Prevention of Stroke in Urban China
A Community-Based Intervention Trial
Background and Purpose—Stroke has been the second leading cause of death in large cities in China since the 1980s. Meanwhile, the prevalences of hypertension and smoking have steadily increased over the last 2 decades. Therefore, a community-based intervention trial was initiated in 7 Chinese cities in 1987. The overall goal of the study was to evaluate the effectiveness of an intervention aimed at reducing multiple risk factors for stroke. The primary study objective was to reduce the incidence of stroke by 25% over 3.5 years of intervention.
Methods—In May 1987 in each of 7 the cities, 2 geographically separated communities with a registered population of about 10 000 each were selected as either intervention or control communities. In each community, a cohort containing about 2700 subjects (≥35 years old) free of stroke was sampled, and a survey was administered to obtain baseline data and screen the eligible subjects for intervention. In each city, a program of treatment for hypertension, heart disease, and diabetes was instituted in the intervention cohort (n≈2700) and health education was provided to the full intervention community (n≈10 000). A follow-up survey was conducted in 1990. Comparisons of intervention and control cohorts in each city were pooled to yield a single summary.
Results—A total of 18 786 subjects were recruited to the intervention cohort and 18 876 to the control cohort from 7 cities. After 3.5 years, 174 new stroke cases had occurred in the intervention cohort and 253 in the control cohort. The 3.5-year cumulative incidence of total stroke was significantly lower in the intervention cohort than the control cohort (0.93% versus 1.34%; RR=0.69; 95% CI, 0.57 to 0.84). The incidence rates of nonfatal and fatal stroke, as well as ischemic and hemorrhagic stroke, were significantly lower in the intervention cohort than the control cohort. The prevalence of hypertension increased by 4.3% in the intervention cohort and by 7.8% in the control cohort. The average systolic and diastolic blood pressures increased more in the control cohort than in the intervention cohort. Among hypertensive individuals in the intervention cohort, awareness of hypertension increased by 6.7% and the percentage of hypertensives who regularly took antihypertensive medication increased 13.2%. All of these indices became worse in the control cohort. The prevalence of heart diseases and diabetes increased significantly in the both cohorts (P<0.01). The prevalence of consumption of alcohol increased slightly, and that of smoking remained constant in both cohorts.
Conclusions—A community-based intervention for stroke reduction is feasible and effective in the cities of China. The reduction, due to the intervention, in the incidence of stroke in the intervention cohort was statistically significant after 3.5 years of intervention. The sharp reduction in the incidence of stroke may be due to the interventions having blunted the expected increase in hypertension that accompanies aging as well as to better and earlier treatment of hypertension, particularly borderline hypertension. Applied health education to all the residents of the community may have prevented some normotensive individuals from developing hypertension and improved overall health awareness and knowledge.
Stroke is the third leading cause of death in western countries and is characterized by high rates of case fatality and disability. In the last 2 decades, stroke has become a major public health problem in China. Surveys made in the early 1980s showed that the incidence per year of all stroke was 220.7/100 000 and that of mortality 74.3/100 000 in urban areas of China.1 Meanwhile, national surveys of hypertension and smoking showed that the prevalence of hypertension doubled from 1959 to 1979,2 and the prevalence of smoking increased to as high as 56% in the male population in some areas.3 4 Knowledge about the importance of a healthy lifestyle has been poor. Supported and supervised by the Ministry of Health of the People’s Republic of China and organized by the Beijing Neurosurgical Institute and the Shanghai Medical University, an intervention study was initiated in 7 Chinese cities, including Beijing, Shanghai, Harbin, Changchun, Zhengzhou, Changsha, and Yinchuan. Data from this study have been included in a recent meta-analysis showing a strong relation between hypertension and stroke in eastern Asia and suggesting that population-wide lowering of blood pressure has the potential to produce large reductions in stroke.5 The purpose of the current study was to evaluate the effectiveness of an intervention aimed at reducing multiple risk factors for stroke. The primary study objective was to reduce the incidence of stroke by 25% after 3.5 years of intervention.
Subjects and Methods
Community and Healthcare Systems in Urban China
The structure of society in large cities in China makes feasible an intervention trial based on the community. Community life is well administered, with every resident registered for infectious disease prevention, and everyone is identified by that system. Some retired people living within the local communities voluntarily cooperate with the relevant departments of the local government and hospitals for the above purposes.
The Chinese healthcare system is also unique. Generally, primary healthcare branches in large hospitals located in the urban communities provide to their local communities prenatal and natal health surveillance, family planning, and vaccination of all susceptible children. The staff in these branches are regularly given training in disease prevention. Retired doctors or nurses living within the communities serve in local health stations and provide medical care to local residents.
Organization of Health Care Systems and Quality Control
All 7 collaborating centers were teaching hospitals affiliated with medical universities or colleges and possessed excellent facilities. A healthcare system was established in every city involved in the study. Each institute or department was responsible for different tasks. The Beijing Neurosurgical Institute and the Shanghai Medical University were jointly responsible for designing and conducting the project. A steering committee composed of the principal investigators of the 7 collaborating centers was responsible for the scientific leadership of the study. To ensure uniformity of research methods, a manual of operations was compiled to standardize the research methods and procedures across the communities. The contents included how to select the intervention and control communities, determine the cohort sample size, interview the subjects and perform physical examinations, follow up subjects for the risk factors of stroke, and report and monitor new cases of stroke, and the criteria for the diagnosis of stroke, hypertension, heart disease, and diabetes. All workers who participated in the project were required to receive training in the basics of the study’s design and conduct and to be familiar with the manual of operations before participating in the project. An annual meeting was held with the main researchers from each collaborating center to review the research progress and address problems encountered.
Every week doctors from collaborating center hospitals and primary prevention branches went to the local health stations to see study patients. The workers in local health stations helped by referring the patients to the doctors, distributing leaflets on health education, and identifying and reporting the possible stroke patients in their communities.
The study was reviewed and approved by the human subjects review committees at each of the 7 collaborating teaching hospitals.
Selection of Intervention and Control Cohort and Sample Size
In every city, 2 geographically separated communities with a registered population of approximately 10 000 each and similar demographic characteristics were selected as either intervention or control communities. The determination of intervention community was not random. For logistic reasons, the community that was nearer to the collaborating hospital was selected to be the intervention community. In each community, a cohort containing about 2700 subjects (≥35 years old) was given a baseline survey. Residents were enlisted by study investigators to take the survey door to door until the number of the subjects recruited met the desired sample size. The required sample size in each cohort was estimated on the basis of the following assumptions: According to the baseline survey, the average incidence of stroke among the 7 cities in 1986 ranged from 229 to 237/100 000. Therefore, it was assumed that the incidence of stroke would be 230/100 000 in the control cohort and that the incidence of stroke was hypothesized to decrease by 25% to 180/100 000 in the intervention group after 3.5 years of follow-up. Based on these assumptions and α=0.05 and β=0.10, a sample size of approximately 2000 subjects was needed in each of the cohorts.6 The sample size was increased by an additional 20% to compensate for possible loss of subjects to follow-up.
Baseline and Follow-Up Surveys
In all the cities, a survey was conducted of both intervention and control cohorts from May to July 1987 to obtain baseline data and screen the eligible subjects for intervention. All residents with a history of stroke were excluded from the study. The questionnaire included a medical history and smoking and drinking status. An extensive physical and neurological examination was performed in the study clinics. Seated blood pressure was measured in the right arm after a minimum of 5 minutes’ rest and the avoidance of eating, smoking, and strenuous activity during the previous 15 minutes. The same measurement procedure was repeated after waiting for at least 1 minute. The average of 2 measurements of the first (systolic) and fourth (diastolic) Korotkoff sounds was used for analyses. Anthropometric measurements included weight and standing height. Height and weight measurements were performed using a stadiometer and beam balance scale, with the subjects wearing their usual light indoor clothing without shoes. Body mass index (BMI) was calculated according to the following formula: body weight (kilograms)/body height (meters) squared. The follow-up survey was carried out in mid-1990.
All of the data from baseline and follow-up surveys were checked, coded, and keyed into computers by specially trained workers in each city, and were then sent to Beijing Neurosurgical Institute, where the consistency of the data was checked.
Diagnosis and Classification of Stroke and the Stroke Report System
We use the definition of stroke from the World Health Organization (WHO) International Classification of Diseases, 9th Revision (ICD-9), codes 430 through 438: “rapidly developing clinical symptoms and/or signs of focal, and at times global, loss of cerebral functions, with symptoms lasting more than 24 hours or death, with no apparent cause other than of vascular origin.” Stroke cases were subcategorized as follows: hemorrhagic (ICD-9 codes 430, 431, and 432), ischemic (433 through 434), and ill-defined stroke (436). A fatal stroke was defined as one for which the patient died within 28 days from the onset of symptoms. Before 1990, CT was not widely used, and >50% of stroke diagnoses and classification in our study were based on clinical information alone.
An active surveillance system was established to identify and ascertain any patient with an incident stroke. The details of this system were described earlier.7 In brief, all possible stroke cases were reported by the patients or their family to the workers in local health station (usually retired physicians or nurses), who in turn notified the collaborating center hospital. The diagnosis was verified by some or all of the following: examination of the patient by neurologists or neurosurgeons associated with the study; hospital records; and, in the case of patients who died at home, reports from eyewitnesses. In addition, the government records on death certificates were reviewed annually, and during the follow-up survey in mid-1990, a special effort was made to identified those subjects lost to follow-up who might have had a nonfatal or fatal stroke.
Definitions of Conditions
Subjects were considered hypertensive if the systolic blood pressure (SBP) was ≥140 mm Hg or diastolic blood pressure (DBP) was ≥90 mm Hg. Subjects who were taking antihypertensive medication within the 2 weeks prior to the survey were also considered hypertensive. Hypertensive subjects were categorized as having borderline hypertension (SBP between 140 and 159 mm Hg or DBP between 90 and 94 mm Hg) or established hypertension (SBP ≥160 mm Hg or DBP ≥95 mm Hg or under antihypertensive drug therapy within the 2 weeks preceding the survey).
Subjects were considered to have heart disease or diabetes if they reported a history of heart disease or diabetes diagnosed by a doctor. Details were not ascertained.
Intervention activities included managing the subjects with hypertension, heart disease, or diabetes in the intervention cohort (n≈2700) and provision of health education to the full community (n≈10 000). Subjects in the intervention cohort with established hypertension and a history of diabetes or heart disease were asked to visit the study clinic every 2 weeks; those with established hypertension only were asked to visit the clinic every 4 weeks; and those with borderline hypertension were asked to visit the clinic every 8 weeks. The doctors from the collaborating hospitals visited the intervention cohort each week to provide treatment to the patients. Treatment was not standardized, and the dosage of drugs was individualized. Blood pressure and therapy were recorded at every visit. The treatment of hypertension included pharmacologic treatment as well as lifestyle modification. Lifestyle modification included recommended weight reduction, increased physical activity, and moderation of dietary sodium and alcohol intake. For those with mild hypertension, traditional Chinese antihypertension remedies were first attempted. If this approach failed to control the hypertension, pharmacological treatment was used. Generally, various combinations of drugs from among dihydralazine, triamterene, reserpine, hydrochlorothiazide, and chlordiazeporide were used. For hypertensives with coronary heart disease, beta-blockers were prescribed.
Health education was also made available to all the residents living in the intervention communities but not to those living in the control communities. Information related to stroke prevention was disseminated door-to-door by leaflets, and posters or stickers were distributed throughout the communities. These materials explained the role of hypertension, coronary heart disease, and diabetes in stroke risk and described the risk factors for these illnesses. The goals of these educational efforts were to reduce the chances of people developing these risk factors for stroke and to promote the identification and treatment of those people who developed ≥1 of these illnesses. Those identified were treated as described above.
Differences in means were tested by use of the 2-sample t test, and differences in frequencies were tested with the χ2 test. The Mantel-Haenszel test was used to test the overall difference in stroke incidence between the intervention and the control cohort. Logistic regression analyses were performed to determine whether imbalance in any of the risk factors for stroke might explain the intervention effect. Adjustment was made for age, sex, baseline systolic and diastolic blood pressures, smoking, alcohol use, history of heart disease or diabetes, and BMI. The data were analyzed with SPSS 7.0 (SPSS Inc) and Epi Info 2.0 (CDC/WHO).
A total of 37 662 subjects recruited in baseline survey, 18 786 assigned to the intervention cohort and 18 876 to the control cohort. Characteristics of the intervention and control cohorts are presented in Table 1⇓. The cohorts had similar male-to-female ratios, SBP, DBP, as well as prevalences of smoking, hypertension, and diabetes. Compared with the control cohort, the subjects in the intervention cohort were 1 year younger and had a significantly higher BMI, prevalence of heart disease (all with P<0.001), and alcohol use (P=0.05). The follow-up survey was carried out 3.5 years later: 80.7% (15 164/18 786) of the subjects in the intervention cohort and 78.5% (14 827/18 876) of the subjects in the control cohort participated in the follow-up survey. The reasons for being lost to follow-up were refusal to be interviewed or having moved out of the area. Table 2⇓ gives the comparison of the characteristics between the subjects participating in the follow-up examination and those not participating (including those lost to follow-up and those who died from causes other than stroke before the follow-up). There were more male than female nonparticipants in both cohorts. The nonparticipant subjects from the intervention cohort were 1.2 years older and had higher levels of SBP and DBP and a higher frequency of heart disease than those completing the 1990 follow-up. In the control cohort, most of the characteristics between the participants and nonparticipants were similar except for the level of DBP and the prevalence of diabetes.
The incidence of stroke in the year prior to the baseline examination (1986) was comparable in the intervention (229/100 000) and the control communities (237/100 000, P=0.802). The incidence of stroke varied from 325/100 000 in Harbin to 151/100 000 in Yinchuan (Table 3⇓), although the differences was not statistically significant.
As shown in Table 4⇓, 174 new cases of stroke (116 nonfatal and 58 fatal) occurred in the intervention cohort and 253 (156 nonfatal and 97 fatal) in the control cohort during the intervention period (May 1987 through December 1990). The 3.5-year cumulative incidence of stroke was significantly lower in the intervention cohort than the control cohort (0.93% versus 1.34%; RR=0.69; 95% CI, 0.57 to 0.84). Differences existed for both nonfatal and fatal stroke. The nonfatal stroke rate was 25% lower (RR=0.75; 95% CI, 0.58 to 0.96) and the fatal stroke rate 40% lower (RR=0.60; 95% CI, 0.43 to 0.84) in the intervention cohort than the control cohort. Adjustment for other risk factors had essentially no effect on the relative risks for intervention versus control (data not shown). The absolute difference in the cumulative incidence of all strokes between the 2 groups was 1.34%−0.93%=0.41%. The number needed to treat with the intervention to prevent 1 stroke would be 100/(0.41) =243. Thus, approximately 243 people would need to be screened and followed-up with appropriate intervention to prevent 1 stroke during a 3.5 year period.
In the intervention cohort, the 3.5-year cumulative rates were 0.48% for ischemic stroke and 0.43% for hemorrhagic stroke; for the control cohort, rates were 0.69% for ischemic stroke and 0.65% for hemorrhagic stroke (Table 4⇑). There were 2 ill-defined stroke cases in the intervention cohort and 1 in the control cohort. The ratio of ischemic to hemorrhagic stroke was 1:1.1 in both cohorts. The rates of both subtypes of stroke were significantly reduced in the intervention cohort compared with the control cohort: 29% (RR=0.71; 95% CI, 0.53 to 0.93) for ischemic stroke and 33% (RR=0.67; 95% CI, 0.50 to 0.89) for hemorrhagic stroke, suggesting that the intervention was effective for the different subtypes of stroke. Again, adjustment for the other risk factors had no important effect on these relative risks (data not shown).
The intervention effect by city is not shown in Table 4⇑. A reduction of nonfatal and fatal stroke as well as ischemic and hemorrhagic stroke was observed in the cities of Beijing, Shanghai, Harbin, Zhengzhou, and Changsha but not in the cities of Changchun and Yinchuan. For both fatal and ischemic stroke, the intervention and control cohorts in Changchun differed by only 1 case. The stroke incidence was low in Yinchuan, therefore the absolute number of the stroke cases was small during the study period.
Before the intervention, the prevalence of hypertension and the levels of SBP and DBP were similar between the 2 cohorts (Table 1⇑), but as shown in Table 5⇓, the proportion of established hypertension in the intervention cohort (26.8%) was significantly larger than that in the control cohort (23.9%, P<0.001). While there were more hypertensive individuals aware of their high blood pressure condition, fewer were under regular antihypertensive drug therapy in the intervention cohort than the control cohort. After 3.5 years of intervention, the prevalence of hypertension increased in both cohorts; however, the increment was 7.8% in the control cohort whereas it was only 4.3% in the intervention cohort. As displayed in Table 5⇓, the average SBP and DBP increased more in the control cohort than in the intervention cohort. In addition, among hypertensive individuals, awareness of hypertension increased by 6.7% in the intervention cohort and decreased by 6.4% in the control cohort. The percentage of hypertensives who regularly took hypertensive medication increased 13.2% in the intervention cohort and decreased 1.2% in the control cohort.
The prevalence of heart disease increased from 11.4% to 12.3% (P<0.01) in the intervention cohort and from 10.3% to 11.9% (P<0.01) in the control cohort. The prevalence of diabetes also increased significantly in both cohorts, but the increment was larger in the intervention cohort (from 1.2% to 1.8%, P<0.01) than the control cohort (from 1.0% to 1.4%, P<0.01). The prevalence of current drinkers of alcohol increased slightly but was not significantly different, and the prevalence of smoking remained constant in both cohorts.
The cities involved in the study had varying rates of stroke morbidity and are located in different parts of China: 2 in the northeast (Harbin and Changchun), 2 in the north (Beijing and Zhengzhou), 2 in the south (Shanghai and Changsha) and 1 in the northwest (Yinchuan). A recent meta-analysis showed that geographic variation of stroke incidence and mortality in China is striking, and rates are generally higher in the north than in the south. The differences may be as large as 2- to 8-fold.8 However, in our study, the city of Changsha, which is located in southern China, had the second highest stroke incidence in 1986 (prior to baseline) and the highest cumulative stroke incidence during the 3.5-year study.
The baseline and cumulative stroke data were obtained by different methods. The former were retrospective data and the later were prospective. At the follow-up survey, information pertaining to the missing subjects was collected when possible to identify all subjects who might have experienced a stroke. This approach might account for the different incidence rates of stroke between baseline and cumulative data among the 7 cities.
The cumulative incidence of nonfatal and fatal stroke was lower in the intervention cohort than in the control cohort in most of the cities. In general, the stroke cases were less likely to be missed in the intervention cohort than in the control cohort, because it had closer contacted with the local health services. Thus, any bias due to loss to follow-up would likely have led to an underestimation of the benefit of the intervention. While there was considerable loss to follow-up in both the intervention and control cohorts, those lost to follow-up differed at baseline in only trivial ways from those who were followed up. Further, the loss to follow-up was unlikely to be related to the intervention. Thus, significant bias due to this source is improbable.
Hypertension, coronary heart disease, atrial fibrillation, diabetes, obesity, smoking, and high alcohol consumption are major risk factors of stroke.9 10 11 Also, the risk of stroke doubles or triples for every 10-year increment of age.12 In our study, the proportion of subjects consuming large amounts of alcohol was small, and the distribution of most of the risk factors at the baseline survey was similar between the intervention and the control cohorts. At baseline, subjects in the intervention cohorts were 1 year younger and had higher BMIs and higher prevalences of heart disease and alcohol use than those in the control cohort. However, these differences were not large, and controlling for them had no effect on the relative risk associated with intervention.
Hypertension is considered the strongest predictor of stroke and is believed to account for 70% of stroke.13 It is proposed to increase the risk of stroke by aggravating atherosclerosis in the aortic arch and the carotid, vertebral, and basilar arteries; causing arteriosclerosis and lipohyalinosis in the small-diameter, penetrating arteries of the cerebrum; and contributing to heart disease, of which stroke is a complication.14 Elevated SBP and DBP have been associated with an increase in the incidence of both ischemic and hemorrhagic stroke.15 The meta-analysis of the association between treatment of hypertension and cardiovascular disease from 14 clinical trials showed that treatment of hypertension can effectively reduce the incidence of stroke in both older and middle-aged hypertensives.16 17
To prevent stroke, we focused mainly on primary and secondary prevention of hypertension, namely preventing normotensive subjects from developing hypertension and treating hypertensive patients. Our results showed that intervention was effective in decreasing both nonfatal and fatal stroke. Compared with the control cohort, the overall incidence of nonfatal stroke was 25% less and fatal stroke was 40% less in the intervention cohort. Analyses of the stroke cases by subtype show that the overall incidence of ischemic stroke was slightly higher than that of hemorrhagic stroke and intervention was effective in preventing both ischemic and hemorrhagic stroke. However, the effect of intervention on subtypes of stroke is difficult to assess confidently, because misclassification of subtypes of stroke is possible.
We hypothesize that the decrease in stroke rates in the intervention cohort was due to the reduction of hypertension, because the frequency of heart disease and diabetes increased and the prevalences of smoking and alcohol drinking remained unchanged. Perhaps strategies aimed at preventing or treating these other risk factors, especially smoking, would be associated with an even greater benefit than that seen in the current trial.
This study has many strengths. Prime among them are a healthcare system that lends itself to such a community-based intervention trial, the study’s large size, and its systematic follow-up for the outcomes of interest. It also has several potential shortcomings. Investigators could not perform the intervention and determination of outcome in a blinded fashion. The follow-up survey was not completed in all subjects. In the intervention group, those who did not participate in the follow-up survey had more stroke risk factors than those who participated; such difference were not as marked in the control group. Regardless of whether subjects participated in the follow-up survey, all subjects were monitored for the occurrence of stroke. The intensity of the monitoring may have been greater in the intervention cohort than the control cohort. If such a difference did exist, it would have served to diminish the beneficial effect seen in this trial. Less than half of the patients who suffered a stroke had neuroimaging, so the presentation of results by subtype may be compromised by misclassification. Regardless, the classification employed may be useful in parts of the world in which neuroimaging is not uniformly performed in patients with acute stroke. These places may also be the ones to have the most to gain from the preventive strategy used in this study. If we assume that none of these problems are sufficient to explain away the findings, we are still left with an inability to know from this study what part of the intervention is responsible for the reduced incidence of stroke. This fact is the beauty of epidemiology, namely, that an imperfect or incomplete understanding of a disease or intervention does not preclude investigators from finding effective means of preventing the disease.
Our results demonstrated that compared with other risk factors, prevention of hypertension was more readily accepted and implemented in the community. With the process of aging, the prevalence of hypertension increased in both intervention and control cohorts during the study period, but the magnitudes were different. In the intervention cohort fewer subjects developed hypertension, more hypertensives were aware of their condition, and the average level of SBP decreased. These indices remained the same or became worse in the control cohort.
The prevalence of heart disease and diabetes was increased in both cohorts, as is anticipated in an aging population. While the increase of heart disease was similar in both cohorts, the prevalence of diabetes increased more in the intervention cohort than in the control cohort. We believe that the disproportionate increase in the intervention cohort was due to improved knowledge of diabetic signs and symptoms because of community health education and regular medical contacts. We did not see a significant decrease in smoking or alcohol use in the intervention cohort as expected. These findings suggest that alternative intervention might be necessary to change these unhealthy behaviors.
In conclusion, our study has important public health implications. Our results demonstrate that community-based intervention for stroke is both feasible and effective in China. The reduction, due to the intervention, in the incidence of stroke was statistically significant after 3.5 years of intervention. The reduction of stroke may be due to the interventions having blunted the expected increase in hypertension that accompanies aging as well as better and earlier treatment of hypertension, particularly borderline hypertension. Health education for the residents of communities may prevent normotensive individuals from developing hypertension and improve overall health awareness and knowledge.
This study was supported by the funds from the Ministry of Health, People’s Republic of China. We thank the following collaborative centers and investigators: Drs Jie Chen, Yu-xiang Guo, Su-Ge Bao, and Qiu-ju Bao from Beijing Neurosugical Institute; Drs Guo-Xing Jiang, Zhi-Ping Qu, and Er-Juan Feng from Institute of Neurology, Shanghai Medical University; Dr Ming-Li Rao from Bethune Medical University; Drs Qin-Shun Dai and Yu-Chun Yang from Harbin Medical University; Dr Fang-Zhong Su from Henan Medical University; Drs Xue-Fen Lu and Qi-Dong Yang from Hunan Medical University; and Drs Jia-Ren Song and Fan-Yuan Kong from Ningxia Medical College. We also thank for their help Prof Steve Self and Dr Ashley Hedeen from the University of Washington and Mrs Grace A. Brandault.
- Received September 8, 1998.
- Revision received December 21, 1998.
- Accepted December 21, 1998.
- Copyright © 1999 by American Heart Association
Li WX. [Smoking, and excess mortality]. Chin J Epidemiol.. 1984;5:91–94.
Lemeshow S, Hosmer Jr. DW, Klar J, Lwanga SK. Adequacy of sample size in health studies. New York, NY: John Wiley & Sons; 1990:21–23.
Cheng XM, Ziegler DK, Lai Y-HC, Li S-C, Jiang G-X, Du X-L, Wang W-Z, Wu S-P, Bao S-G, Bao Q-J. Stroke in China, 1986 through 1990. Stroke. 1995;26:1990–1994.
He J, Michael JK, Wu Z, Whelton PK. Stroke in the People’s Republic of China, I: geographic variations in incidence and risk factors. Stroke. 1995;26:2222–2227.
He J, Michael JK, Wu Z, Whelton PK. Stroke in the People’s Republic of China, II: meta-analysis of the hypertension and risk of stroke. Stroke. 1995;26:2228–2232.
Wolf PA, Dawber TR, Thomas HE, Kannel WB. Epidemiological assessment of chronic atrial fibrillation and the risk of stroke: the Framingham Study. Neurology. 1978;28:973–977.
Beilin LJ, Puddey IB, Burke V. Alcohol and hypertension: kill or cure? J Hum Hypertens. 1996;10(suppl 2):S1–S5.
Dunbabin DW, Sandercock PAG. Preventing stroke by modification of risk factors. Stroke. 1990;21(suppl IV):IV-36–IV-39.
MacMahon S, Peto R, Cutler J, Collins R, Sorlie P, Neaton J, Abbott R, Godwin J, Dyer A, Stamler J. Blood pressure, stroke, and coronary heart disease, part 1: prolonged differences in blood pressure: prospective observational studies corrected for the regression dilution bias. Lancet. 1990;335:765–774.
Collins R, Peto R, MacMahon S, Hebert P, Fiebach NH, Eberlein KA, Godwin J, Qizilbash N, Taylor JO, Hennekens CH. Blood pressure, stroke, and coronary heart disease, part 2: short-term reductions in blood pressure: overview of randomized drug trials in their epidemiological context. Lancet. 1990;335:827–838.