Effects of a Long-term Hypertension Control Program on Stroke Incidence and Prevalence in a Rural Community in Northeastern Japan
Background and Purpose—Although randomized clinical trials have demonstrated the benefit of antihypertensive treatment in preventing stroke, the effectiveness of community-based programs is largely unknown. We investigated long-term community-based prevention activities.
Methods—In rural northeastern Japan, people aged ≥30 years numbered 3219 in the full intervention community and 1468 in the minimal intervention community in 1965. Systematic blood pressure screening and health education began in 1963. Stroke was registered through 1987.
Results—More than 80% of people aged 40 to 69 years were screened in both communities in the 1960s. One community charged for screening services after 1968, whereas the other community intensified intervention; subsequently, screening rates and the follow-up of hypertensive individuals declined in the minimal intervention community, especially in men. In men, stroke incidence declined more (P<0.001) in the full intervention (42% in the period 1970 to 1975, 53% in the period 1976 to 1981, and 75% in the period 1982 to 1987) than in the minimal intervention community (5% increase, 20% decrease, and 29% decrease, respectively); in women, the stroke incidence declined about 45% to 65% in both communities. Changes in stroke prevalence paralleled those in stroke incidence. Trends in systolic blood pressure levels tend to explain the differential stroke rates in men.
Conclusions—Delivery of hypertension control services through intensive, free, community-wide screening and health education was effective in prevention of stroke for men in a community.
Experimental trials have demonstrated that treatment of hypertension prevents stroke incidence and reduces stroke mortality in middle-aged and elderly adults in both clinical and community settings.1 2 3 4 5 6 Improvement in the detection, referral, and treatment of hypertension has been demonstrated by community trials compared with control communities.7 8 9 10 However, none of these community trials, primarily designed for prevention of coronary heart disease, have demonstrated reduced stroke rates.11 12
Japan suffers higher mortality from stroke and lower mortality from coronary heart disease than Western countries.13 To ameliorate the epidemic of stroke, community-based programs of hypertension control were launched in Japan in the 1960s, including those we implemented in 1963 in 2 agricultural communities in northeastern Japan.14 Hypertension control efforts permeated both communities. However, the municipal government in 1 of the communities started to charge the participants for blood pressure screenings after 1968 and did not replace the public health nurse when she retired in 1973, whereas the government in the other community continuously supported the community-based intervention program. This circumstance permitted the observation of the long-term effect on blood pressure and its clinical sequelae according to 2 levels of intensity of an intervention program. Furthermore, the full intervention community conducted systematic education classes for detected hypertensives, and used a home broadcasting system for verbal health announcements via a speaker attached to the telephone. No systematic classes or mass-media education was conducted in the minimal intervention community.
We compared the 2 communities for participation in blood pressure screenings and follow-up examinations and for trends in blood pressure level and other stroke risk factors, stroke incidence, prognosis, and prevalence between 1963 and 1987.
Subjects and Methods
The surveyed communities were Ikawa (full intervention) and a district of city H (minimal intervention). These rural rice-farming communities are 45 miles apart in Akita Prefecture, northeastern Japan. Despite geographical closeness, the 2 communities have different local governments and communicate little. In the full intervention community, the censuses for ages ≥30 were 1509 men and 1710 women in 1965, and 1933 men and 2182 women in 1985. The populations aged ≥70 were 96 men and 128 women in 1965, and 252 men and 359 women in 1985. The minimal intervention community had about half as large a population as the full intervention community and a similar age-sex distribution (Table 1⇓). A previous investigation in both communities showed only a 1% emigration in 10 years.14
Blood Pressure Screenings and Surveys
Residents aged ≥30 in both communities were offered identical blood pressure screenings. Initial screening was by physicians and public health officials. Hypertensive individuals were rescreened annually, while the rest of the community was rescreened every 4 years in the full intervention community and every 2 years in the minimal intervention community.
Systolic and fifth-phase diastolic blood pressures were measured by physician epidemiologists using a standard mercury sphygmomanometer on the right arm after at least 5 minutes’ seated rest. The first measurement was used in data analysis. Hypertension was defined as systolic blood pressure ≥160 mm Hg and/or diastolic blood pressure ≥95 mm Hg and/or current use of antihypertensive medication. Controlled hypertensives were treated persons with systolic blood pressure <160 mm Hg and diastolic blood pressure <95 mm Hg.
Color photographs of the right ocular fundus were coded according to Scheie’s classification.15 Grades II or higher hypertensive or arteriosclerotic changes were regarded as a significant hypertensive change in the retinal arterioles. Using the Minnesota Code16 in a supine resting ECG, we regarded a high R wave (3–1) plus ST-T changes (4–1 to 4–3 or 5–1 to 5–3) as hypertensive cardiac changes; ischemic changes were rare.17 We estimated the prevalence of atrial fibrillation (8–3).18
We asked about history of diabetes mellitus, hypertension, stroke, and coronary heart disease since 1963. Starting in 1975, we asked about usual weekly alcohol intake (grams/d)19 and smoking habits.
A surveillance team identified incident strokes in people aged ≥30, based on hospital reports, national insurance claims, local physicians, ambulance records, death certificates, public health nurses, and blood pressure surveys. In a household survey performed between 1964 and 1966 we found no undetected cases.14 In living cases we obtained a history, made systematic neurological examinations, and reviewed medical records of the local clinics and hospitals. Decedents’ histories were obtained from families, attending physicians, and/or medical records. Final standardized diagnoses of stroke (a focal neurological disorder with rapid onset that persisted for 24 hours or more,20 21 excluding transient ischemic attack) were made by a panel of 3 or 4 physician-epidemiologists, who were blinded to risk factor status and the diagnoses of other panelists, aware of the community of residence, but unaware of the intention eventually to compare the 2 communities. The surveillance and stroke diagnosis methods were identical across communities.
We did not examine trends in stroke mortality or type of stroke, because the population size was small, the accuracy of the death certificate diagnosis was questionable before the introduction of CT in the 1980s,22 and many stroke victims ultimately die of other causes.23
1-Year Stroke Prognosis and Stroke Prevalence
Stroke prognosis 1 year after onset was dead, dependent, or independent with or without neurologic deficit (eg, hemiplegia or dysarthria). Dependent stroke patients were those needing help with at least 1 of the following: bathing, eating, dressing, toileting, or walking. Stroke prevalence was defined as dependent or otherwise neurologically deficient stroke.
Hypertension Control Program: Full Intervention Community
Strategies for hypertension control24 included (1) systematic blood pressure screening for detection of hypertensive individuals; (2) referral of high-risk individuals to local clinics for antihypertensive medication to reduce high blood pressure or treat end-organ effects in the retinal arterioles or ECG; (3) health education for hypertensives at blood pressure screening sites, adult classes, and nurse home visits; (4) training of about 150 “healthy diet” volunteers for health education; and (5) community-wide media-disseminated education to encourage participation in blood pressure screening and reduced salt intake.
The network for the stroke prevention program included task forces, which met twice yearly to discuss program implementation, 3 public health nurses, 4 midwives, 28 community leaders, and about 180 of their coworkers, ie, small district–leaders. The Osaka Medical Center for Cancer and Cardiovascular Disease, the University of Tsukuba, and the Akita Research Institute of Public Health helped plan and organize the program and conducted the population surveys and surveillance. Repeated systematic blood pressure screenings were free of charge. High participation was a priority. Community- and small district–leaders helped recruit residents aged ≥30 using direct contact and recruitment letters and helped arrange blood pressure screenings at community centers and schools. To enhance participation, blood pressure screening was added on Sundays beginning in 1971, and in the evening beginning in 1977.
Between 1964 and 1967, about 100 to 150 hypertensive individuals requiring medication were visited annually to confirm a referral, provide health education, and enhance compliance. Hypertensives requiring medication at the initial screening (n=519) were invited by 3 nurses and 4 midwives to 6 to 10 adult education classes annually from 1968 through 1970 and 4 adult classes annually in the later years. Hypertensives not needing medication (n=806) were invited to 2 adult classes annually. Newly detected hypertensives were also invited to adult classes. Approximately 70% to 80% of the hypertensive individuals attended adult classes. The rest were visited at home 1 to 2 times annually.
Treatment of hypertension was by local physicians, using primarily thiazide diuretics and secondarily β-adrenergic blocking agents. Calcium channel antagonists and angiotensin-converting enzyme inhibitors were rarely used before the mid-1980s.
Classes dealt with blood pressure measurement and management and how to reduce salt intake, including taste tests of low-salt soy bean soup and pickles. Education focused on salt because average sodium intake in the 1960s was 20 g/d14 in a traditional Japanese diet: a high intake of rice, salty soybean soup, and salt-preserved pickles and fish, with low intake of meat, eggs, and dairy foods.14 24 Reduction of excessive alcohol intake such as 5 drinks or more per day was also emphasized. We recommended that farmers rest adequately because farm work was extremely hard, but did not emphasize weight control because most hypertensive individuals were not obese in the 1960s.14
In 1967, about 150 volunteers for diet improvement were trained through 10 annual classes that enhanced knowledge of stroke and practical ways of modifying diet and lifestyle. The volunteers offered health education to about 1000 people per year at blood pressure screenings and meetings at local public centers.
Municipal announcements were transmitted via a speaker attached to each household telephone, originally intended for fire or earthquake emergencies. Individuals could turn the speaker off. Announcements recruited participants to blood pressure screenings and adult classes 1 week before and during the events and for a regular program on cardiovascular health, aired for 3 minutes at 6:30 am, 12:30 pm, and 6:30 pm every Thursday. Rotating topics included reduction of salt intake, the importance of balanced diet, and the importance of proper rest.
Hypertension Control Program: Minimal Intervention Community
A similar organization was established in 1963 in the minimal intervention community, including 1 public health nurse and 1 local clinic. Strategies of the program were similar to those in the full intervention community, but did not include adult classes or community-wide media education. The municipal announcement system was used only for recruiting participants to blood pressure screenings. The initial blood pressure screening performed from 1964 through 1968 was free of charge. However, the local government started to charge the participants for blood pressure screening beginning in 1969.24 Between 1964 and 1972, the enthusiastic nurse visited about 200 hypertensives each year. After her retirement in 1973, the systematic visits ended. Three fourths of the referred hypertensives attended a clinic outside the community.
To evaluate exposure to the intervention program, participation rates at 4 blood pressure screenings between the 1960s and the 1980s were compared between the 2 communities using the χ2 test. Participation in follow-up examinations of stroke-free hypertensive individuals detected at the initial blood pressure screening was compared between communities by the Wilcoxon rank sum test.
Blood pressure levels were age-adjusted, and their population differences were tested in each time period using ANCOVA. Linear trends in 6-year stroke incidence per 1000 people and 1-year prognosis of stroke were assessed using a χ2 test for trend for the periods 1964 to 1969, 1970 to 1975, 1976 to 1981, and 1982 to 1987 (Figure 1⇓). Trends in stroke prevalence were also assessed by a χ2 test for trend in 1972, 1977, 1982, and 1987 (Figure 2⇓). Community differences in stroke incidence and prevalence in each time period were examined by χ2 test. Direct age-adjustment of rates and proportions was calculated using 10-year categories of the 1960s’ pooled populations.
The age-adjusted participation rate in both communities in the 1960s blood pressure screening was approximately 70% to 80% for men and 80% to 90% for women (Table 1⇑). Between the 1960s and the mid-1980s, the age-adjusted participation rates remained high in the full intervention community, but declined in the minimal intervention community to about 50% in men and 60% in women after the mid-1970s, particularly in ages 30 to 39 and ages ≥70. The following analyses of blood pressure were restricted to men and women aged 40 to 69 years, groups whose baseline participation rates were high.
Seventy-seven percent of the initially screened hypertensive men (n=266) in the full intervention community had 2 or more follow-up screenings in 10 years compared with 59% of hypertensive men (n=112) in the minimal intervention community; median numbers of follow-up screenings were 4 and 2, respectively (P<0.01). The respective percentages in 19 years were 82% and 65%; median numbers of follow-up screenings were 5 and 4 (P<0.01). For women, no community difference was evident: 79% of hypertensives (n=217) in the full intervention community versus 75% of hypertensives (n=114) in the minimal intervention community in 10 years and 84% versus 76% in 19 years; median number of participation was 4 versus 4 in 10 years and 6 versus 5 in 19 years.
Sex and age-adjusted 6-year stroke incidence rates among persons aged ≥30 did not differ between the 2 communities in the period 1964 to 1969 (Table 2⇓, Figure 1⇑). Stroke incidence declined in men in subsequent periods (test for trend within each community: P<0.001) and was significantly greater in the full intervention than in the minimal intervention community: 42% decrease versus 5% increase in the period 1970 to 1975 (P<0.01), 53 versus 19% decrease in the period 1976 to 1981 (P<0.05) and 75 versus 29% decrease in the period 1982 to 1987 (P<0.001). For women, there was no difference in stroke incidence in the 3 later periods.
No significant difference in 1-year prognosis was found between the communities in the period 1964 to 1969. The proportion of fatal strokes were 33% to 43% in the period 1964 to 1969 and subsequently declined by half (P for trend <0.05) in both communities (except for full intervention community men, who rebounded to baseline levels in the period 1982 to 1987). No consistent trend was seen in the proportions in the other prognosis categories. Sex and age-adjusted prevalence of stroke among persons aged ≥30 did not differ between the 2 communities in 1972 (Table 3⇓, Figure 2⇑). For men, stroke prevalence declined in the full intervention community in subsequent periods (test for trend: P<0.001), while the prevalence did not change significantly in the minimal intervention community. For women, stroke prevalence declined in both communities (P<0.001); prevalence was reduced in full intervention compared with the minimal intervention community only in 1982.
The only source of information about blood pressure and hypertension status is the blood pressure screenings, which may be biased by differential response rates. Among men and women aged 40 to 69 years in the 1960s, age-adjusted mean values of screening systolic and diastolic blood pressures, prevalence of hypertension, and hypertensive end-organ defects were almost identical between the 2 communities for both sexes (Table 4⇓). The proportion of hypertensive individuals who were treated was only 17% to 25% for men and women in the 2 communities. As a result of screening, approximately 80% of the detected hypertensives were referred to local physicians for treatment in both communities. For men, the full intervention community showed a 3 to 4 mm Hg lower mean systolic blood pressure than the minimal intervention community in the early 1970s and the early 1980s but not in the mid-1980s. A community difference in systolic blood pressure for women was found in the early 1980s but not in the other periods. The prevalence of hypertensive end-organ defects was lower in the full intervention community than in the minimal intervention community for men and women in the early 1970s and for women in the mid-1980s. The proportion of hypertensives was consistently smaller in the full intervention community than in the minimal intervention community after the early 1970s for both sexes. Contrary to expectations, among hypertensives the proportion previously undetected was consistently greater and the proportion treated and controlled was lower in the full intervention community than in the minimal intervention community.
For both men and women, the prevalence of atrial fibrillation and diabetes mellitus ascertained at screening was ≤2% in both communities in the 1960s. The prevalence of atrial fibrillation did not change over time in either community. The prevalence of diabetes mellitus increased over time similarly in both communities. The proportions of current smokers and heavy drinkers did not change in the latest 2 time periods for either community, although these behaviors were less frequently reported in the minimal intervention community than in the full intervention community.
The present study showed a larger decline in stroke incidence and prevalence in men aged ≥30 in a rural Japanese community that between the 1960s and the mid-1980s received a full range of community-wide hypertension interventions than that seen in a demographically similar community that received minimal intervention after 1968. The overall decline in incidence for men was 75% in the full intervention community and 29% in the minimal intervention community. Stroke occurrence in women was less affected, although the prevalence of stroke was differential between the 2 communities in 1982. The incidence of coronary heart disease, which has been reported elsewhere,14 24 was less than 10% of the stroke incidence in the 1960s and did not change over time in either community.
We interpret these findings as indicating that widespread screening, referral, and follow-up for hypertension, supplemented by community-based health education and broad citizen support (including specific support roles for over 10% of the population over age 30), were successful in reducing stroke rates for men in this Japanese rural environment.
After 1968, when the municipal government minimized its support, participation in blood pressure screenings and hypertensive follow-up decreased in the minimal intervention community, especially in men. Men’s work schedules and generally low level of health-consciousness may have contributed to reduced participation.24 On the other hand, participation by men in the full intervention community may have been maintained both by the absence of charges and by more frequent and flexibly scheduled blood pressure screenings. In addition, the success of the intervention in men may be due partly to initial high risk (average 149 mm Hg for systolic blood pressure and 86 to 87 mm Hg for diastolic blood pressure) and a high prevalence of hypertensive end-organ defects.
Why was the community difference in stroke occurrence minimal in women? First, the initial blood pressure levels and stroke incidence were lower in women than in men. Second, women were probably more health conscious than men and were more available during the daytime. Decline in participation in the minimal intervention community was not as great among hypertensive women initially screened as it was among men. This suggests that there were similar changes in health-related behaviors in women between the 2 communities.
We hypothesized that systematic screening, referral, and follow-up examinations, coupled with intensive health education and recognition of personal risk in hypertensive individuals and the whole population, are fundamental to community blood pressure control. In a community heart disease prevention trial in the United States, systematic screening encouraged health behavior changes resulting in lower blood pressures and blood cholesterol levels.25 If screening is the intermediary for stroke reduction, blood pressure should change in proportion to the intensity of screening and related activities. In support of this hypothesis, in our study mean systolic blood pressures in men declined more in the full intervention community than in the minimal intervention community in the early 1970s and the early 1980s, although no community difference was seen in the mid-1980s. This blood pressure trend corresponded with the greater decline of stroke incidence and prevalence in the full intervention community. The community difference in mean systolic blood pressure was only seen in the early 1980s for women, when stroke prevalence declined in favor of the full intervention community. Diastolic blood pressure, which contributes less to the development of stroke than systolic pressure,26 did not change between communities.
We consider it likely that the reduced screening participation (by 16% to 30% in men and 17% to 25% in women) and a consequent failure to detect newly developed hypertensives in the minimal intervention community contributed to a smaller reduction of stroke incidence there. In the minimal intervention community, undetected hypertensive individuals may have been less likely to participate in blood pressure screenings, resulting in the observation of a consistently smaller proportion of previously undetected hypertensives and a greater proportion of treated and controlled hypertensives after the 1970s. Our recent study27 in a different community supports this viewpoint. There, 46% (2062/4496) of the population aged 40 to 69 years responded to a survey; 2 years later, we examined 87% (312/360) of a random sample of nonparticipants. In nonparticipants, age-adjusted systolic blood pressure levels were higher (11 mm Hg in men and 4 mm Hg in women) and the prevalence of antihypertensive medication use was lower (3% in men and 5% in women) than in participants.27
A recent study indicated that treatment of hypertension had little impact on the population risk of stroke, although high blood pressure, current smoking, atrial fibrillation, and diabetes mellitus increased risk of stroke.28 29 We examined trends in these stroke risk factors as well as heavy drinking,18 19 28 but found no community difference in their trends. Although these risk factors may cause stroke, they did not confound findings in this study.
Greater improvements in health behaviors and environmental factors, such as reduction of sodium intake29 for men in the full intervention community, may have contributed to the greater reduction of stroke rates. In the full intervention community, average dietary salt intake for men aged 40 to 59 years was 20 g/d in 1969 and 14 g/d in the period 1980 to 1983 according to a nutrition study.14 The 1980s estimate of salt intake was confirmed by 24-hour urine collection.30 Unfortunately, we have no comparable data on sodium intake in the minimal intervention community. However, a community-based sodium intervention in China demonstrated a larger reduction in sodium intake and systolic blood pressure levels in the intervention community than in the control community.31
The results of this study show this to be a good design to use to evaluate community trials intended to reduce cardiovascular diseases. First, the designated communities were culturally similar at baseline, and both suffered from the target disease (stroke). Blood pressure was sufficiently high in both communities that small reductions were likely to have pronounced effects on stroke. Second, penetration of this community-based program was largely under local control and did not cross from the full intervention community because of limited communication between communities. At the time, hypertension control for stroke prevention was seldom discussed nationally. In comparison, failure to demonstrate differential intervention versus control community effects in coronary heart disease risk reduction has been attributed to rapid dissemination of principles via a pervasive health communication network in Minnesota.12 32
Third, the intervention strategy was tailored to existing community and cultural conditions, and thus the intervention itself may be generalized in similar form to other communities of small population size. Fourth, the evaluation of disease end points was unobtrusive, highly specific and standardized, and independent of the intervention dose.
An inherent weakness is that evaluation information about blood pressure changes became biased when the intensity of intervention in the minimal intervention community was scaled back. Therefore, we drew cautious conclusions about the extent to which changes in stroke rate were mediated by changes in blood pressure levels. In the situation of intervention in these 2 communities, well-designed surveillance of blood pressure was probably impossible because introduction of systematic, community-wide blood pressure surveillance would have reintroduced free screening in the minimal intervention community. It is possible that reduced involvement in the minimal intervention community might have resulted in a decline in the detection of stroke cases during the 24 years. If so, the real difference in stroke incidence between the 2 communities would be even greater than reported here.
Other weaknesses are that the study was carried out in only 2 communities, and therefore the potential for chance results cannot be dismissed, and that the design of the study was not preplanned and lacked randomization. However, community-wide interventions are difficult to preplan and carry out from a logistic perspective, and this type of “weakness” may therefore be inherent in community intervention research.33
In conclusion, a community-based hypertension control program augmented the decline in stroke incidence and prevalence among men. We attribute the success of this community program to active participation of existing health resources in the detection and control of hypertension and to consistently high participation in blood pressure screenings, follow-up examinations, and community-wide health education activities. A mix of public health screening, media-based messages and education, and community involvement and activities is widely used in the United States10 32 and Finland7 11 to combat cardiovascular disease. Enhancements in these methods did not lead to accelerated disease reduction in Minnesota,12 although acceleration in the decline in ischemic heart disease mortality occurred in Finland.34 The present study has public health importance because it shows that a similar mix is effective in stroke reduction in a rural Japanese setting.
The authors thank Prof Aaron R. Folsom, University of Minnesota, for valuable comments on the manuscript. This study was supported in part by the Ministry of Health and Welfare, Tokyo, Japan (Research on Cardiovascular Diseases 4C-2, 1992–94).
- Received December 4, 1997.
- Revision received May 13, 1998.
- Accepted May 13, 1998.
- Copyright © 1998 by American Heart Association
Amery A, Birkenhager W, Brixko P, Bulpitt C, Clement D, Deruyttere M, De Schaepdryver A, Dollery C, Fagard R, Forette F, Forte J, Hamdy R, Henry JF, Joossens JV, Leonetti G, Lund-Johansen P, O’Malley K, Petrie J, Strasser T, Tuomilehto J, Williams B. Mortality and morbidity results from the European Working Party on High Blood Pressure in the Elderly Trial. Lancet. 1985;1:1349–1354.
Tuomilehto J, Nissinen A, Wolf E, Geboers J, Piha T, Puska P. Effectiveness of treatment with antihypertensive drugs and trends in mortality from stroke in the community. BMJ. 1985;291:857–861.
Luepker RV, Rastam L, Hannan PJ, Murray DM, Gray C, Baker WL, Crow R, Jacobs DR Jr, Pirie PL, Mascioli SR, Mittelmark MB, Blackburn H. Community education for cardiovascular disease prevention: morbidity and mortality results from the Minnesota Heart Health Program. Am J Epidemiol. 1996;144:351–362.
Association of Health and Welfare Statistics. Kousei no Shihyo (Trends for National Health, and Hygiene [in Japanese]). Tokyo, Japan: Kousaidou Press; 1992:410–411.
Shimamoto T, Komachi Y, Inada H, Doi M, Iso H, Sato S, Kitamura A, Iida M, Konishi M, Nakanishi N, Terao A, Naito Y, Kojima S. Trends for coronary heart disease and stroke and their risk factors in Japan. Circulation. 1989;79:503–515.
Rose GA, Blackburn H. Cardiovascular Survey Methods. Geneva, Switzerland: World Health Organization; 1968.
Kitamura A, Shimamoto T, Doi M, Iida M, Komachi Y. Secular trends in stroke incidence and risk factors for cerebral infarction in a rural Japanese population [in Japanese]. Jpn J Stroke. 1990;12:387–395.
Iso H, Kitamura A, Shimamoto T, Sankai T, Naito Y, Sato S, Kiyama M, Iida M, Komachi Y. Alcohol intake and the risk of cardiovascular disease in middle-aged Japanese men. Stroke. 1995;26:767–773.
The Ministry of Education Study Group, Okinaka S, Chairman. Characteristics of stroke in Japan and standardized criteria for stroke diagnosis [in Japanese]. Tokyo, Japan: The Ministry of Education; 1963.
Millikan CH. A report by an ad hoc committee established by the Advisory Council for the National Institute of Neurological Disease and Blindness, Public Health Service. A classification and outline of cerebrovascular diseases. Neurology. 1958;8:393–433.
Iso H, Jacobs DR Jr, Goldman L. Accuracy of death certificate of intracranial hemorrhage and nonhemorrhagic stroke: The Minnesota Heart Survey. Am J Epidemiol. 1990;132:993–998.
Corwin LI, Wolf PA, Kannel WB, Kannel WB, McNamara PM. Accuracy of death certificate of stroke: The Framingham Study. Stroke. 1982;13:818–821.
Komachi Y. Prevention of cardiovascular disease, a public health strategy for stroke epidemic (in Japanese). Tokyo, Japan: Suma Press; 1978:133–150.
Miyake S. Long-term hypertension control in a community: comparison of stroke incidence and hypertension control between participants and nonparticipants in health examinations (in Japanese). Jpn J Public Health. 1993;40:606–623.
Whisnant JP, Wiebers DO, O’Fallon WM. Sicks JD, Frye RL. A population-based model of risk factors for ischemic stroke: Rochester, Minnesota. Neurology. 1996;47:1420–1428.
Whisnant JP. Effectiveness versus efficacy of treatment of hypertension for stroke prevention. Neurology. 1996;46:301–307.
Doi M. Epidemiology of salt and hypertension. Cooperative study on the interrelation between dietary intake and hypertension. Annual Report of the Research on Cardiovascular Diseases 1984. Osaka, Japan: The National Cardiovascular Center; 1985:20.
Luepker RV, Murray DM, Jacobs DR, Mittelmark MB, Bracht N, Carlaw R, Crow R, Elmer P, Finnegan J, Folsom AR, Grimm R, Hannan PJ, Jeffery R, Lando H, McGovern P, Mullis R, Perry CL, Pechacek T, Pirie P, Sprafka JM, Weisbrod R, Blackburn H. Community education for cardiovascular disease prevention: risk factor changes in the Minnesota Heart Health Program. Am J Public Health. 1994;84:1383–1393.
Salonen JT, Kottke TE, Jacobs DR, Hannan PJ. Analysis of community-based cardiovascular disease prevention studies-evaluation issue in the North Karelia project and the Minnesota Heart Health Program. Int J Epidemiol.. 1986;15:176–182.
Tuomilehto J, Geboers J, Salonen JT, Nissinen A, Kuulasmaa K, Puska P. Decline in cardiovascular mortality in North Karelia and other parts of Finland. BMJ. 1986;293:1068–1071.