Trends in Blood Pressure, Hypertension Control, and Stroke Mortality: The Minnesota Heart Survey

Transcription

1 The American Journal of Medicine (2006) 119, CLINICAL RESEARCH STUDY Trends in Blood Pressure, Hypertension Control, and Stroke Mortality: The Minnesota Heart Survey Russell V. Luepker, MD, a Donna K. Arnett, PhD, b David R. Jacobs, Jr. PhD, a,c Susan J. Duval, PhD, a Aaron R. Folsom, MD, a Christopher Armstrong, PhD, a Henry Blackburn, MD a a University of Minnesota, School of Public Health, Division of Epidemiology and Community Health, Minneapolis, Minn; b University of Alabama, School of Public Health, Department of Epidemiology, Birmingham, Ala; c University of Oslo, Department of Nutrition, Oslo, Norway. ABSTRACT OBJECTIVE: The study s objective was to determine population trends in blood pressure, hypertension prevalence, hypertension control, and stroke mortality. METHODS: We performed population-based surveys of 2906 to 5630 adults from 1980 to 1982, 1985 to 1987, 1990 to 1992, 1995 to 1997, and 2000 to 2002, and stroke mortality from 1980 to 2002, in the Minneapolis/St Paul, Minn metropolitan area (2.63 million population according to the 2000 census). Randomly selected resident adults aged 25 to 74 years (n ) were each screened once. The main outcome measures were standardized measures of blood pressure, treatment and control of hypertension, and stroke mortality rates. RESULTS: The mean systolic blood pressure adjusted for age decreased in men ( 1.5 mm Hg [95% confidence interval 0.3 to 2.7], P.01) and women ( 1.8 mm Hg [95% confidence interval 0.5 to 3.0], P.001) from 1980 to 1982 and 2000 to The mean diastolic blood pressure was unchanged for men (0 mm Hg) and women ( 0.4 mm Hg, not significant). The proportion of the population taking antihypertensive medications decreased in the 1990s but returned to 1980s levels from 2000 to The use of other methods to decrease blood pressure (diet, exercise, and weight loss) peaked in the 1990 to 1992 survey and then decreased. Proportions of hypertensive patients in the aware, treated, and/or controlled categories leveled in the 1980s and 1990s, but improved substantially from 1995 to 1997 and 2000 to 2002 with blood pressure controlled at the less than 140 and/or 90 mm Hg criteria in 44% of the men and 55% of the women. Population mortality trends for stroke paralleled those for hypertension control. CONCLUSIONS: Population data beginning in 1980 to 1982 from the Minnesota Heart Survey indicate a leveling in the detection and control of hypertension in the 1990s followed by improvement from 2000 to Elsevier Inc. All rights reserved. KEYWORDS: Hypertension detection and control; Stroke; Blood pressure The research reported in this publication was supported by the National Heart, Lung, and Blood Institute, Bethesda, Md (RO1-HL23727). Requests for reprints should be addressed to Russell V. Luepker, MD, University of Minnesota, 1300 South Second Street, Suite 300, Minneapolis, MN address: luepker@epi.umn.edu. The detection, treatment, and control of hypertension are important national health objectives. 1,2 Epidemiologic data demonstrate increased cardiovascular disease risk associated with hypertension, and clinical trial data indicate that reduction of elevated blood pressure decreases cardiovascular disease events. 1,2 Observed declines in age-adjusted stroke and myocardial infarction rates in the United States are commonly attributed in part to improved detection and control of hypertension. 3,4 However, recent data from the National Health and Nutrition Examination Survey (NHANES) indicate that earlier trends of improving hypertension detection and control did not continue during the 1990s. 1,5,6 The Minnesota Heart Survey (MHS) is a populationbased study of trends in cardiovascular risk factors, morbidity, and mortality in a large urban area. 7 At 5-year /$ -see front matter 2006 Elsevier Inc. All rights reserved. doi: /j.amjmed

2 Luepker et al The Minnesota Heart Survey 43 intervals, surveys measure cardiovascular risk factors, including hypertension, in a sample of adults randomly selected from the population. Surveillance of stroke-related deaths is also conducted. This report evaluates trends in population blood pressure and hypertension detection, treatment, and control from 1980 to 2002, along with associated stroke mortality. METHODS Survey Population Cross-sectional risk factor and health behavior surveys included noninstitutionalized resident adults aged 25 to 74 years in the 7-county Minneapolis/St Paul metropolitan area from 1980 to 1982, 1985 to 1987, 1990 to 1992, 1995 to 1997, and 2000 to The Minneapolis/St Paul population is predominantly white and has slightly higher levels of education and employment than that of the U.S. population. The 2000 U.S. census counted 2.63 million inhabitants in the 7-county metropolitan area. CLINICAL SIGNIFICANCE Survey Sample Selection The methods for sample selection in MHS have been described in detail. 8,9 In summary, a two-stage cluster design used a sampling frame created from census maps. The 7-county metropolitan area was divided into 704 clusters of approximately 1000 households, and 40 clusters were randomly selected and used for each survey. A sample of households was randomly selected from each cluster. The sampling fraction (5%-10%) varied by survey because of budget considerations. The census maps were updated before each survey to account for new housing developments, and sampling was adjusted accordingly. In 1995, 4 clusters were added to represent new population growth. Once a house was selected, it was removed from sampling in future surveys. There were small differences in the sampling methods between surveys, which have been described. 7 After an introductory letter, selected households were visited by a trained interviewer who performed a household enumeration. All resident individuals were listed, and those eligible were selected. Data Collection The home interview solicited information on history of hypertension, medication use for hypertension, and other health indicators. The home visit was followed by a visit to a local survey center, usually in a nearby commercial or public building. At that site, further information was obtained including blood pressure, body height, and weight Hypertension detection, treatment, and control improved in the 1980s, leveled in the 1990s, and improved again in Stroke mortality trends parallel those of hypertension control. Although hypertension control has improved, many adults are still undetected, untreated, or under-treated. Among normotensives, mean blood pressure is rising associated with rising obesity. measurements. If an individual was unable or unwilling to come to the survey center, a second home visit was scheduled for the additional measurements. Blood pressure was measured with a random zero sphygmomanometer (Hawksley, West Sussex, United Kingdom) from 1980 to 1997 by technicians trained according to the method of Prineas and using an appropriately sized cuff. 10 After 5 minutes of seated rest, 2 blood pressure levels were taken 1 minute apart with the right arm at the level of the heart; the systolic and fifth phase diastolic pressures were recorded. The averages of the 2 systolic and fifth phase diastolic pressures were used for these analyses. Room temperature, time of day, pulse rate, recent smoking, and other characteristics known to affect blood pressure were also recorded as potential adjustment factors. Quality of blood pressure measures was evaluated with periodic calibration of sphygmomanometers, observation of technician procedures, and surveillance of blood pressure distribution values obtained by each technician using a technique to detect those whose measures frequently deviated from those of other technicians. During the 1995 to 1997 survey, a study comparing the random zero sphygmomanometer with the Dinamap (GE Medical Systems Information Technologies, Inc., Milwaukee, Wis) automated device was performed. All clinic subjects blood pressure levels were measured with both the RZ device and the Dinamap Monitor. The order was random with respect to device for the pairs of measurements. In the 1995 to 1997 survey period mean systolic blood pressure was mm Hg using the RZ and mm Hg using the Dinamap (N 4285). RZ fifth phase diastolic blood pressure (DBP) was mm Hg and mm Hg for the Dinamap (N 4285). In 2000 only the Dinamap was used. To correct blood pressure time trends for instrument change, we calibrated the Dinamap to the random zero by regressing the average random zero measures on the average Dinamap measures. The random zero device is known to read slightly, but systematically, lower than a standard mercury sphygmomanometer. 11 On the basis of this prior comparability study, 11 the random zero values were adjusted to the standard mercury sphygmomanometer, which is used in most settings. Participation Participation rates for a fully completed survey ranged from 64% to 69% of all eligible participants. In addition, more extensive data are available on the entire sample from those who participated in the home interview alone ( 80% par-

3 44 The American Journal of Medicine, Vol 119, No 1, January 2006 ticipation). Clinic respondents were slightly more likely than nonrespondents to be married (70% vs 61%), employed (77% vs 67%), better educated (some college: 68% vs 61%), and nonsmokers (smokers: 23% vs 32%). These patterns of nonresponse did not change over time. Stroke Mortality Surveillance Information on stroke mortality for the years 1980 to 2002 was obtained from the Minnesota Department of Health. For each year, all deaths among Minneapolis/St Paul residents aged 30 to 74 years attributed to stroke as the underlying cause were included. Because this period covered two revisions of the International Classification of Disease codes (International Classification of Disease Eighth Revision and Ninth Revision, codes ), adjustments were made for changes in the coding classifications. These are described elsewhere. 12 Population denominators were based on U.S. census data. Human Subjects Protection The study was approved by the Institutional Review Board of the University of Minnesota. Statistical Analysis All analyses were sex-specific to illuminate the male/female differences in blood pressure level detection and treatment. General linear modeling techniques using SAS PROC GLM were used to estimate and test for differences among agespecific mean levels of systolic blood pressure and diastolic blood pressure, and body mass index. The cluster design was included by adding the neighborhood clusters as a random effect. This allowed computation of an unbiased estimate of the variance of the sample means, resulting in an inflation of the variance by approximately 5%. Age-adjusted prevalences of hypertension (using two blood pressure cut-points: 160 and/or 95 mm Hg or 140 and/or 90 mm Hg and/or medication use) in the 5 surveys were obtained from SAS PROC MIXED and compared by pairwise t tests. Orthogonal polynomials were used to test linear and quadratic functions simultaneously. Hypertensive patients at each survey were further classified into 4 mutually exclusive groups: aware, treated, and controlled; aware, treated, and uncontrolled; aware, untreated, and uncontrolled; and unaware. For cost reasons, 40% of the clusters sampled in the 1995 to 1996 survey were not sampled in the 1996 to 1997 survey. To estimate the variance taking this into account, a bootstrap sample (n 2784) was chosen from this set of clusters in the 1995 to 1996 survey. Increased variance of time trend effects because of bootstrap sampling in 1996 and 1997 were estimated, and P values were corrected accordingly. SAS software was used for all analyses (SAS Institute Inc, Version 8.2. Cary, NC: SAS Institute Inc). RESULTS Tables 1 and 2 describe the survey sizes, mean age-adjusted systolic blood pressure, and fifth phase diastolic blood pressure by sex and survey year. Mean systolic blood pressure decreased significantly from 1980 to 1982 and 2000 to 2002 for both men ( 1.6 mm Hg, confidence interval [CI] 0.3 to 2.7, P.001 [linear]) and women ( 2.0 mm Hg, CI 0.5 to 3.0, P.001 [linear]). There were no significant changes in age-adjusted diastolic blood pressure during the 22 years for men or women. Women had consistently lower mean systolic blood pressure and diastolic blood pressure than men. The trends in systolic blood pressure and diastolic blood pressure by subgroups of treated and untreated hypertensive patients ( 140 and/or 90 mm Hg and/or current users of antihypertensive medications) and normotensive patients are also shown in Tables 1 and 2. Among those treated for high blood pressure, levels of systolic blood pressure and diastolic blood pressure decreased substantially and significantly for both men ( 6.5 mm Hg, CI 3.0 to 10.0, P.001 [linear]) and women ( 5.7 mm Hg, CI 2.2 to 8.9, P.001 [linear]) from 1980 to 1982 and 2000 to 2002 (Tables 1 and 2). The untreated hypertensive patients had variable mean systolic blood pressure and diastolic blood pressure with modest changes. However, systolic blood pressure in the normotensive population from 1980 to 1982 and 2000 to 2002 initially decreased and then increased among men ( mm Hg) and women ( mm Hg). A quadratic trend (a U-shape curve) was significant for both men and women (P.001). Diastolic blood pressure from 1980 to 1982 and 2000 to 2002 among normotensive patients increased in a linear fashion among men ( mm Hg, P.001) and women ( mm Hg, P.001). The age-adjusted population prevalence of hypertension by sex and survey is shown in Table 3. By using the lower blood pressure criteria of 140 mm Hg and/or 90 mm Hg or current medication use, the prevalence of hypertension in men decreased from 30.3% to 22.0% (P.01 [linear]). The decrease among women was similar, from 26.5% to 18.2% (P.001 [linear]). Despite a steady decline in overall prevalence for men and women, the percentage using antihypertensive medications followed a U-shape pattern. It was highest in the 1980s, decreased in the 1990s, and increased again from 2000 to Self-reporting hypertensive subjects described advice by their health care providers on other methods of controlling blood pressure such as exercise, weight loss, and diet. These peaked from 1990 to 1992 and decreased subsequently (Table 3). By using the usual categories of awareness, treatment, and control, an interesting pattern emerged (Figure 1). The proportion of men unaware of their high blood pressure remained stable at approximately 40% from 1980 to 1982 and 1995 to 1997, then decreased significantly to 21% in the 2000 to 2002 survey. The unaware were replaced by a shift to the proportion who were aware, treated, and controlled

4 Table 1 Trends in age-adjusted mean systolic blood pressure (mm Hg) by hypertension category Trends P Men N Mean SE N Mean SE N Mean SE N Mean SE N Mean SE L-trend* Q-trend All Normotensives * *** Treated HBP *** Untreated *** HBP Women All ** * ** *** Normotensives * *** Treated HBP * Untreated HBP *** * HBP high blood pressure; SE standard error. Linear trend P value testing H o slope 0. Quadratic trend P value testing H o quadratic slope 0. Orthogonal polynomials were used to test linear and quadratic trends simultaneously. Hypertension 140 and/or 90 mm Hg or on medications. *P.05. **P.01. ***P.001. Luepker et al The Minnesota Heart Survey Table 2 Trends in age-adjusted mean diastolic blood pressure (mm Hg) by hypertension category Trends P Men N Mean SE N Mean SE N Mean SE N Mean SE N Mean SE L-trend* Q-trend All ** *** *** Normotensives * *** *** *** Treated * Untreated *** Women All *** *** Normotensives *** *** *** Treated * *** Untreated * SE standard error. Linear trend P value testing H o slope 0. Quadratic trend P value testing H o quadratic slope 0. Orthogonal polynomials were used to test linear and quadratic trends simultaneously. Hypertension 140 and/or 90 mm Hg or on medications. *P.05. **P.01. ***P

5 46 The American Journal of Medicine, Vol 119, No 1, January 2006 Table 3 Trends in age-adjusted hypertension prevalence and treatment Hypertension L-trend Q-trend Men Prevalence* (%) On medications (%) Other methods Weight loss Dietary changes (less salt/ sugar, more fiber/water) Increase exercise NA NA Women Prevalence*% On medications (%) Other methods Weight loss Dietary changes (less salt/ sugar, more fiber/water) Increase exercise NA NA NA not asked. *Hypertension definition: 140 and/or 90 mm Hg or on medications as percentage of adult population. Percentage of total population on blood pressure medication with blood pressure data. Percentage of individuals self-reporting that they have been advised to lower blood pressure by the nonpharmaceutical methods listed. Categories are not mutually exclusive. Significance testing for differences between prior survey period. P.05. P.01. P.001. from 20% to 24% among men from 1980 to 1982 and 1995 to 1997, and increased to 44% in the most recent survey. Among women (Figure 1) proportions of aware, treatment, and control were always higher than for men. However, as in men, these patterns remained relatively stable from 1980 to 1982 and 1995 to 1997, and then improved in 2000 to 2002 with 55% of hypertensive women in the aware, treated, and controlled group. We also evaluated the higher standard of 160 mm Hg and/or 95 mm Hg, and found few unaware and untreated subjects ( 10%). From 1980 to 1982 and 2000 to 2002, age-adjusted mean body mass index (weight/height 2 ) increased steadily and significantly for both men and women. The increase was 26.4 kg/m 2 to 28.4 kg/m 2 (P.001) for men and was even greater for women from a mean of 25.0 kg/m 2 to 28.4 kg/m 2 (P.001). Stroke mortality trends for the Minneapolis/St Paul metropolitan area are shown in Figure 2. Age-adjusted stroke mortality rates began to level in the 1980s and stabilized in the 1990s for both men and women. Women had lower rates of stroke mortality than did men throughout the period. DISCUSSION The widespread detection, pharmacologic treatment, and control of hypertension are striking health accomplishments of the last 30 years. A wealth of laboratory, epidemiologic, and clinical trial data, followed by education campaigns by the National High Blood Pressure Education Program, industry, and many health organizations, have resulted in significant medical practice changes. 13,14 Antihypertensive medications are among the most commonly prescribed class of drugs. 13 Many observers attribute part of the large decline in mortality in age-adjusted stroke and coronary heart disease to the widespread and improved control of hypertension. 1,2 However, although most agree that significant progress has been made, there are concerns that this progress may not be sustained. 1,5,6 The MHS data provide population-based information on blood pressure levels and hypertension control in an urban setting. We believe that these data are particularly useful because of the large size of the population under observation (2.63 million, 2000 census) and the ability to evaluate trends using consistent, comprehensive sampling and measurement of population characteristics since Overall trends in mean systolic blood pressure and diastolic blood pressure in the 25- to 74-year-old population are small. Systolic blood pressure levels have decreased slightly in both men and women. Diastolic blood pressure has remained basically unchanged over the 22-year surveillance period. The decline in systolic blood pressure over time, observed by others in Belgium, Denmark, France, The Netherlands, and the United States, is not well explained Population changes in hypertension risk factors do not seem to account for these observations, because body mass is increasing while alcohol use and exercise levels are stable. 22 Changing population food intake with less salt, fat, and animal products may be responsible, but these are not well documented.

6 Luepker et al The Minnesota Heart Survey 47 Hypertension aware, treated, controlled - Men 140 and/or 90 mmhg or on medications ATC ATU AUU Unaware Hypertension aware, treated, controlled - Women 140 and/or 90 mmhg or on medications ATC ATU AUU Unaware Figure 1 Hypertensive patients in the categories of aware, treated, and controlled, based on 140 and/or 90 mm Hg or medication use for men and women. ATC aware, treated, controlled; ATU aware, treated, uncontrolled; AUU aware, untreated, uncontrolled. Separation of hypertensives from so-called normotensives is revealing in understanding these overall population trends. Treatment of high blood pressure is both common and very effective in decreasing systolic blood pressure and diastolic blood pressure levels. More hypertensive patients are treated and better controlled as evidenced by declining blood pressures among members of this group. At the same time, both systolic blood pressure and diastolic blood pressure among the normotensive group have increased. This is concordant with increasing obesity in the population. The combination of these 2 divergent trends is a modest decrease in population systolic blood pressure and no change in diastolic blood pressure. The prevalence of hypertension, as defined by blood pressure of 140 and/or 90 mm Hg or taking antihypertensive medications, leveled during the 1990s but decreased significantly for both men and women in 2000 to Medication use, high in the 1980s, decreased during the 1990s but increased again in the most recent survey. Selfreported nonpharmacologic recommendations to decrease blood pressure including weight loss, diet, and exercise peaked from 1990 to 1992 and then declined. Recent data from NHANES representing the entire U.S. population from 1980 to 2000 finds a modest increase in hypertension prevalence during the decade of the 1990s. 5,6 However, the NHANES change for non-hispanic whites (similar to the Minnesota population) was not significant. The NHANES population covers a larger age window than MHS from 18 years to more than 74 years. NHANES has not yet published data extending into the 21st century, for which we observed declines in hypertension prevalence and improved treatment. Among the striking observations presented here are the dramatic increases in awareness, treatment, and control of hypertension. We find few individuals at the higher 160 and/or 95 mm Hg cut-point in the population. It is difficult to see how additional progress could be made in the detection and treatment of these individuals. Severe hypertension may be disappearing, or those who have this condition are under effective treatment. 5 By using the usual cut-point ( 140 and/or 90 mm Hg), there is significant improvement among those aware, treated, and controlled in the most recent survey. From 2000 to 2002, 44% of men and 55% of women were treated and controlled. At the same time, the number of unaware patients among those with hypertension decreased for both men and women. These data contrast with the 1999 NHANES information in which 37% of men and 31% of women with hypertension were treated and controlled. 5 It is not surprising that high blood pressure detection and control is higher in Minnesota. The state has among the lowest cardiovascular disease rates in the nation and the highest rates of health insurance coverage. Minnesota has been a leader in cardiovascular disease prevention and treatment. 23 In addition, the major health insurers in the Minneapolis/St Paul metropolitan area launched a blood pressure detection and treatment campaign in the late 1990s based on prior MHS data. The campaign was combined with increased scrutiny of clinic practices by health maintenance organizations and insurers for quality of hypertension care. In most regions of the world as in Minnesota, stroke mortality has decreased over the past several decades Some of this improvement may be the result of improved acute stroke care, but most is thought to be the result of improved detection and treatment of hypertension. 30 However, stroke mortality leveled in recent years. 24,25,28,29 Some authors speculate that the maximum benefits of hypertension treatment are already attained. In Rochester, Minn, stroke incidence declined steadily from 1950 to 1979 but increased in the early 1980s for both men and women. 31 This occurred despite improved blood pressure control in Rochester. 31 Those investigators offered several possible explanations. They recognized that the advent of widespread diagnostic imaging with computed tomography and magnetic resonance imaging began during the period when stroke incidence began to increase. More sensitive tests may increase case finding. The decreasing case fatality rate, in the setting of no dramatic changes in care of stroke victims,

7 48 The American Journal of Medicine, Vol 119, No 1, January 2006 R a t e / 1 0 0, Stroke Mortality Minneapolis - St. Pa ul Men Wome n Year Figure 2 Curve smoothed to 3 ye ar runni n g average Adjusted to 2000 census Age-adjusted fatal stroke trends, , in the Minneapolis/St Paul metropolitan area for men and women. suggested that greater numbers of milder cases were detected, thus modifying the rate. 32,33 These and other investigators also hypothesize that better survival from ischemic heart disease, with increasing prevalence of atrial fibrillation and chronic coronary heart disease, may play a role in increasing stroke incidence by adding to embolic events The strengths of the Minnesota Heart Survey data include the large size of the population under observation, the longitudinal data collection, and the comprehensiveness of the information. The weaknesses are found in nonparticipation rates and the lack of a substantial minority population. The changing of measurement methods for blood pressure in the most recent surveys are also a potential weakness despite the overlap studies and corrective adjustments. In summary, blood pressure and stroke trends in the population are changing. Hypertension detection, treatment, and control improved dramatically in the 1970s. 34 After a plateau in the late 1980s and 1990s, hypertension detection, treatment, and control again improved in recent years. Treatment with medications is widespread and apparently more effective, whereas nonpharmacologic treatments are in decline. Stroke mortality trends seem to parallel the control of hypertension. At the same time, average blood pressure levels among normotensives are increasing. This may be, in part, attributable to increasing obesity or other environmental factors. Despite the improvements observed, hypertension and its cardiovascular complications continue to present major medical and public health challenges. References 1. National Heart, Lung, and Blood Institute. The Sixth Report of the Joint National Committee on Detection, Evaluation, and Treatment of High Blood Pressure. Bethesda, MD: U.S. Department of Health and Human Services, National Institutes of Health; NIH publication no ; Chobanian AV, Bakris GL, Black HR, et al. The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: The JNC 7 Report. Hypertension. 2003;42: Lenfant C, Roccella EJ. Trends in hypertension control in the United States. Chest. 1984;86: Lever AF, Ramsay LE. Treatment of hypertension in the elderly. J Hypertens. 1995;13: Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of hypertension in the United States, JAMA. 2003; 290: Fields LE, Burt VL, Jeffery AC, et al. The burden of adult hypertension in the United States 1999 to 2000: a rising tide. Hypertension. 2004;44: Gillum RF, Prineas RJ, Luepker RV, et al. Decline in coronary deaths: a search for explanation. Minn Med. 1982;65: Luepker RV, Jacobs DR Jr, Gillum RF, et al. Population risk of cardiovascular disease: the Minnesota Heart Survey. J Chronic Dis. 1985;38: Luepker RV, Jacobs DR Jr, Folsom AR, et al. Cardiovascular risk factor change to : the Minnesota Heart Survey. J Clin Epidemiol. 1988;41: Prineas RJ. Blood Pressure Sounds: The Measurement and Meaning. Miami, FL: Gamma Medical Products Corporation; de Gaudemaris R, Folsom AR, Prineas RJ, Luepker RV. The random-zero versus the standard mercury sphygmomanometer: a systematic blood pressure difference. Am J Epidemiol. 1985;121: Shahar E, McGovern PG, Pankow JS, et al. Stroke rates during the 1980: the Minnesota Stroke Survey. Stroke. 1997;28:

8 Luepker et al The Minnesota Heart Survey National High Blood Pressure Education Program Working Group. National High Blood Pressure Education Program working group report on hypertension in the elderly. Hypertension. 1994;23: National High Blood Pressure Education Program. National High Blood Pressure Education Program Working Group Report on Primary Prevention of Hypertension. Bethesda, MD: U.S. Department of Health and Human Services, National Heart, Lung, and Blood Institute; NIH publication no ; Sjøl A, Korsgård Thomsen K, Schroll M. Secular trends in blood pressure levels in Denmark Int J Epidemiol. 1998;27: Kumanyika SK, Landis JR, Matthews-Cook YL, et al. Systolic blood pressure trends in US adults between 1960 and 1980: influence of antihypertensive drug therapy. Am J Epidemiol. 1998;148: Fortmann SP, Winkleby MA, Flora JA, et al. Effect of long-term community health education on blood pressure and hypertension control: the Stanford Five-City Project. Am J Epidemiol. 1990;132: van Leer EM, Seidell JC, Kromhout D. Levels and trends in blood pressure and prevalence and treatment of hypertension in the Netherlands, Am J Prev Med. 1994;10: Garraway WM, Whisnant JP. The changing pattern of hypertension and the declining incidence of stroke. JAMA. 1987;258: Marques-Vidal P, Ruidavets J-B, Cambou J-P, Ferrières J. Trends in hypertension prevalence and management in Southwestern France, J Clin Epidemiol. 2000;53: De Henauw S, De Bacquer D, Fonteyne W, et al. Trends in the prevalence, detection, treatment and control of arterial hypertension in the Belgian adult population. J Hypertens. 1998;16: Arnett DK, McGovern PG, Jacobs DR, et al. Fifteen-year trends in cardiovascular risk factors ( through ): the Minnesota Heart Survey. Am J Epidemiol. 2002;156: McGovern PG, Pankow JS, Shahar E, et al. Recent trends in acute coronary heart disease mortality, morbidity, medical care and risk factors. New Engl J Med. 1996;334: Keli SO, Feskens EJM, Naarden EN, Kromhout D. Decrease in stroke mortality in the Netherlands Antilles and the Netherlands: evidence for a recent slow down. The role of blood pressure, alcohol and diet. Thesis. 1995; Bonneux L, Looman CWN, Barendregt JJ, Van der Maas PJ. Regression analysis of recent changes in cardiovascular morbidity and mortality in the Netherlands. BMJ. 1997;314: Gillum RF. Cerebrovascular disease morbidity in the United States, : age, sex, region, and vascular surgery. Stroke. 1986;17: Brown RD Jr, Whisnant JP, Sicks JD, et al. Stroke incidence, prevalence, and survival: secular trends in Rochester, Minnesota, through Stroke. 1996;27: Harmsen P, Tsipogianni A, Wilhelmsen L. Stroke incidence rates were unchanged, while fatality rates declined, during in Göteborg, Sweden. Stroke. 1992;23: Cooper R, Sempos C, Hsieh SC, Kovar MG. Slowdown in the decline of stroke mortality in the United States, Stroke. 1990;21: Jacobs DR, McGovern PG, Blackburn H. The U.S. decline in stroke mortality: what does ecological analysis tell us? Am J Public Health. 1992;82: Whisnant JP, O Fallon WM, Sicks J, Ingall T. Stroke incidence with hypertension and ischemic heart disease in Rochester, Minnesota. Ann Epidemiol. 1993;3: Broderick JP, Phillips SJ, Whisnant JP, et al. Incidence rates of stroke in the eighties: the end of the decline in stroke? Stroke. 1989;20: Higgins M, Thom T. Trends in stroke risk factors in the United States. Ann Epidemiol. 1993;3: Folsom AR, Luepker RV, Gillum RF, et al. Improvement in hypertension detection and control from to : the Minnesota Heart Survey experience. JAMA. 1983;250: