmol/l)
Triglyecrides (mmol/l)
HDL (mmol/l)
LDL (mmol/l)

NGT

Abnormal glucose metabolism in acute stroke

499

Table 2 Characteristics of the subjects that were dysglycaemic on admission and who were re-tested after 3 months
NGT

Dysglycaemia

n
Age (years)
Female gender
Previous hypertension
History of smoking
Previous dyslipidaemia
Family history of diabetes mellitus
Admission
BMI (kg/m2)
Waist circumference (cm)
Systolic BP (mmHg)
Diastolic BP (mmHg)
HbA1c (%)
OGTT
FPG (mmol/l)
2-h plasma glucose (mmol/l)
Total cholesterol (mmol/l)
Triglyecrides (mmol/l)
HDL (mmol/l)
LDL (mmol/l)
3 months
OGTT
FPG (mmol/l)
2-h plasma glucose (mmol/l)
Fasting serum insulin (mU/l)
HOMA
Time to re-testing (days)

26
64.0 (5272)
13 (50)
18 (69)
11 (42)
2 (8)
7 (27)

18
65 (5373)
13 (72)
12 (67)
11 (61)
1 (6)
5 (28)

0.95
0.14
0.86
0.22
0.64a
0.95

26
93
167
94
5.5

(2329)
(7899)
(150191)
(85104)
(5.26.1)

27
93
159
87
6.0

(2430)
(87100)
(140210)
(76104)
(5.56.4)

0.33
0.90
0.70
0.21
0.04

5.7
9.1
5.5
1.3
1.1
3.5

(5.36.3)
(8.210.3)
(4.96.0)
(1.11.5)
(0.91.3)
(3.04.1)

6.4
10.9
5.3
1.5
1.1
3.3

(5.86.8)
(9.414.4)
(4.76.1)
(1.22.3)
(0.81.2)
(2.93.9)

0.15
0.004
0.69
0.12
0.39
0.65

5.0
5.7
6.5
1.4
147

(4.85.3)
(5.06.7)
(4.311.0)
(0.92.2)
(122203)

6.0
8.4
14.4
3.5
130

(5.66.6)
(6.210.1)
(11.019.1)
(2.74.7)
(98217)

<0.001
<0.001
0.001
<0.001
0.41

Categorical variables presented as number (%) with P-values by the 2 test. Continuous variables presented as median (IQR)
with P-values by the MannWhitney Utest for non-parametric data. BP: blood pressure.
a
Fischer-exact test.
Table 3 Changing prevalence of glycaemic status of the dysglycaemic patients from admission to re-testing
Glycaemic status on admission

Diabetes mellitus
IGT

26
39

Glycaemic status 3 months

DM

IGT

NGT

5 (19)
1 (3)

6 (23)
6 (15)

6 (23)
20 (51)

Died

Lost to F/U

5 (19)
8 (21)

4 (15)
4 (10)

Data are n (%). F/U: follow-up.

Systematic review and meta-analysis

The search strategy yielded 935 citations. Records
were screened by title after which 48 articles were
deemed to be potentially relevant. Abstracts were
evaluated by two observers working independently
and 43 studies were excluded; following full-text
scrutiny, one publication was excluded, while for
a further two, the full text was unavailable.
Reasons for exclusion were known diabetics
included (n = 3), no glycaemic data provided

(n = 5), no admission OGTT and no follow-up data

at 3 months (n = 35).

Characteristics of the included studies

One publication (Vancheri) satisfied our inclusion
criteria (Table 5). Together with our unpublished
data, the eligible studies pertained to two distinct
study populations comprising 213 participants with
median ages of 71.0 and 61.0 years, respectively.
The proportions of men were 61 and 53%,

Downloaded from by guest on May 6, 2015

Parameter

500

J.A. Dave et al.

Table 4 Sensitivity, specificity and predictive values of the 2-h post-load glucose on admission for predicting dysglycaemia
at 3 months after discharge
Two-hour post-load
glucose (mmol/l)
on admission

Sensitivity (%)

Specificity (%)

PPV (%)

NPV (%)

6
7
8
9
10
11
12
13
14
15

100.0
100.0
94.4
88.9
72.2
50.0
38.9
38.9
27.8
22.2

3.8
7.7
15.4
50.0
65.4
84.6
88.5
96.2
96.2
96.2

41.9
42.9
43.6
55.2
59.1
69.2
70.0
87.5
83.3
80.0

100.0
100.0
80.0
86.7
77.3
71.0
67.7
69.4
65.8
64.1

(82.4100)
(82.4100)
(74.299.0)
(67.296.9)
(49.187.5)
(29.071.0)
(20.361.4)
(20.361.4)
(12.550.9)
(9.045.2)

(0.718.9)
(2.124.1)
(6.233.5)
(32.167.9)
(46.280.6)
(66.5093.9)
(71.096.0)
(81.199.3)
(81.199.3)
(81.199.3)

(28.456.7)
(29.157.8)
(29.359.0)
(37.671.6)
(38.876.7)
(42.487.3)
(39.789.2)
(52.997.8)
(43.797.0)
(37.696.4)

(20.7100)
(34.2100)
(37.696.4)
(62.196.3)
(56.689.9)
(53.483.9)
(50.880.9)
(53.182.0)
(49.978.8)
(48.477.3)

Table 5 General characteristics of studies included in the systematic review and meta-analysis
Dave n (%)

106
81 (76)
65 (61)
Italy
Hospital
15 (16)
81 (84)
21 (26)
60 (74)
10 (10)

107
44a (41)
57 (53)
South Africa
Hospital
42 (39)
65 (61)
26 (59)
18 (41)
21 (32)

Combined
ES (95% CI)

0.58 (0.25; 0.90)

n: number; ES: effect size.

a
Only patients that were dysglycaemic at admission were re-evaluated.

respectively. For the outcome of interest, a total of

125 subjects were included in the analysis. Both
were observational studies conducted in Italy and
South Africa, respectively and both classified impaired glucose metabolism according to WHO and
ADA criteria.
Meta-analysis of prevalence data of dysglycaemia
in non-diabetic individuals 3 months after having
had an acute stroke revealed a combined prevalence of 58% (95% CI: 25.490.5%) (Table 5).
Statistically significant heterogeneity [heterogeneity
2 significant (P < 0.01), I2 > 90%] was found across
the studies.

Discussion
In this prospective study, the second to evaluate glucose homeostasis in patients without a prior diagnosis of diabetes and an acute stroke, using a FPG and

OGTT on admission and again within 312 months

after discharge, 61% of patients were found to be
dysglycaemic (24% had diabetes and 36%
pre-diabetes) on admission. However, by 312
months after discharge the majority (59%) had
NGT leaving an overall prevalence of persistent dysglycaemia of 21% (7% with diabetes and 14% with
pre-diabetes). The 2-h post-load blood glucose on
admission was most predictive of dysglycaemia at
3 months.
Although the prevalence of dysglycaemia on admission was high in this study, it was lower than that
reported in the only other study using a FPG and
OGTT to evaluate glucose homeostasis on admission and again at least 3 months later.16 In that
study, the prevalence of dysglycaemia on admission
was 84.3% (45.8% had diabetes and 38.5% IGT).
The higher prevalence may be due to the older age
of their patients (69.6 years, IQR 63.276.7 vs.
61 years, IQR 51.071.0). Interestingly, the majority

Downloaded from by guest on May 6, 2015

n at start
n at follow-up
n (male)
Region
Source
Normoglycaemia at admission
Dysglycaemia at admission
Normoglycaemia at 3 months
Dysglycaemia at 3 months
Missing/excluded/died

Vancheri n (%)

Abnormal glucose metabolism in acute stroke

in large part to patients giving incorrect contact

details so as to qualify for admission to their hospital
of choice rather than the hospital closest to where
they live. If one assumes that all those not
re-examined became euglycaemic (best case
scenario) or that they remained dysglycaemic
(worst case scenario) then the lowest and highest
prevalences of dysglycaemia are 28 and 60%,
respectively. Even at 60%, the worst case scenario
provides a much lower prevalence than described
by Vancheri et al.16
A dearth of factors examined on admission
(including age, gender, smoking, previous history
of hypertension, family history of diabetes, BP,
lipids, BMI, waist circumference, HbA1c, fasting insulin and HOMA) proved to be predictive of dysglycaemia at follow-up. Both the present study and that
of Vancheri et al. found the 2-h post-load plasma
glucose on admission to be most predictive of dysglycaemia at 3 months. This may be somewhat surprising as the poor reproducibility of the 2-h
post-load glucose would be expected to cast doubt
on its ability to be a predictive test. The use of a
composite score including multiple diabetes risk
factors such as waist girth or BMI, family history of
diabetes, age and levels of physical activity may
prove to be more useful and its utility should be
explored in a larger cohort. This would permit
early introduction of appropriate glucose lowering
therapy and attainment of euglycaemia or close to
euglycaemia immediately after the stroke, with the
recognized benefits.
We attempted to analyse our results in the context
of existing studies. However this systematic review
highlights the lack of well-designed prospective studies utilizing both a FPG and 2-h post-load plasma
glucose to identify dysglycaemic patients (as recommended by the WHO and ADA). We believe that
most studies were likely to have missed patients with
abnormal glucose metabolism on admission as they
did not use a FPG and OGTT to diagnose dysglycaemia. In addition, most studies do not have
follow-up glycaemic data and are therefore unable
to assess the prevalence of persistent dysglycaemia.
The studies by Kernan et al.14 and Gray et al.31 contain follow-up glycaemic data and are similar to
those included in the meta-analysis, but are likely
to have missed patients with abnormal glucose metabolism, as no OGTT was done on admission and
only patients whose fasting blood glucose was
within a specifically defined range were included.
Lam et al. report a prevalence of 17 and 26% of
diabetes and IGT, respectively in Chinese patients
with a stroke and no prior diagnosis of diabetes.28 As
their report does not contain glycaemic data on admission, it is not possible to determine whether their

Downloaded from by guest on May 6, 2015

(74%) of their patients with dysglycaemia on admission remained dysglycaemic when re-tested
3 months after discharge (43% had diabetes and
31% IGT) leaving their overall prevalence of dysglycaemia at 65% (37.5% with diabetes and 27.1%
with IGT), suggesting a pre-existent abnormality of
glucose metabolism in the majority of their patients.
On the other hand, the majority of dysglycaemic
patients in our study reverted to euglycaemia, indicating that the hyperglycaemia on admission was
likely to be due to the acute stress response. It has
been debated whether the acute stress response
plays a significant physiological role, nevertheless,
it is well-documented that the diabetogenic hormones cortisol and catecholamines are elevated in
some patients with an acute stroke.11,12,24 Whilst
even in a population-based survey in rural
Tanzania 80% of subjects with IGT reverted to
NGT within 5 days, partly attributable to the orienting reflex in BP measurements in a population unfamiliar with blood testing.25
Patients in developing countries present later in
the course of their illness, which in the case of an
acute stroke may result in a larger infarct area and
higher blood glucose. The causal relationship
between hyperglycaemia and larger infarcts remains
unknown, but is speculated to be due to increased
oxidative stress and inflammation found in the setting of hyperglycaemia.26 Interestingly, the patients
in this study that died had a higher fasting blood
glucose level on admission than those that survived
yet there was no difference in HbA1c, suggesting
pre-stroke euglycaemia and possibly larger and
more severe cerebral infarcts. This is consistent
with the study by Murros et al. that showed that
pre-stroke hyperglycaemia (as suggested by an
increased HbA1c) did not have any predictive
value concerning stroke outcome but that
post-stroke fasting hyperglycaemia correlated
strongly with stroke severity and predicted stroke
outcome.27 They suggest that a high fasting blood
glucose after a stroke reflects a stress response to a
more severe ischemic brain lesion. Other studies in
animals and humans have shown an association
between hyperglycaemia and worse outcome
after stroke in terms of both mortality and
morbidity.18,2830
Our study differs from most other studies that have
assessed hyperglycaemia in the acute stroke setting
in that it was designed a priori to examine the question of persistent hyperglycaemia. For that reason,
only patients with hyperglycaemia on admission
were re-examined. However, we were only able to
re-investigate 68% of the dysglycaemic patients as
13 (20%) had died and 8 (12%) were lost to
follow-up. The latter observation is probably due

501

502

J.A. Dave et al.

Acknowledgement
The authors wish to thank Dr Frances Wilson for
helping with the recruitment of patients.

Funding
The Medical Research Council of South Africa and
the University of Cape Town.
Conflict of interest: None declared.

References
1. Roglic G, Unwin N, Bennett PH, Mathers C, Tuommilehto J,
Nag S, et al. The burden of mortality attributable to diabetes
realistic estimates for the year 2000. Diab Care 2005;
28:21305.
2. Wild S, Roglic G, Green A, Sicree R, King H.
Global prevalence of diabetesestimates for the year
2000 and projections for 2030. Diab Care 2004;
27:104753.
3. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK
Prospective Diabetes Study (UKPDS) Group. Lancet 1998;
352:83753.
4. Holman RR, Paul SK, Bethel MA, Matthews DR, Neil HAW.
10-year follow-up of intensive glucose control in type 2 diabetes. New Engl J Med 2008; 359:157789.
5. Nathan DM, Cleary PA, Backlund JY, Genuth SM, Lachin JM,
Orchard TJ, et al. Intensive diabetes treatment and cardiovascular disease in patients with type 1 diabetes. N Engl J Med
2005; 353:264353.
6. Norhammar A, Tenerz A, Nilsson G, Hamsten A, Efendic S,
Ryden L, et al. Glucose metabolism in patients with
acute myocardial infarction and no previous diagnosis of
diabetes mellitus: a prospective study. Lancet 2002;
359:21404.
7. Lankisch M, Futh R, Guker H, Lapp H, Bufe A, Haastert B,
et al. Screening for undiagnosed diabetes in patients with
acute myocardial infarction. Clin Res Cardiol 2008;
97:7539.
8. Echouffo-Tcheugui JB, Sargeant LA, Prevost AT,
Williams KM, Barling RS, Butler R, et al. How much might
cardiovascular disease risk be reduced by intensive therapy
in people with screen-detected diabetes? Diabet Med 2008;
25:14339.
9. Al Himyari FA, Abbas FN. Stress hyperglycemia in nondiabetic Iraqi patients presenting with acute stroke. Endocr Pract
2007; 13:6912.
10. Allport LE, Butcher KS, Baird TA, Macgregor L,
Desmond PM, Tress BM, et al. Insular cortical ischemia is
independently associated with acute stress hyperglycemia.
Stroke 2004; 35:188691.
11. Christensen H, Boysen G, Johannesen HH. Serum-cortisol
reflects severity and mortality in acute stroke. J Neurol Sci
2004; 217:17580.
12. ONeill PA, Davies I, Fullerton KJ, Bennett D. Stress hormone
and blood glucose response following acute stroke in the
elderly. Stroke 1991; 22:8427.
13. Lam KS, Ma JT, Woo E, Lam C, Yu YL. High
prevalence of undiagnosed diabetes among Chinese
patients with ischaemic stroke. Diab Res Clin Pract 1991;
14:1337.
14. Kernan WN, Viscoli CM, Inzucchi SE, Brass LM, Bravata DM,
Shulman GI, et al. Prevalence of abnormal glucose tolerance
following a transient ischemic attack or ischemic stroke.
Arch Intern Med 2005; 165:22733.
15. Gray CS, Scott JF, French JM, Alberti KGMM, OConnell JE.
Prevalence and prediction of unrecognised diabetes mellitus
and impaired glucose tolrance following acute stroke.
Age Ageing 2004; 33:717.

Downloaded from by guest on May 6, 2015

reported prevalence is of persistent dysgycaemia or

whether the dysglycaemia developed in these patients as a result of the stroke i.e. less physical activity, weight gain and the use of diabetogenic drugs
such as b-blockers and/or thiazide diuretics. In our
meta-analysis, the results showed significant heterogeneity. Procedurally, the studies were similar in
almost every respect and thus, we suspect that this
variation is due to the difference in the number of
participants for which follow-up data were available. Nevertheless, a meta-analysis of these studies
is still useful in providing an idea of the overall
prevalence which indicates a combined prevalence
of persistent dysglycaemia of 58%.
It is concerning that a significant number of stroke
patients with no prior history of diabetes have their
first clinical presentation of dysglycaemia as an
acute stroke. It seems that opportunistic screening
of individuals with risk factors for diabetes who
attend a health service will allow for earlier diagnosis and initiation of therapy and a chance to avert or
delay micro and macrovascular complications.
Indeed, the ADDITION-Cambridge study has
shown that people with screen-detected type 2 diabetes have an adverse cardiovascular risk profile
and that a significant absolute reduction in this risk
is achievable through multifactorial therapies.8
Since a proportion of hyperglycaemic non-diabetic
stroke patients will have transient hyperglycaemia
or prediabetes, it is important that these patients
undergo strict lifestyle modification and are
re-assessed with an OGTT at least 3 months after
discharge.
In conclusion, we report a high prevalence of
transient dysglycaemia in patients with an acute
stroke. It is concerning that at least 21% of patients
in our study had undiagnosed dysglycaemia. In
agreement with studies post-myocardial infarction
and in the absence of significant predictive factors
we suggest a follow-up OGTT at least 3 months after
discharge in hyperglycaemic acute stroke patients
with no prior history of diabetes.

Abnormal glucose metabolism in acute stroke

503

16. Vancheri F, Curcio M, Burgio A, Salvaggio S, Gruttadauria G,

Lunetta MC, et al. Impaired glucose metabolism in patients
with acute stroke and no previous diagnosis of diabetes mellitus. QJM 2005; 98:8718.

23. World Health Organization. Definition, Diagnosis

and Classification of Diabetes Mellitus. Part I. Diagnosis
and Classification of Diabetes Mellitus. Geneva, WHO,
1999.

17. Candelise L, Landi G, Orazio EN, Boccardi E. Prognostic

significance of hyperglycemia in acute stroke. Arch Neurol
1985; 42:6613.

24. Myers MG, Norris JW, Hachniski VC, Sole MJ. Plasma norepinephrine in stroke. Stroke 1981; 12:2004.

18. Baird TA, Parsons MW, Phanh T, Butcher KS, Desmond PM,
Tress BM, et al. Persistent poststroke hyperglycemia is independently associated with infarct expansion and worse clinical outcome. Stroke 2003; 34:220814.
19. Gray CS, Hildreth AJ, Sandercock PA, OConnell JE,
Johnston DE, Cartlidge NE, et al. Glucose-potassium-insulin
infusions in the management of post-stroke hyperglycaemia:
the UK Glucose Insulin in Stroke Trial (GIST-UK). Lancet
Neurol 2007; 6:397406.
20. Stroke1989. Recommendations on stroke prevention, diagnosis, and therapy. Report of the WHO Task Force on
Stroke and other Cerebrovascular Disorders. Stroke 1989;
20:14071431.
21. Bonora E, Targher G, Alberiche M, Bonadonna RC,
Saggiani F, Zenere MB, et al. Homeostasis model assessment
closely mirrors the glucose clamp technique in the sssessment of insulin sensitivity studies in subjects with various
degrees of glucose tolerance and insulin sensitivity. Diab
Care 2000; 23:5763.

26. Bemeur C, Ste-Marie L, Montgomery J. Increased oxidative

stress during hyperglycemic cerebral ischemia. Neurochem
Int 2007; 50:890904.
27. Murros K, Fogelholm R, Kettunen S, Vuorela AL, Valve J.
Blood glucose, glycosylated haemoglobin, and outcome of
ischemic brain infarction. J Neurol Sci 1992; 111:5964.
28. Woo E, Chan YW, Yu YL, Huang CY. Admission glucose
level in relation to mortality and morbidity outcome in 252
stroke patients. Stroke 1988; 19:18591.
29. Pulsinelli W, Waldman S, Sigsbee B, Rawlinson D, Scherer P,
Plum F. Experimental hyperglycemia and diabetes mellitus
worsen stroke outcome. Trans Am Neurol Assoc 1980;
105:214.
30. Kagansky N, Levy S, Knobler H. The role of hyperglycemia in
acute stroke. Arch Neurol 2001; 58:120912.
31. Gray CS, Scott JF, French JM, Alberti KG, OConnell JE.
Prevalence and prediction of unrecognised diabetes mellitus
and impaired glucose tolerance following acute stroke.
Age Ageing 2004; 33:717.

Downloaded from by guest on May 6, 2015

22. American Diabetes Association. Diagnosis and classification

of diabetes. Diab Care 2008; 31:s627.

25. Yudkin JS, Alberti KG, Mclarty DG, Swai AB. Impaired glucose tolerance. BMJ 1990; 301:397402.