In contrast to Western populations, stroke is the leading. Is There a Stroke Belt in China and Why?

Transcription

1 Is There a Stroke Belt in China and Why? Gelin Xu, MD, PhD; Minmin Ma, MD, PhD; Xinfeng Liu, MD, PhD; Graeme J. Hankey, MD, FRACP, FRCP Background and Purpose In 1960s, a stroke belt with high stroke mortality was discovered in the southeast United States. In China, where stroke is the leading cause of death, we aimed to determine whether a focal region of high stroke incidence (stroke belt) exits and, if so, the possible causal and modifiable factors. Methods We systematically reviewed all studies of stroke incidence in China between 1980 and 2010, and included those which met our criteria for a high-quality study. Criteria for a provincial region of high stroke incidence were ranking in the top one third of all provinces for stroke incidence and ranking of more than one third of prefectural regions within the province in the top two sevenths of all prefectural regions for stroke incidence. We also reviewed regional distribution of major vascular risk factors, socioeconomic status, and demographic profiles in China. Results Nine eligible studies provided data on the incidence of stroke in 32 of 34 provincial regions of China (with Hong Kong and Macao as exceptions) and 52% of the 347 prefectural regions. Nine provincial regions (Heilongjiang, Tibet, Jilin, Liaoning, Xinjiang, Hebei, Inner Mongolia, Beijing, and Ningxia) met our criteria for a region of high stroke incidence and constitute a stroke belt in north and west China. The incidence of stroke in the stroke belt was per population compared with in regions outside the belt (rate ratio, 2.16; 95% confidence interval, ). The mean population prevalence of hypertension and overweight (body mass index, >25) was greater in the stroke belt than that in other regions (15.3% versus 10.3%, P<0.001; 21.1% versus 12.3%, P=0.013, respectively). The prevalence of hypertension and overweight also correlated significantly with regional stroke incidence (R=0.642, P<0.001; R=0.438, P=0.014, respectively, by Spearman rank correlation). Conclusions A stroke belt of high stroke incidence exists in 9 provincial regions of north and west China. The stroke belt may be caused, at least in part, by a higher population prevalence of hypertension and excess body weight. Lowering blood pressure and body weight in the stroke belt may reduce the geographic disparity in stroke risk and incidence in China. (Stroke. 2013;44: ) Key words: China epidemiology incidence risk factors stroke In contrast to Western, stroke is the leading cause of death in China, where it accounted for 22.5% of deaths (about 2 million) in Stroke mortality in China is still increasing and is projected to double during the next 2 decades, from 2 to 4 million deaths. 2 For survivors of stroke in China, the socioeconomic burden is substantial. 3 Health policies to circumvent the epidemic of stroke in China require an accurate understanding of its nature and its causes. 4 In the 1960s, a stroke belt or region was identified in the southeastern United States, where the rates of stroke mortality among its residents were 50% higher than the remainder of the United States. 5 Subsequent studies suggested that geographic and racial variations in stroke mortality relate to differences in incidence, not case fatality, and that geographic disparities in stroke risk factors, such as hypertension and diabetes mellitus, may underlie the observed differences in stroke mortality. 6 If these findings are valid, interventions to reduce geographic disparities in stroke factors may reduce geographic disparities in stroke incidence (and mortality). We hypothesized that a similar phenomenon of geographic variation in stroke incidence and mortality could exist in China, which has been observed and discussed in individual studies, 7 but has not been study systemically. As mortality is determined by both incidence and case fatality, and as a major priority is the prevention of incident stroke (to complement effective treatments that reduce case fatality), 8 we undertook a systematic review of all studies of the incidence of stroke in China. The aim was to determine whether stroke incidence rates vary geographically throughout the provincial and prefectural regions of China, Received February 20, 2013; final revision received April 2, 2013; accepted April 4, From the Department of Neurology, Jinling Hospital, Nanjing University School of Medicine, Nanjing, Jiangsu Province, China (G.X., M.M., X.L.); School of Medicine and Pharmacology, The University of Western Australia, Western Australia, Australia (G.J.H.); and Stroke Unit, Department of Neurology, Royal Perth Hospital, Perth, Western Australia, Australia (G.J.H.). The online-only Data Supplement is available with this article at /DC1. Correspondence to Graeme J. Hankey, MD, FRACP, FRCP, Stroke Unit, Department of Neurology, Royal Perth Hospital, 197 Wellington St, Perth, Western Australia 6001, Australia. graeme.hankey@uwa.edu.au American Heart Association, Inc. Stroke is available at DOI: /STROKEAHA

2 1776 Stroke July 2013 whether a region with a consistently high stroke incidence (stroke belt) exists, and, if so, whether there may be major contributing factors to the stroke belt that could be modifiable. Methods Search Strategy We searched the international databases PubMed, Scopus, ISI Web of Science, and Science Direct, and the Chinese databases China Hospital Knowledge Database and China National Knowledge Infrastructure, from January 1980 to December 2010 with the words stroke OR cerebrovascular diseases AND incidence OR prevalence AND China OR Chinese. We further checked other literature sources, such as Xinhua Booklist, Amazon online, Google Scholar, and the website of the Ministry of Health of China for extra information concerning the included studies. Some data were obtained or confirmed by contacting the authors of the publications. Selection Criteria Eligibility criteria were as follows: complete population-based case ascertainment in China based on multiple overlapping sources of information (hospitals, outpatient clinics, and death certificates), standardized World Health Organization definition of stroke, 9 incident stroke cases reported, data collection over whole years, stroke incidence estimates were age-adjusted or the age constitution ratios of the sample population were provided, prospective study design. Pathological subtypes of stroke included ischemic stroke, intracerebral hemorrhage, and subarachnoid hemorrhage. Transient ischemic attack was not included. Only studies published in English or Chinese were evaluated. Classification of Geographical Regions There are 5 levels of regions in China: province, prefecture, county, township, and village. Since the 17th century, provincial and prefectural boundaries in China have remained relatively static. The provinces serve an important cultural role in China, as people tend to identify with their native province. At present, there are 34 provincial, 333 prefectural, 2862 county, or township level regions in China. 10 There are 4 types of provincial regions, which included 23 provinces (include Taiwan), 5 autonomous regions, 4 municipalities, and 2 special administrative regions (Hong Kong and Macao). There are 23 counties directly affiliated to the corresponding provinces. To match the territory and population sizes, some neighboring counties of this kind were merged as presumed prefectural regions. Each of the 4 municipalities (Beijing, Shanghai, Tianjin, and Chongqing), which has a comparable territory of prefectural regions but with a larger population, was treated as a whole when assessing prefectural stoke risk. These arrangements resulted in 347 factual or presumed prefectural regions. Sources of Data for Regional Stroke Incidences If data for the incidence of stroke in a region were available from only 1 study, the single-source was accepted. However, if data for the incidence of stroke in a region were available from more than 1 eligible study, the incidence rates from all sources were combined to produce an average that was weighted according to their sample sizes. If data for the incidence of stroke in a region were not available, the regional incidence was calculated on the basis of available incidence data from the surrounding regions. These calculations used a strategy reported in previous studies for determining the stroke belt in the Unites States. 11 Briefly, the stroke incidence for a particular region without observed incidence was estimated as a weighted average of all observed estimates of stroke incidence in the surrounding regions. The weighting incorporated the size of the population sampled and (the reciprocal of) the distance to the center of each of the surrounding regions. Geographical Classification of Stroke Incidence and Definition of the Stroke Belt Stroke incidence estimates in provincial and prefectural regions were ranked ordinally. Provincial regions with a stroke incidence in the top third and prefectural regions with a stroke incidence in the top two sevenths were arbitrarily classified as regions with a high stroke incidence. The stroke belt was defined as a region containing provinces that met the following 2 criteria for a region of high stroke incidence: (1) the province was ranked in the top one third of provinces for stroke incidence (ie, met the above criteria for high stroke incidence) and (2) more than one third (>33.3%) of the prefectural regions within the province were ranked in the top two sevenths of prefectural regions for stroke incidence (ie, met the above criteria for high stroke incidence). The rationale for using 2 criteria for determining the stroke belt was to exclude provinces with extremely high incidence in only 1 or 2 prefectural regions. Exploring Possible Causes of the Stroke Belt To explore the possible causes of the stroke belt, we sought and analyzed several sources of data. We systematically reviewed Chinese nationwide epidemiological surveys to obtain the prevalence of risk factors for cardiovascular diseases, such as hypertension, 12 diabetes mellitus, 13 smoking, 14 and overweight 15 in each province. The methods of diagnosing and ascertaining cases with these risk factors are described in the original publications We also reviewed 2 Chinese nationwide nutrition surveys 15,16 and the China Yearbook of Statistics 10 to ascertain per capita consumption of classified foods. Population morphology, socioeconomic indices, environmental and geographic parameters were retrieved from China Yearbook of Statistics and China Yearbook of Health Statistics. 10,17 To increase homogeneity, we did not combine the multisource data, but those from the most representative survey were selected. If only 1 survey was available for the distribution of a given factor, the data from this single-source survey were accepted. If multiple surveys were available for the distribution of a given risk factor, the data from the survey with larger sample size and more provincial regions involved were selected. When surveys with similar sample size and number of provincial regions were available for multiple years, those of, or nearest to 2000 were accepted. Statistical Analysis Stroke incidence in the 31 provincial regions of Mainland China and Taiwan was analyzed. Crude stroke incidence rates were reported as the number of new cases of stroke in a whole year per individuals. Age-standardized stroke incidence according to the Segi s world population 18 or US 1960 population was retrieved from the enrolled studies. For studies in which stroke incidence was reported in a restricted age range (eg, aged >35 years in the Taiwan study), 19 we used imputed estimates of stroke incidence for the missing ages from those obtained in a large survey of stroke incidence among all ages. 20 The impact of this transformation was minimal because the stroke incidence in population aged 35 years was extremely low (1.2 per compared with in population aged >35 years). 20 After determining the stroke belt, the overall stroke incidence in the stroke belt was calculated from the corresponding provincial incidence rates by weighting the provincial population size. Distribution of risk factors for stroke, such as hypertension, smoking, diabetes mellitus, and overweight in provincial regions, was obtained from nationwide epidemiological surveys. Using these provincial data, distribution of risk factors in stroke belt and regions outside was calculated by weighting the provincial population size. Provincial stroke incidences and possible influential factors in population were ranked, with their relationship analyzed by Spearman rank correlation. Differences in risk factor distribution between stroke belt and other regions were compared using 1-way ANOVA, with provincial data treated as unit of analysis. A P<0.05 was deemed as statistical significance.

3 Xu et al Stroke Belt in China 1777 Results Results of Literature Searching and Quality of the Included Studies The database search for studies on stroke incidence in China yielded 3776 articles. The titles and abstracts of these were screened and 3652 did not meet the eligibility criteria. After reading the full text of the remaining 124 papers, and tracking any relevant references, an additional 31 publications were identified. Two reviewers (G.X. and M.M.) appraised these 151 potentially relevant publications independently, and 9 studies (12 publications) met the selection criteria for eligibility and analysis. 7,19 29 Table 1 shows the characteristics of these included studies. Of these 9 studies, the 2 Monitoring Trends and Determinants in Cardiovascular Disease (MONICA) studies 26,27 had a large proportion of overlapped sample population. We, therefore, only included the one 26 with larger sample in data analysis. The 9 eligible studies provided data for the incidence of stroke in 32 (94.1%, with Hong and Macao as exceptions) of the 34 provincial regions and 182 (52.4%) of 347 prefectural regions in total. All 9 included studies used the door-to-door survey method in all or part of their sampling. The sample population was initially screened by means of a questionnaire which identified cases of suspected stroke, who were then examined by a specialist team. The diagnosis of stroke was confirmed or rejected by evaluating the results of medical records, physical examinations, and laboratory and imaging tests. Although first-ever stroke cases in Sino-MONICA study 26,27 were diagnosed on the basis of dynamic reports of grassroots health station personnel and hospital records, the diagnosis of stroke was confirmed retrospectively by door-todoor survey in a randomly selected 10% of the study population. Fatal cases were confirmed by the local death registry system and medical records. For stroke diagnostic criteria, 5 of 9 studies used World Health Organization criteria, and the other 4 studies used similar criteria (Table 1). The distribution of the all sampling sites from the enrolled 9 studies is shown in Figure 1. The sampling sites were disproportionally distributed across the country, with more sampling sites in east coast regions and fewer in west regions. There was only 1 sampling site in the southwest region (Tibet; Table I in the online-only Data Supplement). Stroke Incidence in Provincial and Prefectural Regions Table 2 shows the stroke incidence in the 31 provincial regions of Mainland China and Taiwan in rank order. The incidence of stroke was highest in Heilongjiang (466.9 per per year) followed by Tibet (450.4), and lowest in Guangxi (73.1) followed by Zhejiang (75.5). The difference in stroke incidence between the highest and the lowest provincial regions was more than 6 times. Figure 2 is a color-coded map of China showing the incidence of stroke in the 347 prefectural regions with total population. Because most studies provided data for the incidence of stroke in women and men separately, in addition to the overall population, it was possible to generate separate color-coded maps of stroke incidence according to prefectural regions among men (Figure I in the online-only Data Supplement) and women (Figure II in the online-only Data Supplement). The Stroke Belt The 9 provincial regions that met both criteria for the stroke belt included Heilongjiang, Tibet, Jilin, Liaoning, Xinjiang, Hebei, Inner Mongolia, Beijing, and Ningxia (Table 2; Figure 2). Although Henan had a high stroke incidence of 174.0/ , among the top third, only 1 of its 18 (5.6%) prefectural regions had a high stroke incidence. Similarly, Hunan had a stroke incidence of 160.0/ , among the top third, only 2 of its 14 (14.2%) prefectural regions had a high stroke incidence. Both Henan and Hunan, therefore, did not meet our criteria for inclusion in the stroke belt. The small map in Figure 2 shows that the 9 provincial regions that met the criteria for the stroke belt of high stroke incidence form a near continuous belt. The belt ranges along the boundary of China from southwest (Tibet) to northwest (Xinjiang and Ningxia), and then to north (Inner Mongolia, Beijing, and Hebei) and northeast (Heilongjiang, Jilin, and Liaoning). This C-shaped belt is interrupted by a very slim area of reported lower stroke incidence between Xinjiang and Inner Mongolia. According to the 2001 China Census, the national population (excluding Taiwan, Hong Kong, and Macao) was billion, of whom 239 million (18.9%) resided in the stroke belt, and 1023 million (81.1%) outside the belt. 10 The stroke belt covered 53.6% (515/960 million km 2 ) of China s territory. The overall incidence of stroke in Mainland China was 135.6/ , but the incidence of stroke in the stroke belt was per compared with per in other regions; P<0.001). The incidence of stroke in the belt was more than 2 times (rate ratio, 2.16; 95% confidence interval, ) of that in other regions. Similar patterns were apparent for both men and women; detailed information can be found in the Table II in the online-only Data Supplement. Prevalence of Vascular Risk Factors Within the Stroke Belt Table 2 shows that the 9 provincial regions within the stroke belt were among the top 12 ranked regions for the prevalence of hypertension the major causal risk factor for stroke. 12 The prevalence of hypertension also correlated very significantly with regional stroke incidence (R=0.642; P<0.001, by Spearman rank correlation). Table 3 shows that the mean population prevalence of hypertension was greater in the stroke belt than that in other regions (15.3% versus 10.3%; P<0.001). Six of the 9 provincial regions within the stroke belt were also among the top 9 ranked regions for body mass index >25 (overweight). 13 The prevalence of overweight also correlated significantly with regional stroke incidence (R=0.438; P=0.014). Overweight prevalence also correlated significantly with hypertension prevalence (R=0.664; P<0.001; Table 2). The mean population prevalence of overweight (body mass index, >25) was greater in the stroke belt than that in other regions (21.1% versus 12.3%; P=0.013; Table 3). Table 3 shows that, compared with other regions, the stroke belt was also characterized by a higher proportion of minority

4 1778 Stroke July 2013 Table 1. Characteristics of the Included Epidemiological Surveys on Stroke Incidence in China Name of the Study Stroke Survey in Urban China First Author and Year of Publication Study Period Study Area Li ; urban Wang Stroke Survey in Rural Li rural China 7-5 China Stroke Prevention Project Epidemiological Study Xue ; on Cerebrovascular PLA group Diseases in China ; He ; WHO MONICA study* Wu WHO MONICA study* Wu Xinjiang Stroke Incidence Study Community-Based Stroke Prevention Trial Epidemiological Study of Stroke, Diabetes Mellitus and Cardiovascular Disease (Taiwan) Cheng urban Tan Jiang Hu rural and urban rural and urban rural and urban rural and 1 urban urban rural and 4 urban Sample Size Age Range, y Female in Sample, % Diagnostic Criteria for Stroke All 50.4 Criteria in Rochester Stroke Study Door-to-door survey All 50.3 Criteria in Door-to-door Rochester survey Stroke Study All 52 WHO criteria Door-to-door for stroke survey All China Stroke Door-to-door Conference survey criteria WHO criteria Grassroots for stroke health station reports and death certificates, door-to-door survey in 10% randomly selected study WHO criteria Grassroots for stroke health station reports and death certificates, door-to-door survey in 10% randomly selected study All NA WHO criteria Door-to-door for stroke survey All NA Criteria in ARIC study Case CT/MRI in Case Ascertainment Certification, % Door-to-door survey 8562 > WHO criteria Door-to-door for stroke survey Number of First-Ever Stroke Age-Adjusted Incidence Per Population Years Not reported ( ) Not reported % % Not reported NA in males; in females Not reported NA in males; in females NA ( ) 81.7% % ARIC indicates atherosclerosis risk in communities; CT, computed tomography; PLA, People's Liberation Army; NA, not available; and WHO MONICA, World Health Organization Monitoring Trends and Determinants in Cardiovascular Disease. *A large proportion of the sample population in these two studies 26,27 may be overlapped. We only included the Wu et al 26 study in data analysis because its sample is larger. (non-han) ethnicities (14.80% versus 6.98%; P=0.038), less housing area per population (15.70 versus m 2 / population; P=0.005), more patient beds per 1000 individuals (3.34 versus 2.33; P=0.005), more doctors per 1000 individuals (2.16 versus 1.52; P=0.009), and higher expenditure on health care (8.08% versus 5.62%; P=0.002). Capital cities within the stroke belt are colder (annual average temperature 8.67 C versus C; P<0.001), drier (annual average humidity 57.2% versus 72.1%, P<0.001; annual rainfall versus mm, P=0.001), and with more sunshine hours ( versus hours of sunshine per year; P<0.001). Discussion We have identified a stroke belt of high stroke incidence in China. The belt stretches across 9 provincial regions located in north and west China, where the incidence of stroke was compared with 109.7/ in regions outside the

5 Xu et al Stroke Belt in China 1779 Figure 1. The regional distribution of sampling sites of the 9 enrolled studies. Note that there were few sampling sites in west China where population are sparse. Islets in the South China Sea and East China Sea are not shown in this map. belt (rate ratio, 2.155; 95% confidence interval, ). The population prevalence of hypertension and overweight correlated significantly with regional stroke incidence. A strength of this study is the use of stroke incidence, rather than mortality (as used in identifying the US stroke belt) in evaluating geographic variation in stroke incidence and risk, as it avoids biases because of variation in treatment patterns among different regions. Several limitations should be emphasized when interpreting the results of this study. Retrieving age-adjusted stroke incidence rates from multiple epidemiological surveys may generate bias because of discrepancies in sampling and other methodology. The geographical distribution of the sampling sites of the 9 enrolled studies was uneven. Although the population density in China is uneven, and much denser in east China than west China (eg, >700/km 2 in Jiangsu versus <10/ km 2 in Tibet), there were fewer sampling sites in the less populated regions of western China than in eastern China and, in particular, there was only 1 sampling site in southwest China (Tibet; Table III in the online-only Data Supplement). We have made an assumption that the estimates of a high stroke incidence in Tibet are similar in neighboring regions of southwestern China, where there are no data about stroke incidence. We have included Tibet and surrounding regions in the stroke belt, on the basis of this assumption, but acknowledge that there are limited data for this region and our assumption may be incorrect. The 9 enrolled studies used different reference for evaluating age-adjusted stroke incidence (US 1960 population in 4 study and Segi s world population in 5 studies). However, the effect of using 2 different reference in evaluating age-adjusted stroke incidence seems mild and acceptable in the light of the 2.2 times difference in stroke incidence observed between stroke belt and other regions (Table III in the online-only Data Supplement). The enrolled studies were conducted during 1983 and We cannot evaluate the temporal change of geographic distribution of stroke incidence because of the absence of recent data. Figure 2. Map of China showing age-adjusted stroke incidence in the total population of China according to prefectural regions. The small map (inset) highlights the location of the stroke belt of provincial territories with a high stroke incidence in China. Islets in the South China Sea and East China Sea are not shown in this map. However, stroke risk may change over time in a given region, as observed in some local longitudinal studies. 29 Our data suggest that some of the variation in stroke incidence in China may be attributable to different prevalences of hypertension and obesity among the provinces. The overall prevalence of hypertension in adult Chinese ( 18 years) was estimated to be 18.8% in 2002, 30 a 31% increase compared with Unlike in United States, where only minor geographic variation in blood pressure and hypertension prevalence was observed, 31 4 nationwide hypertension surveys in China conducted in 1959, , , 12 and ; all observed marked regional discrepancies in blood pressure and the prevalence of hypertension. In the 1991 survey, 12 the prevalence of hypertension was greatest in Tibet (19.54%) and lowest in Hainan (6.75%). The prevalence of hypertension in China follows a striking north-south gradient, highest in the north and lowest in the south, which is in concordance with the north-south gradient of stroke incidence observed in other studies. 7,20,23,25 27 Why is hypertension more prevalent in the north than in the south? Daily salt intake has long been proposed as a cause. 34,35 People in a province usually share similar cooking habits and dietary preferences. There are 8 major cuisines in China: Hui (Anhui), Yue (Cantonese), Min (Fujian), Xiang (Hunan), Su (Jiangsu), Chuan (Sichuan), Zhe (Zhejiang), and Lu (Shandong). One distinguishing feature of these cuisines is the preference of salt and other spices. For example, the Lu cuisine, which is popular in north China, favors salty savor. 36 The influences of these regional dietary preferences

6 1780 Stroke July 2013 Table 2. Ranking of Provincial Stroke Incidence and Exposure Ratios of Possible Influential Factors in Population Provincial Region Age-adjusted Stroke Incidence 1/ Urbanization, % GDP per capita RMB Life Expectancy Elderly, % Hypertension, %* Smoking, % BMI>25%* DM, % Heilongjiang (1) 51.5 (6) 8562 (10) 72.4 (13) 5.24 (26) (7) (17) (4) NA (18) Tibet (2) 18.9 (31) 4559 (27) 64.4 (31) 4.50 (29) (1) (31) (22) NA (18) Jilin (3) 49.7 (7) 6847 (15) 73.1 (9) 5.85 (23) (6) (4) (9) 3.27 (6) Liaoning (4) 54.2 (5) (8) 73.3 (7) 7.83 (8) (8) (30) (5) NA (18) Hebei (5) 26.1(26) 7663 (11) 72.5 (12) 6.86 (14) (4) (10) (8) NA (18) Xinjiang (6) 33.8 (17) 7470 (12) 67.4 (27) 4.53 (28) (10) (28) (16) NA (18) Inner Mongolia (7) 42.7 (9) 5872 (16) 69.9 (24) 5.35 (25) (3) (3) 20.97(7) 2.23 (30) Henan (8) 23.2 (30) 5444 (18) 71.5 (17) 6.96 (13) (11) (22) (13) 3.72 (4) Hunan (9) 29.8 (21) 5639 (17) 70.7 (21) 7.29 (11) 8.67 (27) (18) 9.36 (24) NA (18) Beijing (10) 77.5 (2) (2) 76.1 (2) 8.36 (4) (2) (16) (1) 4.56 (1) Ningxia (11) 32.4 (19) 4839 (24) 70.2 (22) 4.47 (30) (12) (13) (14) 2.28 (29) Shandong (12) 38.0 (14) 9555 (9) 73.9 (5) 8.03 (6) (9) (29) (10) 3.48 (5) Hubei (13) 40.2 (12) 7188 (13) 71.1 (20) 6.31 (17) 11.51(13) (11) 11.53(18) NA (18) Guangdong (14) 55.0 (4) (5) 73.3 (8) 6.05 (20) 9.29 (25) (15) (15) 3.11 (28) Chongqing (15) 33.1 (18) 5157 (19) 71.7 (15) 7.90 (7) 8.26 (29) (8) 7.19 (25) 4.37 (2) Gansu (16) 24.0 (27) 3838 (30) 67.5 (26) 5.00 (27) 9.25 (26) (19) (20) 3.24 (7) Tianjing (17) 72.0 (3) (3) 74.9 (3) 8.33 (5) (5) (25) (2) NA (18) Shanghai (18) 88.3 (1) (1) 78.1 (1) (1) (20) (23) (3) NA (18) Hainan (19) 40.1 (13) 6894 (14) 72.9 (10) 6.58 (15) 6.75 (31) (27) 6.17 (28) NA (18) Fujian (20) 41.6 (10) (7) 72.6 (11) 6.54 (16) (18) (24) (19) NA (18) Shaanxi (21) 32.3 (20) 4549 (28) 70.1 (23) 5.93 (22) 9.79 (21) (12) 6.96 (27) NA (18) Qinghai (22) 34.8 (16) 5087 (21) 66.0 (28) 4.33 (31) (15) (7) 9.49 (23) NA (18) Sichuan (23) 26.7 (25) 4784 (25) 71.2 (19) 7.45 (10) 8.26 (30) (9) 6.98 (26) 4.37 (3) Jiangxi (24) 27.7 (24) 4851 (23) 69.0 (25) 6.11 (19) (19) (6) 4.84 (30) NA (18) Yunnan 89.0 (25) 23.4 (29) 4637 (26) 65.5 (30) 6.00 (21) 9.39 (23) (1) 6.02 (29) NA (18) Jiangsu 85.8 (26) 41.5 (11) (6) 73.9 (6) 8.76 (3) (17) (26) (6) 3.23 (8) Anhui 83.0 (27) 27.8 (23) 4867 (22) 71.9 (14) 7.45 (9) (14) (14) (17) NA (18) Shanxi 82.7 (28) 34.9 (15) 5137 (20) 71.7 (16) 6.20 (18) (16) (21) (11) NA (18) Guizhou 78.0 (29) 23.9 (28) 2662 (31) 66.0 (29) 5.79 (24) 8.49 (28) (2) (21) NA (18) Zhejiang 75.5 (30) 48.7 (8) (4) 74.7 (4) 8.84 (2) 9.60 (22) (20) (12) 1.19 (31) Guangxi 73.1 (31) 28.2 (22) 4319 (29) 71.3 (18) 7.12 (12) 9.31 (24) (5) 4.78 (31) NA (18) Taiwan NA NA NA NA NA NA NA NA Correlation with R stroke incidence P Provincial regions in shadow are those enclosed in the stroke belt. Numbers in brackets are descending ranks. Hypertension prevalence was retrieved from a sample population with age limitation of >15 y. 13 DM prevalence was retrieved from a sample population with age imitation of y. 14 Smoking prevalence was retrieved from a sample population with age imitation of y. 15 Prevalence of BMI>25 was retrieved from a sample population with age limitation of y. 16 Data on possible influential factors were retrieved from 2001 China Yearbook of Statistics if not indicated otherwise. 11 Data on Chongqing before 1997 were retrieved from that of Sichuan, from which Chongqing separated as a municipality in RMB (Ren-min-bi) 0.12 US dollar in 2000, and 0.15 US dollar in Urbanization means proportion of rural population in the total population in a given region. Elderly means aged 65 y. Hong Kong and Macao were not included because of the absence of data. Stroke incidence in Taiwan was listed here but not ranked or analyzed because of the absence of other data. BMI indicates body mass index; DM, diabetes mellitus; GDP, gross domestic product; and NA, not available. *Prevalence of hypertension correlated significantly with prevalence of BMI>25 (R=0.664; P=0.000) by Spearman rank correlation. Missing data were replaced by overall prevalence for ranking. Analyzed by Spearman rank correlation. (eg, salty food) on blood pressure and hypertension risk have been confirmed by nutrition surveys and urinary sodium measurements. 34,35,37 However, in this study no direct association between annual salt consumption and regional stroke incidence was observed. This may reflect the fact that in this study the annual salt and spice consumption per capita retrieved from the nationwide sampling survey was based on the resident s recall of total salt consumption. Therefore, the amount may not reflect the real salt intake because salt may be used for purposes other than flavoring or processing foods, as evidenced by the relatively higher salt and spices consumption ( 11 kg annually, or 30 g daily) per capita.

7 Xu et al Stroke Belt in China 1781 Table 3. Exposure Ratios of Possible Influential Factors Against Stroke Belt Factors Stroke Belt* (9 Provinces) Other Regions* (22 Provinces) P Value Stroke incidence (per ) Life expectancy, y Elderly population ( 65 y, %) Illiteracy population, % Population with advanced education, % Population density, per km Urbanization, % Non-Han population, % Sex ratio (men/women) Prevalence of hypertension, % Prevalence of smoking, % Prevalence of DM, % Population with BMI>25, % GDP per capita (RMB) Capital annual consumption expenditure (RMB) Engel coefficient Healthcare expenditure in total consumption, % Sports expenditure in total consumption, % Annual cigarettes consumption per capita Annual alcohol consumption per capita, kg Annual cereal consumption per capita, kg Annual tea consumption per capita, kg Annual vegetable consumption per capita, kg Annual salt and spice consumption per capita, kg Annual fruits consumption per capita, kg Annual fish consumption per capita, kg Annual red meats consumption per capita, kg Annual milk consumption per capita, kg Housing per capita, m Number of patient beds per Number of doctors per Annual SO 2 release, ton/km Annual liquid waste release, ton/km Annual average temperature, C Annual average humidity, % Annual precipitation, mm Annual sunshine, h Mean altitude, m Stroke incidences were age-adjusted according to world standard population. 19 Hypertension prevalence was retrieved from a sample population with age limitation of >15 y. 13 DM prevalence was retrieved from a sample population with age imitation of y. 14 Smoking prevalence was retrieved from a sample population with age imitation of y. 15 Prevalence of BMI>25 was retrieved from a sample population with age limitation of y. 16 Data on possible influential factors were retrieved from 2001 China Yearbook of Statistics or China Yearbook of Health Statistics if not indicated otherwise. 11,18 Engel coefficient indicates proportion of income spent on food. Urbanization means proportion of rural population in the total population in a given region. Elderly means aged 65 y. Salt and spice consumption mainly include salt, monosodium glutamate, and soy sauce. Fish consumption mainly includes aquatic animal products. Red meats consumption mainly includes poultry, beef, mutton, and pork. Data on Chongqing before 1997 were retrieved from that of Sichuan, from which Chongqing separated as a municipality in RMB (Ren-min-bi) 0.12 US dollar in 2000, and 0.15 US dollar in BMI indicates body mass index; DM, diabetes mellitus; GDP, gross domestic product. Taiwan, Hong Kong and Macao were not included due to the absence of data. *Data were calculated from provincial data weighted by population size if not indicated otherwise. Comparison was made by ANOVA, with provincial data treated as individual variables. Data were arithmetic means of the parameters from the capital cities of the corresponding provinces.

8 1782 Stroke July 2013 The cold weather in north China may limit outdoor physical activity in the long winter, enhance fat reservation, 38 and consequently increase the northerners risk of obesity and related metabolic abnormalities. 39 This study observed overweight (body mass index, >25; as defined in the Third National Nutrition Survey) 16 was more prevalent in the stroke belt (north China), and the distribution of overweight correlated significantly with regional hypertension prevalence. Although geographically located in south China, Tibet has a high stroke incidence. Preference for meats (beef and mutton) and salty butter tea (su-you-cha) increased the risk for hypertension and hyperlipidemia in Tibetans. 40,41 The hypobaric hypoxia related to high altitude in Tibet can reactively increase the number of red blood cells (high altitude polycythemia), hematocrit and blood viscosity, which consequently increases the risk of hypertension and stroke. 42 Hypoxic vasoconstriction encountered at altitude may raise pulmonary vascular resistance and arterial pressure and increase the likelihood of right to left shunting, and consequently increase the risk of cryptogenic stroke because of paradoxical embolism. 43 Although multiple data sources have indicated Tibet as a region with high stroke incidence and risk, this presumption cannot be confirmed by the limited data on stroke incidence in this region at present. Future studies of stroke incidence should include more sampling sites in this less populated region. In conclusion, we have identified a stroke belt of high stroke incidence across 9 provincial regions located in west and north China. Higher regional prevalence of hypertension and excess body weight may be responsible and appropriate targets for population-based stroke prevention. Acknowledgments We thank Professor Stephen M. Davis in Melbourne Brain Center, Royal Melbourne Hospital, Parkville, VIC, Australia for his help in formulating the initial hypothesis of a stroke belt in China and planning the study. None. Disclosures References 1. Chen Z. The Third Nationwide Survey on Causes of Death. Beijing: The Peking Union Medical College Press, China s battle with stroke. Lancet Neurol. 2008;7: Heeley E, Anderson CS, Huang Y, Jan S, Li Y, Liu M, et al.; ChinaQUEST Investigators. Role of health insurance in averting economic hardship in families after acute stroke in China. Stroke. 2009;40: Liu L, Wang D, Wong KS, Wang Y. Stroke and stroke care in China: huge burden, significant workload, and a national priority. Stroke. 2011;42: Lanska DJ. Geographic distribution of stroke mortality in the United States: to Neurology. 1993;43: Cushman M, Cantrell RA, McClure LA, Howard G, Prineas RJ, Moy CS, et al. Estimated 10-year stroke risk by region and race in the United States: geographic and racial differences in stroke risk. Ann Neurol. 2008;64: He J, Klag MJ, Wu Z, Whelton PK. Stroke in the People s Republic of China. I. Geographic variations in incidence and risk factors. Stroke. 1995;26: Hankey GJ. Ischaemic stroke prevention is better than cure. J R Coll Physicians Edinb. 2010;40: Hatano S. Experience from a multicentre stroke register: a preliminary report. Bull World Health Organ. 1976;54: National Bureau of Statistics of China. China Statistical Yearbook Beijing: China Statistical Press; Howard G. Why do we have a stroke belt in the southeastern United States? A review of unlikely and uninvestigated potential causes. Am J Med. Sci. 1999;317: PRC National Blood Pressure Survey Cooperative Group. Prevalence and development trends of hypertension in China. Gao Xue Ya Zhazhi. 1995;3: Wang K, Li T, Xiang H. Study on the epidemiological characteristics of diabetes mellitus and IGT in China. Zhonghua Liu Xing Bing Xue Za Zhi. 1998;19: Yang G, Fan L, Tan J, Qi G, Zhang Y, Samet JM, et al. Smoking in China: findings of the 1996 National Prevalence Survey. JAMA. 1999;282: Ge K, Zhai F, Yan H. The third national nutrition survey. Wei Sheng Yan Jiu. 1996; 25: Wang L. China National Nutrition and Health Survey Beijing: People s Medical Publishing House; Ministry of Health of The People s Republic of China. China Yearbook of Health Statistics Beijing: The Peking Union Medical College Press; Segi M. Cancer Mortality for Selected Sites in 24 Countries ( ). Sendai, Japan: Department of Public Health, Tohoku University of Medicine; Hu HH, Sheng WY, Chu FL, Lan CF, Chiang BN. Incidence of stroke in Taiwan. Stroke. 1992;23: Xue GB, Yu BX, Wang XZ, Wang GQ, Wang ZY. Stroke in urban and rural areas of China. Chin Med J (Engl). 1991;104: Li SC, Schoenberg BS, Wang CC, Cheng XM, Bolis CL, Wang KJ. Cerebrovascular disease in the People s Republic of China: epidemiologic and clinical features. Neurology. 1985;35: Wang CC, Cheng XM, Li SC, Wang WZ, Wu SP, Wang KX,et al., Epidemiological survey of neurological disorders in six urban areas of People s Republic of China. Zhong-Hua Shen-Jing Wai-ke Za-Zhi 1985;1: Li ZS, Yang QD, Chen SM, Shu Q, Fu YG. Epidemiological survey of cerebrovascular disease in rural areas of China. Zhong-Hua Shen-Jing Wai-ke Za-Zhi 1989; 5: Cheng XM, Ziegler DK, Lai YH, Li SC, Jiang GX, Du XL, et al. Stroke in China, 1986 through Stroke. 1995;26: PLA Cooperative Group on Cerebrovascular Diseases Survey. Epidemiological Study on Cerebrovascular Diseases in China. Beijing: People s Military Health Press; Wu GX, Wu ZS, He BL. Epidemiological characteristics of stroke in 16 provinces of China. Zhonghua Yi Xue Za Zhi. 1994;74: , Wu Z, Yao C, Zhao D, Wu G, Wang W, Liu J, et al. Sino-MONICA project: a collaborative study on trends and determinants in cardiovascular diseases in China, part I: morbidity and mortality monitoring. Circulation. 2001;103: Tan Y, Yin F, Ya K, Are Z, Pei Z. Epidemiological study on cerebrovascular disease in Xinjiang. Zhongguo Man Xing Bing Yu Fang Yu Kong Zhi.1998;6: Jiang B, Wang WZ, Chen H, Hong Z, Yang QD, Wu SP, et al. Incidence and trends of stroke and its subtypes in China: results from three large cities. Stroke. 2006;37: Li L. China National Nutrition and Health Survey 2002: Hypertension. Beijing: People s Medical Publishing House; Hall WD, Ferrario CM, Moore MA, Hall JE, Flack JM, Cooper W, et al. Hypertension-related morbidity and mortality in the southeastern United States. Am J Med. Sci. 1997;313: Proceedings of China National Cardiovascular Diseases Conference. Beijing: People s Medical Publishing House; Wu YK, Lu CQ, Gao RC, Yu JS, Liu GC. Nation-wide hypertension screening in China during Chin Med J (Engl). 1982;95: Liu LS, Xie JX, Fang WQ. Urinary cations and blood pressure: a collaborative study of 16 districts in China. J Hypertens. 1988;6(Suppl 4):S591 S Intersalt Cooperative Research Group. Intersalt: an international study of electrolyte excretion and blood pressure. Results for 24 hour urinary sodium and potassium excretion. BMJ. 1988;297: Cheng C. The culture of Chinese diet: regional differentiation and developing trends. Acta Geographica Sinica. 1994; 49:

9 Xu et al Stroke Belt in China Hankey GJ. Nutrition and the risk of stroke. Lancet Neurol. 2012;11: Tucker P, Gilliland J. The effect of season and weather on physical activity: a systematic review. Public Health. 2007;121: Zhuo Q, Wang Z, Piao J, Ma G, Zhai F, He Y, et al. Geographic variation in the prevalence of overweight and economic status in Chinese adults. Br J Nutr. 2009;102: Liu L, Liu L, Ding Y, Huang Z, He B, Sun S, et al. Ethnic and environmental differences in various markers of dietary intake and blood pressure among Chinese Han and three other minority peoples of China: results from the WHO Cardiovascular Diseases and Alimentary Comparison (CARDIAC) Study. Hypertens Res. 2001;24: Sherpa LY, Deji, Stigum H, Chongsuvivatwong V, Luobu O, Thelle DS, et al. Lipid profile and its association with risk factors for coronary heart disease in the highlanders of Lhasa, Tibet. High Alt Med Biol. 2011;12: Jha SK, Anand AC, Sharma V, Kumar N, Adya CM. Stroke at high altitude: Indian experience. High Alt Med Biol. 2002;3: Szawarski P, Tam EW, Richards P. Stroke at high altitude. Hong Kong Med J. 2012;18:262.