OBJECTIVE

To examine the association of diabetes and glycated hemoglobin (HbA1c) with the risk of intracerebral hemorrhage (ICH) in a large population-based cohort.

RESEARCH DESIGN AND METHODS

The computerized database of the largest health care provider in Israel was used to identify adult members aged 40 years or older and alive at 1 January 2010 (297,486 with diabetes and 1,167,585 without diabetes). The cohort was followed until 31 December 2017 for incidence of ICH. Multivariable Cox proportional hazards regression models, adjusted for baseline disease risk score, were applied to estimate the hazard ratio (HR) of ICH.

RESULTS

Overall 4,170 ICH cases occurred during 10,730,915 person-years of follow-up. Diabetes was independently associated with increased ICH risk, with hazard ratio (HR) 1.36 (95% CI 1.27–1.45), and increased with longer diabetes duration: 1.23 (1.12–1.35) and 1.44 (1.34–1.56) for diabetes duration ≤5 years and >5 years, respectively. The increased ICH risk associated with diabetes was more pronounced in patients ≤60 years old (Pinteraction <0.001). Among patients with diabetes, HbA1c had a nonlinear J-shaped relationship with ICH (P for nonlinearity = 0.0186). Compared to the fourth HbA1c decile, 6.5–6.7% (48–50 mmol/mol), the HR for ICH was 1.27 (1.01–1.59) and 2.19 (1.75–2.73) in the lowest HbA1c decile, ≤6.0% (≤42 mmol/mol), and highest HbA1c decile, >9.3% (>78 mmol/mol), respectively.

CONCLUSIONS

Diabetes is associated with increased risk of ICH that is directly associated with diabetes duration. ICH and HbA1c appear to have a J-shaped relationship, suggesting that both poor control as well as extreme intensive diabetes control might be associated with increased risk.

Spontaneous intracerebral hemorrhage (ICH) is a devastating condition accounting for 10–15% of all stroke cases. It is associated with a dismal prognosis, as only 38% of affected patients survive the first year (1).

Type 2 diabetes affects more than 415 million adults worldwide and is a well-known contributor to cardiovascular morbidity, cognitive decline, and all-cause mortality (2). Although diabetes is an independent risk factor for ischemic stroke (3), as yet there is no conclusive evidence for the association between diabetes and ICH, as previous studies showed conflicting results (48).

In addition, low and high fasting blood glucose concentrations were associated with higher risk of incident ICH in populations with diabetes and populations without diabetes (8,9). No previous studies however have looked at the association of glycated hemoglobin (HbA1c), a much more accurate measure of glycemic control than glucose levels (10), and ICH risk. We sought to determine 1) the association of diabetes and ICH and 2) the relationship between HbA1c levels and ICH in a large nationwide population-based cohort.

Source of Data

This study is based on data from the computerized database of Clalit Health Services (CHS), which provides inclusive health care for more than half of the Israeli population. Health care coverage in Israel is mandatory according to the National Health Insurance Law (1995) and is provided by four groups akin to not-for-profit HMOs. All members of the different HMOs have a similar health insurance plan and similar access to health services. The electronic medical record database of CHS includes data from multiple sources: records of primary care physicians, community specialty clinics, hospitalizations, laboratories, and pharmacies. A registry of chronic disease diagnoses is compiled from these data sources. Diagnoses are captured in the registry by diagnosis-specific algorithms, using ICD-9 code reading, text reading, laboratory test results, and disease-specific drug usage. A record is kept of the data sources and dates used to establish the diagnosis, with the earliest recorded date, from any source, considered to be the defining date of diagnosis. A number of high-quality, population-based studies have been conducted based on the data retrieved from the CHS database (11,12).

Selection of Study Population

The study cohort consisted of all CHS adult members aged 40 years or older and alive at 1 January 2010 (n = 1,496,094). Individuals with <2 years of continuous membership in the CHS prior to cohort entry date were excluded (n = 15,379). Of the remaining 1,480,715 individuals, 313,130 patients had a preexisting diagnosis of diabetes and 1,167,585 individuals were without diabetes. Patients with diabetes had to have at least one test result for HbA1c in the 2 years before cohort entry (n = 297,486). Cohort participants (n = 1,465,071) were followed-up until reaching the study outcome (ICH), death, loss to follow-up, or end of follow-up at 31 December 2017—whichever came first.

Study Variables

The outcome of interest was ICH, defined as primary discharge diagnosis with ICH (ICD-9 code 431). A systematic review of studies that evaluated the accuracy of using administrative data for identifying ICH showed that the positive predictive value ranged from 79 to 97% for the inpatient ICD-9 code 431 (13). We conducted a validation study to evaluate the accuracy of using inpatient ICD-9 code 431 to identify ICH in the CHS database. We reviewed the medical files of 1,614 cases of ICH that were detected in 297,486 patients with diabetes; unfortunately 110 files could not be reached, and of the remaining 1,504 files there was a convincing evidence for secondary ICH in 78 cases and spontaneous ICH in 1,222 cases (805 deep and 355 lobar, and in 62 patients location could not be determined), yielding a positive predictive value 86.4%, in line with the systemic review (13).

The diagnosis of diabetes was retrieved from the CHS chronic disease registry that relies on different sources, including clinical diagnosis of diabetes (ICD-9 code 250), HbA1c ≥6.5% [>48 mmol/mol], and diabetes-specific drug usage (ATC code A10). Data on medication use are considered to be complete because of the very low copayment required in the CHS making it unlikely that prescription medications are purchased in non-CHS pharmacies. It should be noted that diabetes has been used along with five other areas of health care to assess the quality of community health care in Israel; hence, health care providers, including CHS, have invested efforts to achieve accurate diagnosis of diabetes (14).

HbA1c test results performed, in patients with diabetes, during the 2 years prior to cohort entry were retrieved from the CHS laboratory database. The most recent HbA1c test performed during this period was used to assess the association between HbA1c and ICH in the main analysis, and the median HbA1c of all tests performed in this period was used in separate sensitivity analyses. HbA1c was studied as a continuous variable and for descriptive purpose was also classified into deciles based on its distribution among patients with diabetes.

In addition, for each patient the following baseline data were retrieved from the computerized database of the CHS: demographic and other descriptive variables, health habits, socioeconomic status (SES) defined based on the SES score of the clinic neighborhood as defined by the Israel Central Bureau of Statistics, LDL cholesterol (LDL), HDL cholesterol (HDL), presence of selected chronic medical conditions and ICH risk factors, medication use of selected drug categories, and medical services utilization as shown in Table 1. SES, LDL, and HDL had missing values (0.47%, 16.3%, and 15.1%, respectively); hence, these variables were used as categorical variables that include a category of missing values.

Statistical Methods

Statistical analyses were performed using IBM SPSS Statistics 22.0 (IBM, New York, NY), and SAS, version 9.4, software. For all analyses, P < 0.05 for the two-tailed tests was considered statistically significant. The association between diabetes and ICH was evaluated using Cox proportional hazards regression models to estimate the crude and the adjusted hazard ratio (HR). Further, patients with diabetes were classified according to their HbA1c distribution into deciles, with the group without diabetes serving as the reference category. In the analysis limited to patients with diabetes, HbA1c was first modeled as a continuous variable. Nonlinearity of the relationship of HbA1c with ICH was tested using a likelihood ratio test comparing two Cox regression models: one including only linear effect and the second including also quadratic and cubic terms. For graphical presentation, a smoothed plot of adjusted HR (relative to an HbA1c reference value of 6.5% [48 mmol/mol]) was estimated along with point-wise 95% CI. For this purpose, HbA1c was flexibly modeled in a Cox regression using restricted cubic spline function with five knots corresponding to ∼5%, 25%, 50%, 75%, and 95% percentiles of HbA1c among patients with diabetes (15). In addition, because of the nonlinear relationship with ICH, HbA1c was also modeled as a categorical variable (deciles), with the fourth decile 6.5–6.7% (48–50 mmol/mol) used as the reference category.

Because of the large number of potential confounders, we performed adjustment for a disease risk score (DRS), a summary measure of disease probability. The DRS was estimated using a Cox proportional hazards regression model for ICH outcome that included most clinically relevant ICH risk factors and other clinical covariates likely to be correlated with ICH (Table 1). In comparison with conventional multivariate analyses, adjustment for the single variable DRS increases the efficiency of the analyses (16,17). It has been shown than the DRS and propensity score methods had comparable performance and that DRS has an advantage when multiple comparison groups are studied (16,17).

In addition, we performed the following sensitivity analyses: 1) for assessment of the association between diabetes and ICH, instead of adjustment for DRS we adjusted for propensity score (probability of having diabetes) that was calculated using a logistic regression model that included all variables in Table 1; 2) we repeated the analysis by adjusting for the most clinically relevant variables (i.e., standard multivariable regression); 3) because in patients without diabetes ICH may be mediated by newly diagnosed diabetes occurring after study entry, we performed two sensitivity analyses, firstly by censoring at the time of diabetes diagnosis and secondly by excluding individuals with diabetes detected during follow-up; 4) instead of using the most recent HbA1c test result for assessing the association with ICH, we repeated all analyses using the median HbA1c of all tests performed in the 2 years prior to study entry; 5) we examined whether the association between diabetes and ICH was modified by age, sex, and hypertension. Finally, we performed an additional sensitivity analysis to assess the relationship between HbA1c and ICH type (deep vs. lobar). This analysis was restricted to patients with diabetes (n = 297,486), in whom the diagnosis of ICH was validated and classified into lobar and deep ICH by reviewing patients’ files, as described above. The multiple imputation procedure (SAS MI procedure) was performed to account for missing type of ICH. Five imputation data sets were created, where all model variables were included in the imputation process.

A total of 1,465,071 patients (297,486 with diabetes and 1,167,585 without diabetes) with mean (SD) age 60.1 (13.4) years were included in the study, of whom 792,882 (54.1%) were female. Table 1 shows the distribution of baseline demographic and clinical characteristics among the study population with and without diabetes. Patients with diabetes were older with lower SES and generally were more likely to have comorbidities and to use medication and health services. Previous history of ICH was detected in 0.30% of patients with diabetes compared with 0.12% in patients without diabetes (P < 0.001). Anticoagulant use was more frequent (4.7%) among patients with diabetes compared with (2.0%) in patients without diabetes (P < 0.001).

Overall 4,170 patients had incident ICH during a mean (SD) follow-up of 7.3 (1.8) years and 10,730,915 person-years, reflecting an ICH crude incidence rate of 38.8 per 100,000 person-years. The univariate and multivariate association of each variable, included in Table 1, with ICH is shown in Supplementary Table 1. The strongest risk factors for ICH were prior ICH, prior stroke/transient ischemic attack (TIA), use of anticoagulation, hypertension, alcohol abuse, male sex, Arab ethnicity, chronic liver disease, and older age. Of note, statin use was associated with decreased risk of ICH (Supplementary Table 1), in line with our previous report (11) looking at statin users only.

Diabetes and ICH

The crude incidence rate of ICH was 78.9 per 100,000 person-years among patients with diabetes and 29.4 per 100,000 person-years among patients without diabetes (crude HR 2.69 [95% CI 2.53–2.87]) (Table 2). Diabetes remained significantly associated with ICH after adjustment for DRS (1.36 [1.27–1.45]). The results were unchanged after exclusion of new cases of diabetes and after censoring at the time of new diabetes diagnosis occurring during follow-up: DRS-adjusted HR 1.37 (95% CI 1.28–1.46) and 1.38 (1.29–1.47), respectively. Similar results were reached with adjustment for propensity score, 1.47 (1.36–1.59), and for the clinically relevant risk factors, 1.51 (1.40–1.62) (Table 3). Predefined subgroup analyses had shown statistically significant interactions with age (Pinteraction <0.001) and hypertension (Pinteraction <0.001) but not with sex: ICH risk associated with diabetes was more pronounced in patients ≤60 years old and in nonhypertensive patients, although the CI overlapped in hypertensive and nonhypertensive patients (Supplementary Fig. 1).

The risk of ICH was directly associated with diabetes duration. Compared with the group without diabetes, the DRS-adjusted HR was 1.23 (95% CI 1.12–1.35) and 1.44 (1.34–1.56) for diabetes duration ≤5 years and >5 years, respectively. The corresponding HRs with adjustment for propensity score were 1.27 (1.15–1.41) and 1.65 (1.50–1.80), respectively (Table 3).

The risk of ICH associated with diabetes was clearly related to HbA1c in a J-shaped manner. The HRs associated with each HbA1c decile in comparison with the group without diabetes are shown in Table 2. The lowest risk was observed in the fourth HbA1c decile, 6.5–6.7% (48–50 mmol/mol), and was comparable with the risk in the group without diabetes, with DRS-adjusted HR 0.98 (95% CI 0.82–1.18).

HbA1c and ICH Among Patients With Diabetes

HbA1c was significantly associated with ICH among patients with diabetes: adjusted HR 1.14 (95% CI 1.10–1.17) for each 1% increase in HbA1c (Table 4). However, HbA1c appears to have a nonlinear J-shaped relationship with ICH (Pnonlinearity = 0.0186), with the lowest risk observed at HbA1c of 6.5% (48 mmol/mol). A smoothed plot of adjusted HRs (relative to HbA1c reference level of 6.5% [48 mmol/mol]) is presented in Supplementary Fig. 2.

To avoid assuming linearity, we studied HbA1c as a categorical variable according to deciles of the distribution, with the fourth decile, 6.5–6.7% (48–50 mmol/mol), serving as the reference group. Compared with the reference group, the HR for ICH was 1.27 (95% CI 1.01–1.59) and 2.19 (1.75–2.73) in the lowest HbA1c decile, ≤6.0% (<42 mmol/mol), and highest HbA1c decile, >9.30% (>78 mmol/mol), respectively (Table 4). The effect of HbA1c was modified by age (Pinteraction = 0.025). No statistically significant interactions were observed with sex, hypertension, or diabetes duration (Supplementary Table 2). We reached similar results when, instead of using the most recent HbA1c result, we used the median HbA1c, of all tests performed in the 2 years prior to study entry, to assess the association with ICH.

The magnitude of the association between HbA1c and ICH type was more pronounced for deep ICH than for lobar ICH. For each 1% increase in HbA1c, adjusted HR was 1.17 (95% CI 1.13–1.22) for deep ICH and 1.09 (1.02–1.17) for lobar ICH (Pinteraction = 0.066).

To the best of our knowledge, this is the largest nationwide study to date looking at the association between diabetes and ICH encompassing HbA1c levels. We found that diabetes and diabetes duration are associated with increased risk for ICH, also after adjustment for potential confounders, and that low, as well as high, HbA1c levels are associated with ICH risk.

It is puzzling that despite the fact that diabetes promotes overall cardiovascular morbidity, as yet no consensus has emerged concerning the role of diabetes in the occurrence of ICH, as results across studies showed inconsistent associations. Some studies found no significant relationship between diabetes and ICH (18,19), while others found significant positive association (5,20). Another Mendelian randomization study showed no significant association of diabetes with ICH risk (21). Surprisingly, other studies found that that ICH occurs significantly less in patients with diabetes (22,23). For explaining the latter, authors speculated that diabetes-induced thickening of basement membrane and the proliferation of the endothelium render the cerebral vessels less prone to rupture. Another explanation was that in diabetes aggregation, coagulation and plasminogen activator inhibitor levels are increased, while fibrinolytic activity is decreased (23).

A meta-analysis including 102 prospective studies with 698,782 cases showed that diabetes is a risk factor with relative risk of 1.56 for ICH (5). Of note, most of the individual studies included in this meta-analysis focused on coronary heart disease and major stroke subtypes, encompassing only a small number of ICH cases for each, and information for duration of diabetes was lacking; therefore, generalization of association was hard.

A more recent systemic meta-analysis (4), conducted in view of the inconsistencies among previous individual studies and systemic reviews, found association between diabetes and ICH occurrence in 19 case-control studies, whereas association was not identified in three population-based cohort studies (4).

The other main finding in our study is the J-shaped relationship between HbA1c and ICH risk, with the lowest risk observed at HbA1c of 6.5% (48 mmol/mol). The risk of ICH among patients with HbA1c of 6.5–6.7% (48–50 mmol/mol) was comparable with the risk in patients without diabetes, suggesting that albeit having diabetes, patients with good, but not extreme, diabetes control do not appear to have excess risk of ICH compared with patients without diabetes. Previous studies showed similar relations between fasting blood glucose and ICH (8,9), reinforcing the validity of our results. None, however, have looked at the association between HbA1c, reflecting glycemic control much more reliably, and ICH.

Indeed, previous studies have demonstrated J-shape or U-shape relationships between fasting glucose levels and cardiovascular diseases, all-cause mortality (24), and ICH (8). It has been suggested that hypoglycemia as well as rapid changes in plasma glucose may lead to elevation of counterregulatory hormones such as epinephrine or neuroepinephrine, promoting vasoconstrictions and hemodynamic alternations that may end in ICH (25). Another suggested explanation is that low glucose levels may reflect disease burden or frailty (8).

To date, the exact mechanisms underlying the association between diabetes, HbA1c, and ICH remain unknown. Hopefully, in the future more will be learned about the pathogenesis of this potential association.

A notable strength of this study is being a population-based study with a large number of ICH cases (n = 4,170) encompassing both recent HbA1c result and median HbA1c. As opposed to some previous studies, we also included duration of diabetes in our analysis. Controlling for DRS as a single variable increases the efficiency of the analysis compared with conventional multivariate analyses. Yet, this study has some limitations; firstly, despite the multivariate model adjustment that accounted for a large number of potential confounders, some residual confounding might still exist owing to unmeasured variables. Hence, because our retrospective cohort study is observational in nature, this study could not prove a cause-effect relationship; such evidence should rely on randomized controlled clinical studies. Secondly, we rely solely on an administrative computerized database that was not specifically designed for the current study. In addition, we had no information about prevalence of type 1 or 2 diabetes. We also looked at ICH location only among a subgroup of patients with diabetes (n = 1,614). Furthermore, HbA1c results were available only in patients with diabetes; hence, we cannot argue for a relationship between HbA1c and ICH in patients without diabetes. However, HbA1c in patients with diabetes, which are by definition <6.5% (48 mmol/mol), is not expected to be associated with increased risk of ICH. This seems to be unlikely because HbA1c <6.5% (48 mmol/mol) in diabetes may be a marker of hypoglycemia that probably mediates increased risk for ICH, whereas in patients without diabetes, lower HbA1c is not a marker of hypoglycemia. In addition, we have no data on cerebral amyloid angiopathy status among study participants.

In summary, our study suggests that diabetes is associated with increased risk of ICH that is directly associated with diabetes duration. ICH and HbA1c appear to have a J-shaped relationship, suggesting that both poor control as well as extreme intensive diabetes control might be associated with increased risk. These findings and their clinical implications need to be assessed with future randomized clinical trials.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. W.S. designed the study, researched and analyzed data, and wrote the manuscript. O.B.-G. analyzed data and reviewed and edited the manuscript. N.G. researched data and reviewed and edited the manuscript. J.M. contributed to discussion and reviewed and edited the manuscript. J.N. reviewed and edited the manuscript. G.R. reviewed and edited the manuscript. E.A. designed the study and wrote the manuscript. E.A. is the guarantor of this work and, as such, had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the Annual Meeting of the Israel Neurological Association, Rishon Le Zion, Israel, 11–12 December 2018.

1.
Qureshi
AI
,
Tuhrim
S
,
Broderick
JP
,
Batjer
HH
,
Hondo
H
,
Hanley
DF
.
Spontaneous intracerebral hemorrhage
.
N Engl J Med
2001
;
344
:
1450
1460
[PubMed]
2.
IDF Diabetes Atlas
. 7th ed.
Brussels, Belgium
,
International Diabetes Federation
,
2015
3.
Peters
SA
,
Huxley
RR
,
Woodward
M
.
Diabetes as a risk factor for stroke in women compared with men: a systematic review and meta-analysis of 64 cohorts, including 775,385 individuals and 12,539 strokes
.
Lancet
2014
;
383
:
1973
1980
[PubMed]
4.
Boulanger
M
,
Poon
MT
,
Wild
SH
,
Al-Shahi Salman
R
.
Association between diabetes mellitus and the occurrence and outcome of intracerebral hemorrhage
.
Neurology
2016
;
87
:
870
878
[PubMed]
5.
Sarwar
N
,
Gao
P
,
Seshasai
SR
, et al.;
Emerging Risk Factors Collaboration
.
Diabetes mellitus, fasting blood glucose concentration, and risk of vascular disease: a collaborative meta-analysis of 102 prospective studies
[published correction appears in Lancet 2010;376:958].
Lancet
2010
;
375
:
2215
2222
[PubMed]
6.
Bragg
F
,
Li
L
,
Bennett
D
, et al.;
China Kadoorie Biobank (CKB) Collaborative Group
.
Association of random plasma glucose levels with the risk for cardiovascular disease among Chinese adults without known diabetes
.
JAMA Cardiol
2016
;
1
:
813
823
[PubMed]
7.
O’Donnell
MJ
,
Chin
SL
,
Rangarajan
S
, et al.;
INTERSTROKE Investigators
.
Global and regional effects of potentially modifiable risk factors associated with acute stroke in 32 countries (INTERSTROKE): a case-control study
.
Lancet
2016
;
388
:
761
775
8.
Jin
C
,
Li
G
,
Rexrode
KM
, et al
.
Prospective study of fasting blood glucose and intracerebral hemorrhagic risk
.
Stroke
2018
;
49
:
27
33
[PubMed]
9.
Park
C
,
Guallar
E
,
Linton
JA
, et al
.
Fasting glucose level and the risk of incident atherosclerotic cardiovascular diseases
.
Diabetes Care
2013
;
36
:
1988
1993
[PubMed]
10.
American Diabetes Association
.
6. Glycemic targets:
Standards of Medical Care in Diabetes—2018.
Diabetes Care
2018
;
41
(
Suppl. 1
):
S55
S64
[PubMed]
11.
Saliba
W
,
Rennert
HS
,
Barnett-Griness
O
, et al
.
Association of statin use with spontaneous intracerebral hemorrhage: a cohort study
.
Neurology
2018
;
91
:
e400
e409
12.
Preis
M
,
Hirsch
J
,
Kotler
A
, et al
.
Factor XI deficiency is associated with lower risk for cardiovascular and venous thromboembolism events
.
Blood
2017
;
129
:
1210
1215
[PubMed]
13.
Andrade
SE
,
Harrold
LR
,
Tjia
J
, et al
.
A systematic review of validated methods for identifying cerebrovascular accident or transient ischemic attack using administrative data
.
Pharmacoepidemiol Drug Saf
2012
;
21
(
Suppl. 1
):
100
128
[PubMed]
14.
Jaffe
DH
,
Shmueli
A
,
Ben-Yehuda
A
, et al
.
Community healthcare in Israel: quality indicators 2007-2009
.
Isr J Health Policy Res
2012
;
1
:
3
[PubMed]
15.
Heinzl
H
,
Kaider
A
.
Gaining more flexibility in Cox proportional hazards regression models with cubic spline functions
.
Comput Methods Programs Biomed
1997
;
54
:
201
208
[PubMed]
16.
Arbogast
PG
,
Ray
WA
.
Performance of disease risk scores, propensity scores, and traditional multivariable outcome regression in the presence of multiple confounders
.
Am J Epidemiol
2011
;
174
:
613
620
[PubMed]
17.
Arbogast
PG
,
Kaltenbach
L
,
Ding
H
,
Ray
WA
.
Adjustment for multiple cardiovascular risk factors using a summary risk score
.
Epidemiology
2008
;
19
:
30
37
[PubMed]
18.
Hesami
O
,
Kasmaei
HD
,
Matini
F
,
Assarzadegan
F
,
Mansouri
B
,
Jabbehdari
S
.
Relationship between intracerebral hemorrhage and diabetes mellitus: a case-control study
.
J Clin Diagn Res
2015
;
9
:
OC08
OC10
[PubMed]
19.
Burchfiel
CM
,
Curb
JD
,
Rodriguez
BL
,
Abbott
RD
,
Chiu
D
,
Yano
K
.
Glucose intolerance and 22-year stroke incidence. The Honolulu Heart Program
.
Stroke
1994
;
25
:
951
957
20.
Herzig
R
,
Vlachová
I
,
Mares
J
, et al
.
Occurrence of diabetes mellitus in spontaneous intracerebral hemorrhage
.
Acta Diabetol
2007
;
44
:
201
207
[PubMed]
21.
Liu
J
,
Rutten-Jacobs
L
,
Liu
M
,
Markus
HS
,
Traylor
M
. Causal impact of type 2 diabetes mellitus on cerebral small vessel disease: a Mendelian randomization analysis.
Stroke
2018
;
49
:
1325
1331
[PubMed]
22.
Jørgensen
H
,
Nakayama
H
,
Raaschou
HO
,
Olsen
TS
.
Stroke in patients with diabetes. The Copenhagen Stroke Study
.
Stroke
1994
;
25
:
1977
1984
[PubMed]
23.
Karapanayiotides
T
,
Piechowski-Jozwiak
B
,
van Melle
G
,
Bogousslavsky
J
,
Devuyst
G
.
Stroke patterns, etiology, and prognosis in patients with diabetes mellitus
.
Neurology
2004
;
62
:
1558
1562
[PubMed]
24.
Wei
M
,
Gibbons
LW
,
Mitchell
TL
,
Kampert
JB
,
Stern
MP
,
Blair
SN
.
Low fasting plasma glucose level as a predictor of cardiovascular disease and all-cause mortality
.
Circulation
2000
;
101
:
2047
2052
[PubMed]
25.
Tanne
D
,
Koren-Morag
N
,
Goldbourt
U
.
Fasting plasma glucose and risk of incident ischemic stroke or transient ischemic attacks: a prospective cohort study
.
Stroke
2004
;
35
:
2351
2355
[PubMed]
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.

Supplementary data