Helsinki

Cardiovascular morbidity and multifactorial primary prevention: Fifteen-year follow-up of the Helsinki Businessmen Study* T. E. Strandberg*, U. V. Salomaa**, V A. Naukkarinen***, H. T. Vanhanen*, S. J. Sarna****, and T. A. Miettinen* ' Department of Medicine, University of Helsinki. * National Public Health Institute, Helsinki, ** Jorvi Hospital, Espoo, ' "Department of Public Health, University of Helsinki, Helsinki, Finland Objective: Mortality and morbidity in a long-term follow-up after a multifactorial primary prevention of cardiovascular Design? The five-year randomized controlled multifactorial prevention trial for CVD was performed between 1974-1980. Evaluation of participants (including electrocardiograms) was performed at start, end of trial and five years post-trial. CVD morbidity and mortality follow-up, using ' the country-wide Hospital Discharge Register and Death Certificate Register, was continued up to the end of 1989. Setting: Second Department of Medicine, University of Helsinki. Participants: In all, 3490 business executives born during 1919-34participated in health check-ups in the late 1960s. In 1974,1222 of these men who were clinically healthy, but with CVD risk factors, were entered into the primary prevention trial; 612 randomized to an intervention group and 610 to a control group. In addition, 593 men, who volunteered for the study but were excluded because of low levels of CVD risk factors, were followed- up as a nonrandomized low-risk group. Correspondence to: T. Strandberg, M.D., Department of Medicine, University of Helsinki, FIN-00290 Helsinki, Finland. Key words: prevention, cholesterol, beta-blockers, cardiovascular diseases, Received: 21 December 93; accepted:! June 1994. "Presented partly at the 32nd Annual Conference on Cardiovascular Epidemiology, March 19-20,1992, Memphis IN, and the 65th Scientific Sessions of the American Heart Association, November 16-19,1992, New Orleans, USA Interventions: During the five-year trial in 1974-1980 the subjects of the intervention group visited the investigators every fourth month. They were treated with intensive dietetic-hygienic measures and frequently with hypolipidemic (mainly clofibrate and/or probucol), and antihypertensive (mainly pindolol or propranolol and/or hydrochlorothiazi-de plus amiloride) drugs. The control group and the low-risk group were not treated by the investigators. Main outcome measures: Coronary mortality, nonfatal clinical and silent myocardial infarction (Ml), coronary artery bypass grafting (CABG), fatal and nonfatal ischemic stroke, transient ischemic attack (TIA) leading to hospitali-zation, fatal and nonfatal hemorrhagic stroke (subarachnoid and intracerebral hemorrhage). Results: The total number of fifteen-year coronary events was 75 in the intervention group, 49 in the control group (p<0.05 vs intervention) and 26 in the low-risk group (p<0.05 vs intervention and control). The respective numbers of coronary deaths were 34,14 and 6; of definite or probable MI 32, 30 and 17 (posttrial 16, 24 and 15); of CABG 5, 0 and 0 and of ischemic cerebrovoscular events 8, 20 (p<0.05 vs intervention) and 14. In multifactoriai analysis coronary events were positively associated with serum cholesterol (negatively with HDL-cholesterol), blood pressure and smoking in the intervention and control groups, and with serum triglycerides in the low-risk group. Basal alcohol consumption was a significant predictor of fewer coronary events in the intervention group but not in the control groups. In-trial pindolol treatment of the intervention group, but not the overall beta-blocker treatment, was associated with total (OR 197, 95% CI 103-3.78) and nonfatal coronary events (or 2.51, 95% Cl 1.06-5.93) but not with coronary mortality or with post-trial total or nonfatal coronary events. Associations with other drug treatments tended to be protective only with diuretics. Summary: Low traditional CHD risk factor levels are "ssociated with low coronary morbidity also in this popula-.^jn of middle-aged men. Total coronary events were more frequent, mainly due to coronary deaths, in the intervention group than in the control group. Nonfatal coronary events in the intervention group were significantly associated with the frequent in-trial use ofpindolol, a beta-blocker with intrinsic sympathomimetic activity. Stroke incidence was low in the intervention group. Several multifactorial primary prevention studies were started during the 1970's in order to efficiently control the development of cardiovascular diseases (CVD, for a recent review see 1). Many of these trials had quite meager or even disappointing results, often interpreted to be due to modest risk factor lowering. However, risk factor reduction was substantial in the Helsinki Businessmen Study (2, 3); yet, 15-year follow-up showed increased total, coronary id violent mortality in the intervention group as compa-red to the randomized control group (2, 4, 5). The results have evoked several speculations about causes (6-10), but even though the multidrug treatments of the trial were suspected, no clear-cut explanation could have been elucidated for the result. We have now extended the follow-up to include nonfatal coronary and cerebrovascular events during the 15-year follow-up. For comparison, the respective morbidity will be reported also for a non-randomized low-risk group of the same background population presented in the initial intervention trial (2,3). Study population and intervention methods The details of the Helsinki Businessmen Study have been described in previous reports (2-5). Participants had to be free of vascular disease but have at least one of the following risk factors: 1) serum cholesterol level £ 7.0 mmol/liter; 2) serum triglycerides level > 1.7 mmol/liter; 3) systolic blood pressure > 160 mm Hg; 4) diastolic blood pressure > 95 mmHg; 5) smoking > 10 cigarettes/day; 6) relative body weight 120% or higher; 7) one-hour glucose tolerance (1 g/kg body weight of glucose orally) > 9.0 mmol/liter. In addition to the randomized high-risk control and intervention groups of 610 and 612 men, respectively, we also identified in 1974-75 a group of 593 men who volunteered for the trial, but were excluded because of low level of risk factors. This group has been referred to as the low-risk group. The members of the intervention group visited the investigators every fourth month during the 5-year intervention period, while the control group and the low risk group were in usual health care. Drug treatment (mainly beta-blockers propranolol or pindolol and/or diuretics hydroch-lorothiazide chloride alone or combined with amiloride for hypertension; probucol and clofibrate for hyperlipidemias) was used, if target levels were not reached by hygienic measures (smoking cessation, physical exercise, diet instructions, weight reduction) alone. Diet instructions advised reduced intake of calories, saturated fat, cholesterol, alcohol and sugar, and increased intake of polyunsatu-rated fats, fish, chicken, veal and vegetables. At end-trial most of the individual risk factors were reduced in the intervention group so that the total coronary heart disease (CHD) risk score was substantially smaller in the intervention group than in the control group (2). However, the levels of the low-risk group were not reached (2). At the end-trial all participants were informed of their laboratory results and were advised to continue the health check- ups and possible treatments in usual health care. Probucol therapy was discontinued because the drug is not registered in Finland. The first and second post-trial follow-up evaluations including only mortality were performed in 1985-86 and 1990. They have been described in detail in the respective reports (4,5). The risk factor levels and use of drugs in the three groups were studied in 1985-86 using questionnaires and routine laboratory methods (11) and the results are summarized in Table 1. For the present report the cases of nonfatal myocardial infarction (covering ICD-9 codes 410-412) and ischemic and hemorrhagic cerebrovascular events (ICD-9 codes 430-438) up to December 31, 1989 were collected from the country-wide Hospital Discharge Register. This includes the personal 10-digit code, unique for every citizen in Finland; the hospital code; diagnosis number and the treatment period. Thus, the hospital-made diagnoses of myocardial infarction (MI), transient ischemic attacks (TIA) or stroke can be reliably identified using this register. More TABLE 1
Risk factor levels in the study groups (number of men evaluated in parenthesis) at 5-year post-trial evaluation
Intervention group
Control group
Low-risk group
Age at entry (1974), yr
Relative body weight, %
117 (0.6)
Blood pressure, mmHg
Serum cholesterol, mmol/E
Fasting blood glucose, mmol/l
Smokers, %
Alcohol consumption, g/week
Coffee consumption, cups/d
3.9 (0.1)
3.3 (0.1)"
Exercise, h/week
The values are mean (SD) * p<0.05 vs. control group, ** p<0.01 vs. other groups records invaria-bly contained ECG decryptions (made by physician) and enzyme data relevant to MI. The records were rewieved finally established by a cardiologist (Dr. Matti Romo) than 24 hours the diagnosis of TIA was establi-shed. The cases of hemorrhagic (intracerebral) stroke and subarachnoid hemorrhage were diagnosed using lumbar puncture and/or computerized tomography. One man in the intervention group and Cumulative clinical coronary heart disease events (fatal and Intervention
UJ U. O CC LU
2 4 6 8 10 12 14 16
YEARS FROM RANDOMIZATION L
earlier reports beta-blocker users were analyzed as one group. We have now divided the users according the type of beta-blocker (mainly pindolol or propranolol). Post-trial risk factors and treatments 1 shows that the risk factor levels and drug treat- significantly below the two other groups and the drug treatment was infrequent. demonstrated in Figure 1 and Table 2. In Coronary heart disease (CHD) and cerebrovascular events in the study groups 1974-1989. In-trial (1974-1980) events in parenthesis Intervention
Low-risk
Fatal CHD
Nonfatal Ml
CABG only
ECG only {up to 1985)
Total coronary
Cerebrovascular
15 5 (8) (2)
ischemic stroke
Hemorrhagic stroke
Total coronary or ischemic
cerebrovascular patients
• p<0.05 vs. control group, ** p<Q,Q5 vs. other groups Ml denotes myccardial infarction, CABG denotes coronary artery bypass grafting, TIA denotes hospitalized transient Ischemia attack. the intervention group, the total number of coronary morbi- In fact, during the post-trial ten-year period the respecti-- dity (75 cases) was higher than in the control group (49 ve nonfatal figures are 16 and 24. In addition, five men had cases). The difference was mainly due to coronary deaths, 34 been coronary bypassed without MI in the intervention and vs. 14 cases, while the numbers of the definite non fatal none in the control groups. The lowest morbidity figures events (definite MIs+silent Mis) were similar, 31 vs. 31 cases. (26 cases) were found in the low-risk group, including 6 coronary deaths (all post-trial) and 20 nonfatal (18 post- factors serum cholesterol level and-smoking (plus systolic blood pressure in the intervention group) were significant predictors of total coronary events in both high-risk groups. Risk factors, treatments and coronary morbidity A closer Reported alcohol consumption (log value) was significantly analysis of in-trial beta-blocker users (192 overall, 135 at associated with coronary only in the intervention group the end-trial) showed that 63% used only pindo-lol and 29% (OR 0.41,95% CI 0.26-0.65). Interestingly, only serum tri- propanolol in the intervention group. Of these men 14 used glyceride level was significantly positively associated with both pindolol and propranolol (not concomi-tantly). Only 8 men (4%) used other beta-blockers. In the control group The initial HDL cholesterol levels were unknown becau- 38% of beta-blocker users (23 out of 61) were on se their determination from frozen serum samples turned pindolol, 16% on propranolol and 46% on other beta- out to be unreliable (3). As the effect of HDL-cholesterol on multivariate analyses may be of interest we substituted In a univariate within-group analysis (Table 3) the 15- pre-trial HDL-cholesterol by the levels determined in the year incidence of total CHD events were significantly posi- 1985-86, a time point no more affected by the probucol tively associated with the in-trial clofibrate and pindolol treatment. When controlled for other initial risk factors ise and were higher than in the whole intervention group. shown in Table 4, this surrogate HDL-cholesterol was a The men without medications, in turn, had the lowest inci- significant additional predictor of coronary events in the dence. The respective analysis with Cox proportional three groups. The ORs were 0.29 (95% CI 0.10-0.80), 0.26 hazards models (Table 4) revealed that of the initial risk (95% CI 0.07-0.98) and 0.12 (95% CI 0.02-0.99) in the intervention, control and low risk groups, respectively. The inclusion of different in-trial drug treatments of the intervention group into the Cox model with the initial risk factors is shown in Table 4. The total use of beta-blockers or propranolol alone was not significantly associated with fif-teen-year coronary events (OR 1.73, 95% CI 0.89-3.35 and 1.14, 0.51-2.54, respectively). However, the association of pindolol was significant with total coronary events (OR 1.97, 95% CI 1.03-3.78) and total nonfatal coronary events (OR 2.51, 95% CI 1.06-5.93) but non significant with coronary Treatment**
No. of men treated
Probucol
Clofibrate
All betablocker
Diuretics
Diet or drug < 4 mo
Any intervention
* Includes fatal CHD, nonfatat and silent myocardial Infarctions and CABQ (total number 75) trial drug treatment £ 4 months * Drug at end-trial Odds ratios (OR with 95% confidence interval) of 15-year coronary morbidity for initial risk factors and alcohol consumption without (model A) and with in trial treatment procedures (Model B) Morbidity due to coronary heart disease
Risk factor or treatment (change)*
Intervention, n=75
Control, n=49
Low-risk group, n=26
Age (1 year)
Obesity (1 kg/m1)
Systolic blood pressure (10 mmHg)
Serum cholesterol (1 mmol/i)
Serum trigiycerides {mmol/l, log value) 1.02
One-hour glucose (1 mmol/l)
Smoking (> 10 cig./day)
Alcohol (g/week, log value)
Probucol
Clofibrate
Beta-blocker
Oiuretics
"-tJiet or drug < 4 mo
* p<0.05 *Log transformed values for serum trigiycerides (dichotomous in low-risk group, >1.0 mmol/I) and alcohol use. Smoking (> 10 cigarettes in Intervention and control groups, any smoking in tow-risk group) and treatments are dichotomous variables (drug use > 4 months during intervention period). The diet group includes subjects with no drug treatment, or drugs less than 4 months during the intervention period. # Five-year post-trial values ischemic events in the control, intervention (virtually all nonfatal) was lowest in the intervention group (Fig. 2). The numbers of hemorrhagic strokes were low and similar in all three groups. The Cox proportional hazards model for the combined three groups using various risk factors as covariates revealed that cerebrovascular events were associated only with initial serum cholesterol (OR 1.43, 95% CI 1.06-1.93). Separate group analyses should be interpreted cautiously because of relatively small number of events, but they showed that cholesterol significant predictor in the intervention 2.33,95% CI 1.18-4.60), and smoking in the control group (OR 3.18,95% CI 1.23-8.27). The ORs for systolic ox diastolic blood pressure ranged from 1.42 to 1.09 but all Cls included unity. The Helsinki Businessmen Study /
,
„
»
*
'

c
J
in the fatal events, less so in nonfatal vascular events among all three groups. Actually, if the coronary cerebrovascular ischemic events are combined, there is no significant difference between intervention and control groups; low-risk group still has the lowest incidence. ined with the majority of evidence from decades of to have been exaggerated because the recent trials at least and nonfatal coronary events, but its use does not explain in elderly hypertensive persons have repeatedly shown the excess coronary deaths in the intervention group. As favorable outcomes for both stroke and coronary heart stated earlier (1, 5) and now supported by the favourable disease (32-34). The low OR in diuretics-treated men in results from the low-risk group, we strongly feel that the our trial is in keeping with this, possibly because we almost results of our trial do not question either the importance of xclusively used potassium-sparing preparations. - In traditional risk factors in the development of cardiovascu- contrast to diuretics the in-trial beta-blocker treatment lar diseases, or the value of multifactorial prevention as was significantly associated with fatal plus nonfatal such. We rather think that these results should stimulate coronary events during the intervention period (2) and to more research on the mechanisms of intervention and some extent in the five-year post-trial study (20). However, improved methods to optimally lower cardiovascular risk the association was insignificant with fatal coronary events even in the long-term post-trial follow-up (4, 5). The present new findings show that the in-trial use of pindolol with intrinsic sympathbmimetic activity (ISA) was significantly positively associated with fatal plus nonfatal coronary events over the fifteen-year follow-up period. However, the The study was financially supported by SydantutictmussaatUJ, role of pindolol is more complicated because its in-trial use Suomen Kulttuurirahaston Kymenlaakson Rahasto and Signe och Ane Gyllenbergs Stiftelse. was not related to nonfatal events, mainly to myocardial Dr. Matti Romo MD and Dr. Nan Li MD are gratefully acknow- infarctions, it can be speculated that the harmful effect of ledged for expert oulside review of patients records and electrocar- the drug was not related to arrhythmias, but more likely to diograms; the authors also thank Ms Marja Aarnio and Ms Soile thrombosis and/or progression of atheroslerosis. Evidence has been accumulated that beta-blockers with ISA may confer less benefit in the prevention of CHD (27). Other types of beta-blockers seem to be favourable in primary orevention especially shortly after myocardial infarction 29). Thus, the chemical characteristics of beta-blockers 1. Miettinen TA, Strandberg TE (1992) Implications from recent may make important clinical differences; possible mechani- results of long term multifactorial primary prevention of cardio A highly interesting, but at the moment only speculative 2. Miettinen TA, Huttunen JK, Naukkarinen V, Strandberg T, area, is the possible psychologically detrimental effects of Mattila S, Kumlin T et al (1985) Multifactorial primary preven any intervention methods. These mechanisms may be asso- tion of cardiovascular diseases in middle-aged men. JAMA 254: ciated with intervention causing stress and accelerated atherosclerosis; for instance coronary constriction can be 3. Miettinen TA, Huttunen JK, Naukkarinen V, Starndberg T, mentally induced (35). Equally important may be the Vanhanen H (1986) Long-term use of probucol in the multifacto abrupt end of the intervention period (7). In addition to rial primary prevention of vascular disease. Am J Cardiol 57: 49H-54H drug treatments, our intervention group was submitted to frequent and aggressive health education in order to lower 4. Naukkarinen VA, Strandberg TE, Vanhanen HT, Salomaa VV, Sarna SJ, Miettinen TA (1989) Mortality rates after multifacto the risk factor levels as efficiently as possible. The setting rial primary prevention of cardiovascular diseases. Ann Med 21: clearly differs from ordinary double-blind drug trials, where both the control and the intervention groups feel to 5. Strandberg TE, Salomaa W, Naukkarinen VA, Vanhanen HT, be similarly treated. We have a limited set of psychological Sama SJ, Miettinen TA (1991) Long-term mortality after 5-year data in questionnaires filled in by the participants at the multifactorial primary prevention of cardiovacsular diseases in start and the end of the intervention period and a study is 6. Paul 0, Hennekens CH (1991) The latest report from Finland: a In summary, the present fifteen-year follow-up results of lesson in expectations. JAMA 266:1267-1268 the Helsinki Businessmen Study show significant differen- 7. Anonymous (1991) Should clinical trials carry a health warning? ces between the intervention and control groups after inclusion of nonfatal coronary and cerebrovascular events. 8. Psaty BMf Siscovick DS (1992) Long-term mortality after a car The use of pindolol was significantly associated with total diovascular risk reduction program ACP Journal Club 116:1

Source: http://strophantus.de/mediapool/59/596780/data/helsinki-neu.pdf