zyxwvutsrqp zyxwvutsrqpo zyxwvutsrqpon zyxwvutsrq journal of Znternal Medicine 199 3 : 2 3 3 : 1 5 5-1 6 3 Stroke morbidity in patients treated for hypertension-The Skaraborg Hypertension Project u. LINDBLAD, From the Ikpartment L. OJ RASTAM & J. RANSTAM Community Health Sciences. University OJ zyxw zyx Lund. MrrlrnB. Sweden zyxw J (Department of Community Health Sciences, University of Lund. Malmo. Sweden). Stroke morbidity in patients treated for hypertension-The Skaraborg Hypertension Project. journal 01Internal Medicine 199 3 : 233: 155-163. Abstract. Iindblad U. RBstam L, Ranstam Stroke incidence was analysed in a Swedish cohort of male (n = 1428) and female (n = 18 12) hypertensive patients in comparison with age- and sex-matched population controls ( 1 : 1 ) and with normotensive untreated subjects (1249 men and 1247 women). Mean follow-up was 8.3 years. Patients were aged 40-69 at the start of follow-up in 1977-1981. Relative risks (95%confidence interval [CI]) for stroke morbidity were 1.63 (1.16, 2.29) for men and 1.40 (0.94, 2.09) for women compared to population controls. Corresponding figures for stroke mortality were 1.96 (1.01-3.82) and 1.48 (0.71-3.06). Compared to the normotensive sample with adjustment for smoking and body mass index relative risks for stroke morbidity were 3.07 ( 1 . 9 6 4 . 8 0 )for men and 2.56 (1.46-4.51) for women. The prognosis of treated hypertension with respect to stroke is better than anticipated from previous studies, a fact that should be considered when treatment guidelines are developed. Keywords: cohort study, hypertension, mortality, primary health care, primary prevention, stroke. Introduction Prospective studies have identified high blood pressure as an important risk factor for cardiovascular disease [ 1-31. Numerous clinical trials also have demonstrated that pharmacological treatment of hypertension can prevent, or at least postpone stroke 14-71. During the last few decades much effort has been devoted to improving population control of hypertension [8-101 and today most of the hypertensive subjects in the population are also treated with blood-pressure-lowering medication [ 111. Despite this, the residual risk for complications is claimed to be substantial in treated subjects [12-151. Different explanations for this persisting excess risk have been suggested, i.e. insufficiently lowered blood pressure [ 151, problems with study design or statistical power [ 1h, 171. or other interacting risk factors [ 181. Metabolic side-effects of the hypertension medication have also been claimed to be contributing factors (16. 191. It is claimed both that blood pressure should be lowered more effectively, and that newer and potentially more effective (and also more expensive) drugs need to be used [ 15. 201. Thus, this demand for improved hypertension treatment may have economic consequences both for the patient and for society as a whole [21]. A solid foundation is needed when changes in the routines for hypertension control are suggested. Not least, studies of the prognosis in patients who are representative for the population are needed. In this project a large representative cohort of male and female hypertensive patients treated in primary care was followed for a mean follow-up period of 8.3 years. The aim of this study was to analyse the relative risks, as an aspect of public health, for acute stroke morbidity and stroke mortality along with total mortality compared both to population controls and to a normotensive reference group. As the treatment was aimed at primary prevention, acute stroke morbidity in this study refers to incidence of the first event during follow-up. 155 zyxwvutsrqp zyxwvutsrqp zyxwvutsrqp zyxwvutsrq zyxwvutsr 156 U. L I N D B L A D e&al. Subjects and methods The Skaraborg Hypertension Project The Skaraborg Hypertension Project was launched in 19 77 [ 101with the aim of improving hypertension control. The project included a medical care programme with guidelines for the detection, diagnosis, clinical work-up, treatment and follow-up of 40-69year-old hypertensive patients. This programme was implemented in half of Skaraborg county (the county being divided in half geographically) the other half serving as a reference area. The target population at the start of the project was 4 6 2 3 4 in the study area and 4 2 6 4 0 in the reference area. Hypertension out-patient clinics were organized in the six primary care settings of the study area and cooperation was established with private practitioners (only a few active in the area) and also with major occupational health clinics. According to the programme guidelines, blood pressure limits were age-dependent (Table 1). A patient was considered hypertensive and entered into the hypertension register when the systolic and/or the diastolic blood pressure had exceeded defined limits for high blood pressure on three different occasions and/or when the patient had ongoing pharmacological treatment. The first blood pressure in previously untreated patients could be casual (e.g. when measured at screening) while the second and third always had to be measured after 5 min of rest in the supine position. The goal for the pharmacological treatment was to reach the blood pressure level defined as normal for both systolic and diastolic blood pressure (Table 2). At registration a patient work-up was performed. Data on laboratory tests, body weight and body height were entered into a standardized form. Patients were also interviewed about smoking habits. It was recommended that pharmacological treatment should follow a stepped-care protocol with a beta-blocker (initially propranolol and from 19 79 metoprolol) or a diuretic (bendroflumethiazide) as a first step and the combination of the two as a second. As a third step a vasodilator (hydralazine) could be added. However, the use of this schedule was only a recommendation and physicians were free to prescribe any medications they found to be suitable for their patients. At the end of the first 5-year period (1981) 27.3%were treated with a beta-blocker only, 23.4% of the patients were treated with a thiazide diuretic only, and 21.2% with a combination of the two. The triple combination, also including hydralazine. was used in 7.6% In 6.7%a combination of hydralazine with a beta-blocker (4.9%),a thiazide diuretic (1.6%) or hydralazine alone (0.2%) was used. Of the remainder, 8.4% had treatment including other medications such as spironolactone, furosemide and a-metydopa, while 5.4% had no pharmacological treatment. The main objectives of the project were to evaluate zyxwvutsrqpo zyxwvutsrq Table 1. Normal. borderline and high blood pressure limits. A blood pressure is defined as normal if both systolic and diastolic blood pressures are lower than the limits. and high if one or both limits are exceeded Blood pressure level (mmHg) Age (years) Normal Borderline High 40-60 > 60 < 160/<95 < 170/< 105 <-> <-> > 170/> 105 > 180/> 110 zyxw Table 2. Male and female incidence of Hrst stroke (fatal and non-fatal combined) in the three samples. Relative Asks with the population census sample as reference Incidence rate (per 1000 person years) Relative dsk (with 95% CI) Sample No. of subjects No. of Occurrences Men Hypertensive Normotensive Census 1428 1249 1428 87 32 53 7.4 2.5 4.5 1.63 (1.16, 2.29) 0.51 ( 0 . 3 2 . O.H1) 1.00 0.005 0.004 in12 1247 1812 58 3.9 1.4 2.8 1.40 (0.94. 2.09) 0 . 6 0 (0.34. 1.07) 0.096 0.083 P Women Hypertensive Normotensive Census CI = confidence interval. 18 41 1.00 STROKE MORBIDITY 157 zyxwvutsrq zy zyxwvu whether it was possible to achieve control of hypertension in the population and if this could be done cost-effectively without harmful effects on the patients [ 10). Along with the follow-up of blood pressure reduction, the programme included an evaluation of how well it could postpone hypertensive complications [ 101-this included data obtained during the 5-year trial and included patients who were registered during the years 1977-1 98 1. As part of the evaluation a baseline population study was undertaken in 1977 with a participation rate of 70%. and a terminal population study in 1982. Randomly selected subjects were invited to see a nurse for a health check-up including measurement of blood pressure, body weight and body length. They also completed a questionnaire, including the question: 'Are you taking any drugs for high blood pressure ? '. Smoking habits were also inquired about. the next best match. An index subject could be selected as a control and was thus not replaced. Consequently, the sample was representative of the total population from which the hypertensive patients were initially recruited. This sample contributed with a total of 26742 person years in the study. ( 3 )The normotensive sample. All 2496 subjects ( 1 249 men and 1247 women) who aged 40-70 years participated in the baseline population study in 1977, had a blood pressure corresponding to what was defined as normal in the project (Table l), and responded that they did not take blood-pressurelowering medications. Mean age at entry was 52.3 years and this sample made up 25 861 person years in the study. End-point registration Study populations End-points in this study were acute stroke (fatal or non-fatal) and all causes mortality. Acute stroke was defined according to criteria suggested by WHO [22]. If death occurred within 28 days of the onset of an acute stroke, the event was classified as fatal. Only the first non-fatal stroke that occurred during surveillance was used, together with all fatal events. The procedure of end-point surveillance and record validation has been described in detail elsewhere [22]. In summary, information about possible events was achieved from two sources. The Skaraborg County Council keeps a computerized in-patient register covering all discharges from the three acutecare hospitals in the county. From this register, all episodes with diagnoses most likely to contain occurrence of acute stroke (ED-codes 4 3 0 4 3 8 ) were extracted for subjects in the three study samples. In Sweden the underlying cause of death is certified by a physician. The death certificate is sent to the National Mortality Register where a coding staff decides the underlying cause of death according to a detailed protocol. These data were kept in a register which is updated annually. Information about all deaths for subjects in the three study samples were collected from this register. All extracted information from the three samples, where hospital discharge notes were available, was validated by a n experienced physician who was not otherwise involved in the project [22]. Data from occurrences fulfilling criteria for acute stroke were used for the analysis. In other cases, the underlying zyxwvutsrqpon This study includes data on three study samples: ( 1 ) The hypertensive sample. Three-thousand, twohundred-and-forty hypertensive patients ( 1428 men and 1812 women), registered at any of the hypertensive out-patient clinics during the first 5 years of the project (January 1, 1977-December 31, 1981) were included in the study. This corresponds to 7% of the target population. However, in one of the six municipalities in the study area, Skara, as many as 13% (n = 778) of the target population registered. To be eligible for registration, a person had to be 40-69 years old and fulfil the diagnostic criteria already mentioned. On admittance 2367 patients had ongoing treatment (1023 men and 1344 women) while 87 3 patients (405 men and 468 women) were previously untreated. mean age at entry was 56.8 years and the hypertensives contributed with a total of 2 6 8 79 person years. ( 2 ) The population census sample. One individual control to each index subject was randomly selected from the population census valid on 1 January of the year when the index subject entered the hypertension register. They were matched by age, gender and residency (the municipality where the index subject resided at the time of registration). In case a selected subject died between 1 January and the day when its index subject was registered, he/she was replaced by 158 zyxwvutsrq zyxwvutsrq zyx zyxwvuts zyxwvutsr U. LINDBLAD et al. cause of death in the National Mortality Register was accepted without any further validation. sample on the other hand, had lower relative risks, but also in that case, the difference was only statistically significant in men. Statistical methods Body mass index was calculated by the formula weight (kg)/length2 (m7. Disease-free survival (i.e. time to first stroke) was analysed using the KaplanMeier method [23]. Differences in survival were tested with a generalized Wilcoxon statistics. Multivariate analyses were performed using Cox's proportional hazards model [24]. Statistical significance was presumed when P < 0.05. 94 I .......... ..... Population Results b lncidence offirst stroke As seen in Table 2 . a total of 87 men and 58 women in the hypertensive sample had a first acute stroke during follow-up, the male rate being twice that of the female rate. In the population sample, the sexspecific rates were lower, but these in turn were higher than those in the normotensive sample. Within the samples, the incidence of first stroke was invariably higher in men than in women. Figure 1 presents life tables for combined fatal and non-fatal acute stroke in men and Fig. 2 shows data for women. From the start of follow-up the sexspecific incidence of first stroke in the hypertensive sample was higher than that in the population sample, who in turn invariably had a higher incidence of first stroke than the normotensive sample. Using the population sample as reference in a Cox model, the relative risk for stroke was significantly higher for the hypertensive men but not for the hypertensive women (Table 2). The normotensive 900 2700 1800 Fig. 1. Stroke-free survival in men (Kaplan-Meier). . 96 No. of subjects No. of deaths Men Hypertensive Normotensive Census 142H 1249 1428 26 6 13 Population Normotensives 0 900 2700 1800 1812 1247 1812 I8 1 CI = confidence interval. 12 3600 Follow-up (days) Fig. 2. Stroke-free survival in women (Kaplan-Meier). Mortality rate (per 1000 person years) zyxw Relative risk (with 95% CI) P 1.1 0.5 1.96 (1.01. 3.X2) 0.38 (0.13. 1 . 0 5 ) 1 .oo 0.047 0.063 1.2 0.1 0.8 1.48 (0.71, 3.06) 0 . 1 0 (0.01. 0.80) 1.00 0.296 0.030 2.2 Women Hypertensive Normotensive Census . . . . . . . . . . . Hypertensives Table 3. Male and female stroke mortality in the three samples. Relative risks with the population census sample as reference Sample 36 3 Follow-up (days) zyx zyxwvuts S T R O K E MORBIDITY Using the normotensive sample as reference, there was a statistically significant persisting over-risk in both hypertensive men and women. The relative risk for acute stroke in the hypertensive sample was 3.19 (95% CI: 2.08. 4.90; P < 0.001) for men and 2.33 (95% CI: 1.35. 4.03: P = 0.003) for women. Adjustment of differences between the samples in smoking habits and body mass index slightly decreased the sex difference. (3.07. 95% CI: 1.96. 4.80; P < 0.001 for men and 2.56, 95% CI: 1.46. 4.5 1 ; P = 0.00 1 for women). The aetiologic fraction, calculated from these data, was 25% for men and 17%, for women. The case fatality rate, calculated as the proportion of all first events leading to death within 28 days, was 23% in hypertensive men and 26% in hypertensive women ; corresponding rates for the population sample were 17% in men and 22% in women. In the normotensive sample the case fatality rate for the two genders combined was 10%(the genders in this sample were analysed together because of the few events). 159 P = 0.021), but neither hypertensive men (relative risk 0.90, 95% CI: 0.73, 1.10; P = 0.296), nor hypertensive or normotensive women (relative risk 0.94, 95% CI: 0.74. 1.20; P = 0.619 and relative risk 0.78, 95% CI: 0.59, 1.04; P = 0.092 respectively) differed from the population sample. The differences between the samples were observed from the start of follow-up. With the normotensive sample as reference, relative risk for total mortality was 1.16 (95% CI: 0.93, 1.44; P = 0.19) for men and 1.20 (95% CI: 0.90. 1.61; P = 0.21) for women. Adjustment for smoking and body mass index had no significant impact on these results. zyx zyxwvutsrqp zyxwv Stroke mortality The total number of deaths due to acute stroke was small with incidence rates below 2.2/1000 patientyears in all samples (Table 3). Notably, only one normotensive woman and six normotensive men died from stroke during follow-up. The relative risk of dying from stroke was higher than that of the population sample in hypertensive men and lower in normotensive women, but confidence intervals were wide. With the small number of stroke deaths in the normotensive sample, it was not regarded as worthwhile to calculate relative risk with that sample as reference, as confidence intervals would be too wide to enable interpretation. All causes mortaliQj Acute stroke contributed only a minor proportion to the total mortality in the population sample; 7% in men and 9% in women. Corresponding proportions were 15% (men) and 14% (women) in the hypertensive sample, and 4% and 1%. respectively, in the normotensive sample. The ACM (all causes mortality) was higher among men than among women in all three samples. It was statistically significantly lower among normotensive men with a relative risk of 0.77 (95% CI: 0.62.0.96; Skara To discover whether these results could be considered general for all hypertensives, the patients residing in the Skara municipality (n = 778) were compared with all other patients in the hypertensive sample ( n = 2462). After adjusting data for differences in age, smoking habits and body mass index there were no statistically significant differences in the incidence of first stroke, or in all causes mortality either for men or for women. The number of fatal strokes in the Skara group was too small to permit statistical analysis. Discussion This study, based upon up to 10 years of follow-up of a large number of hypertensive patients treated in primary care, shows that the prognosis of this very common category of patients is reasonably good with all causes mortality at the same level as in the total population, and a slightly increased risk of strokestatistically significant only for men. When compared with normotensive untreated subjects the risk of stroke was about three times higher, but again, total mortality was not significantly increased. Previous studies have demonstrated that this medical care programme was feasible [25]. Mean blood pressure in all treated patients aged 40-69 in the study area decreased compared with the control area, and a sub-group analysis of the terminal population study revealed a statistically significant difference of 8 / 5 mmHg between patients treated at the hypertension clinics and those treated in the control area [26]. At the last annual check-up in 1981,63%of the patients had normal blood pressure, 160 zyxwvutsrqpo zyxwvuts zyxwvutsrqpon U. LINDBLAD et al. 24% had borderline blood pressure and 13% had high blood pressure according to the definitions of high and normal blood pressure in the project (see above). This improved quality of care was followed by a statistically significant trend of decreasing stroke incidence in favour of the study area [27]. For the interpretation of these results, however, it should be born in mind that subjects attending health controls tend to be more concerned about their health, have healthier life-styles and suffer less from diseases than non-attenders [28-301. Specifically, this was true for those subjects from among whom the normotensive sample was selected [29]. Thus, they were not entirely representative of the total population of normotensive untreated subjects and a lower ACM should be expected. Conversely, the population sample was representative of the unselected population ; therefore, this sample would be exposed to a variety of risk factors such as smoking, abuse of alcohol and, not least, hypertension. The prevalence of chronic disease was probably also higher than in the normotensive sample. There would also be differences in socioeconomic characteristics [29, 3 I]. Deciding which of these two control groups to use for reference is, therefore, of the utmost importance for the interpretation of the results. Comparison with a normotensive sample identified at screening would imply overestimation of the impact of the factor studied. An unselected population sample does not have these drawbacks, but on the other hand, one cannot control the potential impact from other risk factors or co-morbid disease. From the baseline population study the prevalence of hypertension in the target population was estimated to be 1 5 1 6 % (including both patients with ongoing treatment at entry and subjects without treatment on admission but fulfilling the diagnostic criteria stated above) [ 111. The 3240 hypertensive patients aged 40-69 included in the hypertensive sample correspond to 7% of the target population (i.e. about half of the eligible patients in this age group in the study area). Also, the hypertensive sample may to some extent be selected, as patients with good socio-economic status and those interested in their health may be more likely to accept treatment and long-term follow-up. Moreover, patients with severe disease would be less likely to be treated in primary health care. This may explain the tendency towards lower total mortality among the hypertensives compared with the population sample. To estimate the impact of this potential bias, the subsample of hypertensives residing in Skara (where practically all hypertensive subjects in the population were included) was compared with the other hypertensive patients; admittedly, the statistical power in this comparison was low. However, only negligible differences were identified and for the purpose of this study the hypertensive sample was agreeably r e p resentative of all hypertensives in these ages. However, there was a n under-representation of older subjects and the results are not applicable to subjects older than 69 years at the start of follow-up. Compared to present guidelines for the treatment of hypertension the blood pressure levels used for the diagnosis of hypertension and the treatment goals was higher and the prevalence of hypertension was accordingly low. However, these definitions are equal to those that were valid in Sweden during the late 1970s and the early 1980s. The risk of stroke is claimed to be higher in men [32, 331, but the Framingham data does not support this [34]. While, in this study, incidence of first stroke was higher in men than in women, relative risk was of the same magnitude in the two genders. The small observed difference might be explained by the fact that hypertension control in women improved more during follow-up than in men [26]. Thus, relative risks were lower than expected [ 12-14] and also attributable risks were lower than those previously reported [35, 361. A previous Swedish study showed considerably higher relative risks when hypertensive subjects were compared with normotensive controls [ 121. However, diagnostic criteria were wide and differed from those of the present study and the confidence intervals were also wide. It has also been argued that the study design was not appropriate [ 171. In another publication produced from that same study, new cases of stroke over a period of 1 0 years, were compared to normotensive controls with regard to previous exposure to high blood pressure [ 3 71. The relative odds of stroke in men with hypertension was estimated to be 12.3 while in females stroke was not analysed separately. Controls were identified at the end of the study period and, therefore, the possible impact of selective mortality in that group was not accounted for. Their blood pressure was also recorded post hoc while the blood pressure of the cases was extracted from medical records. As the controls consisted of survivors the odds ratios may be overestimated-xplaining the high risk found in men. zyxwvutsr zyxwvutsrq zyxwvu STROKE MORBIDITY Mortality in 3 7 8 3 patients at the Glasgow Blood Pressure Clinic with non-malignant hypertension, followed between 1968 and 1983 for an average of 6.5 years, was compared with three reference groups : the general population, a group of screening participants and with a group of hypertensive subjects identified in screening and followed-up at a blood pressure clinic based in general practice [ 131. Significantly higher mortality was found in both genders in comparison with each of these three control groups. Compared to the general population ACM was 2.3 times the general population for men at 10 years and 2.4 times for women. When compared to hypertensives treated in general practice, the difference in mortality was still highly significant: in turn, the mortality of that control group did not differ from their population controls. These data seem consistent with our study. The most probable explanation for the high persisting mortality of the Glasgow patients would be heavier aggregation of risk factors and cardiovascular complications. This view is further supported by high mortality rates during 10 years of follow-up and the high proportion of cardiovascular deaths, (74.9%in men and 74.2% in women). The fact that the Glasgow patients were slightly younger might also have contributed to the differing results, as relative risk decreases with increasing age [ 131. Furthermore, in another hospital-based hypertension clinic mortality for both men and women was studied over a 10-year period with a mean follow-up time of 6.1 years [38]. Total mortality and death from myocardial infarction were similar to that in the local population while stroke mortality was about twice that level. This is in contrast to the Glasgow study but in accordance with our results. Neither of these two British studies, however, included non-fatal end-points. A third study often cited as support for high residual risk associated with treated hypertension is the Primary Prevention Trial in Goteborg [14]. Compared to expected incidences of cardiovascular complications (total mortality, fatal and non-fatal CHD and stroke), calculated from an observational study of middle-aged men where treatment was kept at a minimum, the incidence rates were significantly reduced for treated hypertensives [ 141. However. comparedtonormotensive population controls, cardiovascular complications were still high among the hypertensives [ 141. This control group was identified in screening and in addition to being normotensive, 161 they were also non-smokers and non-hypercholesterolaemic. When a 'low-risk' fraction of the treated hypertensives (non-smoking: non-hypercholesterolaemic : no target organ manifestation : no previous CVD at onset: initial diastolic blood pressure below 120 mmHg) were compared with these normotensive subjects, no difference was revealed in the 10-year cumulative incidence of CVD morbidity [ 141. Thus. treated hypertension per se had no impact on longterm prognosis when no other risk factors were present. Compared with present guidelines, the goal for treatment in the Skaraborg Hypertension Project was quite modest-at least for patients in the 4060-years-old age group [ 101. During the latter half of the 1 9 8 0 ~guidelines ~ recommended that diastolic blood pressure should be lowered below 90 mmHg in all ages whereas systolic blood pressure is regarded to be a less important basis for the institution of treatment [20]. It may be argued that more aggressive treatment might further lower the relative risk of stroke. Conversely more extensive treatment, including the common use of new medications without well-known long-term effects, could bear a risk for potentially harmful side-effects. Excluding systolic blood pressure as a diagnostic criterion may have caused widespread clinical acceptance of high systolic blood pressure levels, which lately has been shown to be associated with increased risk for stroke [l.3 8 4 0 1 . Another possible explanation for the remaining over-risk could be that target organ damage prior to detection and treatment is not fully reversed. Thus, in summary, treated hypertensive patients still suffer increased risk of stroke, but this over-risk seems lower than previously anticipated in spite of the rather high blood pressure levels used as diagnostic criteria in this study. zyxwvuts Acknowledgements This study was supported by grants from the Swedish Heart and Lung Foundation: Lund University: and the Skaraborg County Council. zyxwv zyxwvu zy References 1 Dawber TR. The Framingham Study. the Epiderniologg o j Atherosclerotic Disease Cambridge: Harvard University Press. 1980. 162 zyxwvutsrqpo zyxwvutsr U. L I N D B L A D et al. zyxwvutsrq zyxwvutsr zyxwvutsr zyxwvutsrqp zyxwvuts zyxwvutsrqp 2 The Pooling Project Research Group. Relationship of blood pressure, serum cholesterol. smoking habit, relative weight and ECG abnormalities to incidence of major coronary events. Final report of the Pooling Project. / Chron Dis 1978: 31 : 201-306. 3 MacMahon S. Peto R. Cutler J et a/. 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: 765-74. 4 Hypertension Detection and Follow-up Program Cooperative Group. Five year findings of the hypertension detection and follow-up program. Ill. Reduction in Stroke Incidence Among Persons With High Blood Pressure. / A M A 1982: 247: 633-8. 5 Management Committee. The Australian Therapeutic Trial in Mild Hypertension. Lancet 1980: i: 1261-7. 6 Medical Research Council Working Party. MRC trial of treatment of mild hypertension: principal results. Br Med I 1985: 291: 97-104. 7 Collins R. Peto R. MacMahon S et a/. Blood pressure, stroke, and coronary heart disease. Part 2, short-term reductions in blood pressure: overview of randomised drug trials in their epidemiological context. Lancet 1990: 335: 827-38. 8 NIH. The 1984 Report of the joint National Committee on Detection, Evaluation and Treatment of High Blood Pressure. Washington: National Institutes of Health. 1984: (NIH Publication no. 84-1088). 9 Tuomilehto J. Nissinen A, Wolf E. Geboers 1. Piha T. Puska P. EKectiveness of treatment with antihypertensive drugs and trends in mortality from stroke in the community. Br Med ] 1985: 291: 857-61. 10 Berglund G. lsacsson S-O. Ryden L. The Skaraborg Project-a controlled trial regarding the effect of structured hypertension care. Arta Med Srand 1979: 205: (Suppl 626): 64-8. 11 RBstam L. Berglund G. lsacsson S-0. Ryden L. The Skaraborg Hypertension Project. 1. The prevalence of hypertension. Arta Med Scand 1986: 219: 243-8. 12 Lindholm L. Ejlertsson G. Schersten B. High risk of cerebrocardiovascular morbidity in well treated male hypertensives. A retrospective study of 40-59-year-old hypertensives in a Swedish primary care district. Arta Med Srand 1984: 216: 251-9. 13 Isles GC. Walker LM, Beevers GD et a/. Mortality in patients of the Glasgow blood pressure clinic. /. Hypertens 1986; 4: 141-56. 14 Samuelsson 0, Wilhelmsen L. Elmfelt D. Pennert K. Wedel H, Wikstrand J , Berglund G. Predictors of cardiovascular mor- 19 Pollare T. Lithell H. Berne C. A comparison of the effect of hydro-chlorothiazide and captopril on glucose and lipid metabolism in patients with hypertension. N Erigl/ M r d 1989: 321: 868-73 20 1989 guidelines for the management of mild hypertension: memorandum from a WHOIISH meeting. / H!jpFrterrs 19x9: 7: 689-93. 2 1 Johannesson M. Economic Evaluation of H!jperterisiori Treattirent. Linkoping: University of Linkoping. 199 1. (Thesis). 22 Linblad 11. Ranstam 1. Rlstam 1.. Validity of register data on 23 24 25 26 219: 261-9. 27 Lindblad U. Rlstam L. Ryden L. Ranstam J . Herglund G, Isacsson S-0. Reduced stroke incidence with structured 28 29 30 31 32 33 34 bidity in treated hypertension: results from the Primary Preventive Trial in Goteborg. Sweden. I Hypertens 1985: 3: 167-76. 1 5 The BBB Study Group. The BBB Study: a prospective randomized study of intensified antihypertensive treatment. j Hypertens 1988: 6: 693-7. 16 MacMahon SW. Cutler JA. Furberg C. Payne GH. The effects of drug treatment for hypertension on morbidity and mortality from cardiovascular disease : a review of randomized controlled trials. Prog Curdiova.Fr Dis 1986: 24 (Suppl 1): 99-1 18. 17 Ranstam J. JogreusC. Olsson H. Letter to the editor. Acta Med Scad 1986; 219: 429-31. 18 Samuelsson 0. Wilhelmsen L. Andersson OK, Pennert K. Berglund G. Cardiovascular morbidity in relation to change in blood pressure and serum cholesterol levels in treated hypertension. Results from the primary prevention trial in Goteborg. Sweden. / A M A 1987: 258: 1768-76. acute stroke and acute myocardial infarction : The Skaraborg Hypertension Project. Scand / SOL.Med 1993 (in press). Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. / Arn Stat Assot 19 58 : 5 3 : 4 57-8 1. Kalbfleisch JD. Prentice RL. The Statistical AmI!jsis of Failure Time Data. New York: Wiley. 1979. Rlstam L. Berglund G . lsacsson SO. Ryden 1. The Skaraborg Hypertension Project 11. Feasibility of a medical care program for hypertension. Arta Med Srarid 1986: 219: 249-60. Rlstam L. Berglund G. Isacsson SO. Ryden L. The Skaraborg Hypertension Project 111. Influence on blood pressure of a medical care program for hypertension. Acta Med Scarid 1986: 35 hypertension care: The Skaraborg Hypertension Project. / Hypertens 1990: 8 : 1147-53. Wilhelmsen L. Ljungberg S. Wedel H. Werko I.. A comparison between participants and non-participants in a primary prevention trial. / Chron Dis 1976: 29: 331-9. Haglund BJA. lsacsson S-0. Ryden L. RBstam L. Health protile of Skaraborg county 1977. A Swedish rural cross-sectional study. Srand / Prim Health Care 1983: 1 : 102-1 3. Bengtsson C. Gredmark T. Hallberg L et a/. The popultition study of women in Gothenburg 1980-8 I-the third phase of a longitudinal study. Scarid] SOL.Med 19x9: 2: 141-145. Holme I. Helgeland A. Hjerman I et a/. Coronar!! risk fartors in relation to sorio-economir status. The Oslo study. life insurance companies institute for medical statistics in Oslo and medical out-patient clinic. Ullevaal Hospital, Oslo, 197X. Wolf PA, Kannel WB. Verter J. Current status of risk factors for stroke. Neurol Cliri 1983: 1: 3 1 7 4 3 . Dyken ML. Wolf PA, Barnett HJM er al. A statement for physicians by the sub-committee on risk factors and stroke of the stroke council. Stroke 1984: 1 5 : 1105-11. Wolf PA, Kannel WB. Dawber TR. Prospective investigations: The Framingham study and the epidemiology of stroke. Adv Neurol 1978: 19: 107-20. Salonen JT. Puska P. Tuomilehto J. Homan K. Relation of blood pressure. serum lipids. and smoking to the risk of cerebral stroke. A longitudinal study in eastern Finland. Stroke 1982: 13: 327-33. 36 Wilhelmsen L. Primary and secondary prevention. / Hypertens 1987: 5 : S79-82. 37 Lindholm L. Lithman T. lltility of a population-based case- control study model with a limited number of patients in estimating risks of hypertension. Actn Med Srarid 198 5 : 2 1 7: 403-9. 38 Cruickshank JM. Pennert K. Sorman AE. Thorp IM. Zacharias FM. Zacharias FJ. Inw mortality from all causes, including myocardial infarction. in well-controlled hypertensives treated with a beta-blocker plus other antihypertensives. / H!/pertens 1987: 5: 489-98. zyxwvutsrqpo zyxwvutsr zyxwvutsrqp zyxwvuts zyxwvutsr zyxwvutsr STROKE MORBIDITY 39 Shaper AG. Phillips AN, Pocock SJ,Walker M. Macfarlane PW. Risk factors for stroke in middle aged British men. Br Med / 1991: 302: 1111-5. 4o SHEP Cooperative Research Group, Prevention of stroke by anti-hypertensive drug treatment in older persons with isolated systolic hypertension. Final results of the Systolic Hypertension in the Elderly Program (SHEP). IAMA 199 1 : 265: 3255-64. 163 Received 17 June 1992. accepted 1 7 September 1992. Correspondence: Dr Ulf Lindblad. Department of Community Health Sciences. University of Lund. Malmo General Hospital. S-214 01