Systematic review Risk factors for ischemic stroke and its subtypes in Chinese vs. Caucasians: Systematic review and meta-analysis Chung-Fen Tsai1,2,3, Niall Anderson,4 Brenda Thomas3, and Cathie L. M. Sudlow3,5* Background Chinese populations are reported to have a different distribution of ischemic stroke subtypes compared with Caucasians. Aims To understand this better, we aimed to evaluate the differences in prevalence of risk factors in ischemic stroke and their distributions among ischemic stroke subtypes in Chinese vs. Caucasians. Summary or review We systematically sought studies conducted since 1990 with data on frequency of risk factors among ischemic stroke subtypes in Chinese or Caucasians. For each risk factor, we calculated study-specific and random effects pooled estimates in Chinese and Caucasians separately for: prevalence among ischemic stroke; odds ratios, comparing prevalence for each ischemic stroke subtype vs. all others. We included seven studies among 16 199 Chinese, and eleven among 16 189 Caucasian ischemic stroke patients. Risk factors studied were hypertension, diabetes, atrial fibrillation, ischemic heart disease, hypercholesterolemia, smoking and alcohol. Chinese ischemic stroke patients had younger onset of stroke than Caucasians, similar prevalence of hypertension, diabetes, smoking and alcohol, and significantly lower prevalence of atrial fibrillation, ischemic heart disease and hypercholesterolemia. Risk factor associations with ischemic stroke subtypes were mostly similar among Chinese and Caucasian ischemic stroke patients. Compared with all other ischemic subtypes, diabetes was more common in large artery stroke, atrial fibrillation and ischemic heart disease in cardioembolic stroke, and hypertension and diabetes in lacunar stroke. Conclusion Our study showed a lower prevalence of atrial fibrillation, ischemic heart disease and hypercholesterolemia in Chinese, and mostly similar risk factor associations in Chinese and Caucasian ischemic stroke patients. Further analyses of individual patient data to allow adjustment for confounders are needed to confirm and extend these findings. Key words: Chinese, ischemic stroke, risk factor, subtype, Whites Correspondence: Cathie Sudlow*, Division of Clinical Neurosciences, Centre for Clinical Brain Sciences, University of Edinburgh, Chancellor’s Building, 49 Little France Crescent, Edinburgh EH16 4SB, UK. E-mail:
[email protected] 1 Department of Neurology, Cardinal Tien Hospital, Taiwan 2 School of Medicine, Fu Jen Catholic University, Taiwan 3 Division of Clinical Neurosicences, Centre for Clinical Brain Sciences, University of Edinburgh, Edinburgh, UK 4 Centre for Population Health Sciences, University of Edinburgh, Edinburgh, UK 5 Institute of Genetics and Molecular Medicine (C.L.M.S), University of Edinburgh, Edinburgh, UK Received: 22 September 2014; Accepted: 6 January 2015; Published online 23 April 2015 Conflict of interest: The authors declare no conflict of interest. Funding: C-F. Tsai: scholarships from Cardinal Tien hospital and Ministry of Education, Taiwan. C. Sudlow: Scottish Funding Council. The funding sources had no role in study design, literature search, data acquisition, analysis and interpretation of data or manuscript writing. DOI: 10.1111/ijs.12508 © 2015 World Stroke Organization
Introduction Compared with Caucasians (white populations of European descent), Chinese populations have a higher overall stroke incidence and a higher proportion of intracerebral haemorrhage (1). There is also some evidence to suggest a different distribution of ischemic stroke (IS) subtypes, with a higher proportion of small vessel disease (lacunar) stroke and a lower proportion of cardioembolic stroke in Chinese vs. predominantly white populations of European descent (1,2). The different distribution of IS subtypes in Chinese vs. predominantly white populations may relate to differences in the prevalence of risk factors and their associations with IS subtypes. This may in turn have implications for understanding the causes of ischemic stroke and predicting the effects of prevention strategies. To test the hypothesis that risk factor profiles in IS and its subtypes vary in different populations, we conducted a systematic review and meta-analysis to evaluate the overall prevalence of risk factors and their associations among IS subtypes in Chinese compared with Caucasian IS patients.
Methods Search strategy and selection criteria We used comprehensive strategies in Medline and EMBASE to identify studies published in any language, comparing the frequency of risk factors among IS subtypes in Chinese patients. For comparison, we sought similar studies in predominantly Caucasian patients (white patients of European descent) using searches aimed at identifying existing systematic reviews and metaanalyses of risk factors for IS subtypes (Supplementary Appendices S1 and S2). We also conducted citation searches of relevant reviews and perused reference lists of included primary articles and reviews (3–5). We included studies with prospective recruitment, standard definition of stroke, clear IS subtype classification, and data collection from 1990 [because brain imaging with computed tomography (CT) or magnetic resonance (MR)] were not used widely before this) to April 2013 (6,7). We included both communityand hospital-based studies of first-ever and recurrent strokes. For reliable stroke diagnosis and classification, we required CT/MR brain imaging (or autopsy) in >70% of cases. IS had to be classified into etiological subtypes with the Trial of Org 10172 in Acute Ischemic Stroke (TOAST) classification or similar large artery atherosclerosis (LAA), cardioembolism (CE), intracranial small vessel disease (SVD), other determined or undetermined etiology), or into anatomical subtypes with the Oxfordshire Community Stroke Project (OCSP) classification (lacunar infarct, total anterior circulation infarct, partial anterior circulation infarct, posterior circulation infarct) (8,9). We excluded studies with Vol 10, June 2015, 485–493
485
Systematic review
C-F. Tsai et al.
incomplete risk factor information for IS subtypes, or highly selected patients. We contacted study authors directly for essential information not available in their original publication(s).
(pooled proportion of male 61% in Chinese and 52% in Caucasians, P < 0·001). Brain imaging rates were high where reported (98 to 100%) (Tables 1 and 2).
Data extraction From each included study, we extracted available information on: geographical setting; sources of case ascertainment; inclusion of only first-ever or first-ever and recurrent strokes; study period; stroke definition; subtype classification method; proportion of cases with CT or MR brain imaging; mean age and gender distribution of stroke cases; risk factor definitions; number of cases overall and with each risk factor for all IS and for separate IS subtypes. One author selected studies and extracted data, resolving uncertainties and disagreement through discussion with a second.
Risk factors prevalence for overall IS patients Risk factors studied were hypertension, diabetes, atrial fibrillation (AF), ischemic heart disease (IHD), hypercholesterolemia, smoking and alcohol intake, and their definitions as reported in the original studies are summarized in Supplementary Tables S1 and S2. Chinese vs. predominantly Caucasian IS patients had similar pooled prevalence of hypertension (pooled proportions both 59%), diabetes (25% vs. 21%), smoking (38% vs. 30%) and alcohol intake (21% vs. 15%), and lower prevalence of AF (11% vs. 27%), hypercholesterolemia (9% vs. 30%) and IHD (11% vs. 20%) (Fig. 1). However, heterogeneity within Chinese and Caucasian studies was substantial and there were some differences in risk factor definitions between Chinese and predominantly Caucasians.
Statistical analysis We calculated pooled proportions of each risk factor and male in all IS patients in Chinese and predominantly Caucasians separately. For each risk factor with data available from more than one study, we calculated study-specific and random effects pooled odds ratios (ORs) with 95% confidence intervals (CIs) for each IS subtype vs. all others in Chinese and in Caucasians. We categorized OCSP subtypes into lacunar and non-lacunar groups. We assessed within-ethnic group heterogeneity with both I2 and Cochrane Q χ2 statistics (10). To assess whether pooled ORs differed significantly between Chinese and Caucasian groups, we performed heterogeneity tests, using the within-group pooled OR estimates and their standard errors, with χ2 statistics to test for statistical significance (11). P < 0·10 was taken to indicate significant between-group heterogeneity in our study, since heterogeneity tests are typically applied conservatively for meta-analysis, particularly where the number of studies is smaller, as this compensates for the low power of the heterogeneity chi-square test in this situation (12). We performed analyses with StatsDirect software (http://www.statsdirect.com).
Results Characteristics of included studies From a total of around 9000 publications identified by electronic searches, we eventually included seven studies (three using TOAST, three OCSP, one both) in a total of 16 199 Chinese IS patients (13–20), and 11 studies (eight TOAST, three OCSP) in a total of 16 189 predominantly Caucasian IS patients (4,21–30). Supplementary Figs S1a and S1b show study selection and reasons for exclusions. All seven Chinese studies were hospital-based, while of the 11 Caucasian studies, 5 were community-based and 6 were hospitalbased. Patients in the Chinese studies had younger onset of stroke compared with Caucasians (Chinese: mean age 66, range 61 to 69; Caucasians: mean age 71, range 66 to 74). From the available data of gender for overall ischemic stroke patients (15 791 in Chinese and 12 433 in Caucasians), there was a predominance of males in Chinese as compared with predominantly Caucasian patients
486
Vol 10, June 2015, 485–493
Risk factor associations with IS subtypes Chinese stroke patients had younger onset of stroke compared with Caucasian patients across the major IS subtypes. As for gender, where reported, both Chinese and Caucasian studies showed a predominance of males in LAA and lacunar stroke, but a predominance of females in CE stroke (Supplementary Table S3). Statistically, there was a different distribution of ischemic stroke subtypes between Chinese and Caucasian patients (Supplementary Table S4). Despite these age and gender differences, comparisons of risk factors among IS subtypes showed generally similar directions of results between Chinese and Caucasian IS patients, although the size of the differences sometimes varied. Among studies using the TOAST classification, comparing LAA vs. other IS subtypes, results for Chinese and Caucasians were broadly similar, with no significant between-group heterogeneity detected for any risk factor (Fig. 2a). Diabetes was slightly but significantly more common (ORs 1·34 and 1·42 for Chinese and Caucasians respectively) and AF substantially and significantly less common (ORs 0·27 and 0·19). Hypertension, IHD and smoking were consistently slightly more common in LAA (ORs 1·12 to 1·31), although results were generally not independently significant within ethnic groups. Alcohol consumption appeared commoner in LAA vs. other subtypes in Chinese but not in Caucasians; however, there was no heterogeneity between the ethnic groups (Fig. 2a). Comparing CE vs. other stroke subtypes, there were significant, strongly positive associations with AF (ORs 71·36 and 36·81 in Chinese and Caucasians respectively) and IHD (ORs 3·62 and 1·31). Both associations were significantly more marked in Chinese (Chinese-Caucasians heterogeneity for AF: P < 0·001, IHD: P = 0·021; Fig. 2b). Hypertension, diabetes, smoking and alcohol were all less common in CE vs. other subtypes in both populations, but results were independently significant for hypertension and diabetes in Chinese, and for smoking and alcohol in Caucasians (between-group heterogeneity for hypertension: P = 0·166, diabetes: P = 0·079, smoking: P = 0·011, alcohol: P = 0·028; Fig. 2b). © 2015 World Stroke Organization
Region
© 2015 World Stroke Organization
China, Yinchuan
Yang P (20)
2009-2010
2007-2008
2002-2005
1997-1998
2007-2008
2002-2003
1999
1995
Study period (year)
Hospital-based, consecutive admission Hospital-based, consecutive admission
Hospital-based, consecutive admission
Hospital-based, consecutive admission
Hospital-based, consecutive admission
Hospital-based, consecutive admission Hospital-based, consecutive admission
Hospital-based, consecutive admission
Patient recruitment
First-ever and recurrent
First-ever and recurrent
First-ever and recurrent
First-ever and recurrent
First-ever and recurrent
First-ever stroke
First-ever and recurrent
First-ever and recurrent
Stroke inclusion
61
67
65
68
66
67
69
65
Mean age of IS (years)
69%
62%
54%
58%
62%
66%
NR
57%
Gender (male%)
100%
100%
100%
NR†
100%
98%
100%
100%
CT/MR (%)
932
11657
1314
699
11560
610
408
676
IS patients (n)
Hypertension, diabetes, AF, hyperlipidemia, smoking, alcohol, age, gender
Hypertension, diabetes, AF, IHD, smoking, previous stroke, age, gender Hypertension, diabetes, AF, hyperlipidemia, smoking, alcohol, previous stroke, age, gender Diabetes
Hypertension, diabetes, AF, IHD, CS, hypercholesterolemia, hypertriglyceridemia, smoking, alcohol, previous stroke, age, gender Hypertension, diabetes, AF, IHD, CS, hyperlipidemia, smoking Hypertension, diabetes, AF, IHD, hyperlipidemia, smoking, alcohol, age, gender Hypertension, diabetes, AF, IHD, dyslipidemia, smoking, alcohol, previous stroke, age, gender
Risk factors reported
*NINDS classification. † More than 90% of acute stroke patients had brain CT in this institute from their other publications. Studies ordered according to study period in TOAST and OCSP classifications respectively. AF, atrial fibrillation; CS, carotid stenosis; CT, computed tomography; HD, heart disease; IHD, ischemic heart disease; IS, ischemic stroke; MR, magnetic resonance; n, number.
China, multicentre
China, Hong Kong
OCSP classification Li H (17)
Jia Q (19)
China, multicentre
Wang Y (16)
China, Chengdu
China, Nanjing
Liu X (15)
Li W (18)
Taiwan, Kaohsiung
Lin YT (14)*
TOAST classification Yip PK (13) Taiwan, Taipei
Study (first author)
Table 1 Clinical characteristics of included risk factors studies among ischemic stroke subtypes in Chinese populations
C-F. Tsai et al.
Systematic review
Vol 10, June 2015, 485–493
487
488
Vol 10, June 2015, 485–493
Greece, Athens
Germany, Erlangen Germany, multicentre
Italy, Perugia
United Kingdom, Oxford United Kingdom, London France, Dijon
Vemmos KN (22)
Kolominsky-Rabas PL (23) Grau AJ (24)
Silvestrelli G (25)
Schulz UGR (4)
UK, London
Hajat C† (30) 1995-1998
1993-1994
1991-1992
2005-2006
1999-2005
2002
1998-2002
1998-1999
1999
1992-1997
1977-1994
Study period (year)
Community-based with multiple sources
Hospital-based, consecutive admission Hospital-based, admission
Community-based with multiple sources Community-based with multiple sources Community-based with multiple sources
Hospital-based, consecutive admission
Community-based with multiple sources Hospital-based, admission (multicentre)
Hospital-based, consecutive admission
Hospital-based, consecutive admission
Patient recruitment
First-ever
First-ever
First-ever
First-ever stroke
First-ever and recurrent First-ever
First-ever
First-ever and recurrent
First-ever
First-ever
First-ever
Stroke inclusion
72
71
73
74
71
NR
73
66
73
70
66
Mean age of IS (years)
48%
53%
55%
45%
49%
51%
52%
58%
44%
59%
57%
Gender (male%)
NR‡
NR‡
100%
100%
NR‡
98%
100%
100%
100%
100%
100%
CT/MR (%)
862
2472
166
332
1169
102
1759
5017
531
885
2894
IS patients (n)
Hypertension, diabetes, AF, IHD, smoking, alcohol, TIA, age, gender Hypertension, diabetes, AF, IHD, smoking, alcohol, TIA
AF, CS
Hypertension, diabetes, HD, hyperlipidemia, smoking, PAD, TIA, age, gender Hypertension, diabetes, AF, IHD, hypercholesterolemia, smoking, TIA, age, gender Hypertension, diabetes, HD, smoking Hypertension, diabetes, arrhythmia, IHD, hypercholesterolemia, smoking, alcohol, PAD, TIA, age, gender Hypertension, diabetes, EHD, hyperlipidemia, hypercholesterolemia, hypertriglyceridemia, smoking, alcohol, obesity, TIA, age, gender, Hypertension, diabetes, smoking, TIA, age, gender Hypertension, diabetes, smoking, TIA, age, gender Hypertension, diabetes, hypercholesterolemia, smoking, TIA, age, gender
Risk factors reported
*TOAST classification. † OCSP classification. ‡ Around 80-95% acute stroke patients had brain CT/MR in these institutes from their other published papers. Studies ordered according to study period in TOAST and OCSP classifications respectively. AF, atrial fibrillation; CS, carotid stenosis; CT, computed tomography; EHD, embolic heart disease; HD, heart disease; HD, heart disease; IHD, ischemic heart disease; IS, ischemic stroke; MR, magnetic resonance; n, number; PAD, peripheral artery disease; TIA, transient ischemic attack.
Europe, multinational
Carlo AD (29)
OCSP classification Lindgren A (28)
Bejot Y (27)
Sweden, Lund
Spain, Barcelona
TOAST classification Marti-Vilalta JL (21)
Hajat C* (26),
Region
Study (first author)
Table 2 Clinical characteristics of included risk factors studies among ischemic stroke subtypes in Caucasians
Systematic review C-F. Tsai et al.
© 2015 World Stroke Organization
Systematic review
C-F. Tsai et al. Risk factor (studies) (patients)
Pooled proportion
(95% CI)
Within-group H
HTN-C (7) (16199)
0.59 (0.48, 0.69)
I2=99%
HTN-W (10) (16021)
0.59 (0.55, 0.64)
I2=97%
DM-C (7) (16199)
0.25 (0.20, 0.31)
I2=97%
DM-W (10) (16021)
0.21 (0.18, 0.24)
I2=94%
AF-C (7) (16199)
0.11 (0.09, 0.13)
I2=92%
AF-W (5) (6142)
0.27 (0.20, 0.34)
I2=97%
IHD-C (6) (15627)
0.11 (0.06, 0.18)
I =99%
IHD-W (4) (9236)
0.20 (0.13, 0.28)
I =99%
HC-C (2) (1990)
0.09 (0.02, 0.20)
I =98%
HC-W (4) (7991)
0.30 (0.18, 0.43)
I2=99%
Smoking-C (7) (16199)
0.38 (0.31, 0.45)
I=98%
Smoking-W (10) (16021)
0.30 (0.24, 0.37)
I2=99%
Alcohol-C (5) (15092)
0.21 (0.10, 0.33)
I =99%
Alcohol-W (4) (10108)
0.15 (0.07, 0.27)
I2=99%
0.0
0.2
0.4
0.6
2
Between-group H (p-value) P>0.999
P=0.211
P
