Incidence of myocardial infarction in elderly men being treated with antihypertensive drugs: population based cohort studyBMJ 1996; 313 doi: http://dx.doi.org/10.1136/bmj.313.7055.457 (Published 24 August 1996) Cite this as: BMJ 1996;313:457
- Juan Merlo, doctoral fellowa,
- Jonas Ranstam, biostatisticianb,
- Hans Liedholm, senior registrara,
- Bo Hedblad, epidemiologista,
- Gunnar Lindberg, clinical epidemiologistb,
- Ulf Lindblad, research fellowc,
- Sven-Olof Isacsson, professora,
- Arne Melander, professorb,
- Lennart Rastam, professora
- a Department of Community Medicine, Lund University, Malmo University Hospital, S-205 02 Malmo, Sweden
- b NEPI Foundation (Swedish Network for Pharmacoepidemiology), Malmo and Stockholm, Sweden
- c Skaraborg Institute, Skovde, Sweden
- Correspondence to: Dr Merlo.
- Accepted 17 June 1996
Objective: To analyse the association between use of antihypertensive treatment, diastolic blood pressure, and long term incidence of ischaemic cardiac events in elderly men.
Design: Population based cohort study. Baseline examination in 1982-3 and follow up for up to 10 years.
Setting: Malmo, Sweden.
Subjects: 484 randomly selected men born in 1914 and living in Malmo during 1982.
Main outcome measures: Observational comparisons of incidence rates and rate and hazard ratios of ischaemic cardiac events (myocardial infarction or death due to chronic ischaemic cardiac disease).
Results: The crude incidence rate of ischaemic cardiac events was higher in those subjects who were taking antihypertensive drugs than in those who were not (rate ratio 2.6 (95% confidence interval 1.7 to 3.9)). After adjustment for potential confounders (differences in baseline smoking habits, blood pressure, time since diagnosis of hypertension, ischaemic or other cardiovascular disease, hypercholesterolaemia, hypertriglyceridaemia, diabetes mellitus, obesity, and raised serum creatinine concentration) this rate was reduced but still raised (hazard ratio 1.9 (1.0 to 3.7)). In men with diastolic blood pressure >90 mm Hg, antihypertensive treatment was associated with a twofold increase in the incidence of ischaemic cardiac events (rate ratio 2.0 (1.1 to 3.6)), which vanished after adjustment for potential confounders (hazard ratio 1.1 (0.5 to 2.6)). In those subjects with diastolic blood pressure </=90 mm Hg, antihypertensive treatment was associated with fourfold increase in incidence (rate ratio 3.9 (2.1 to 7.1)), which remained after adjustment for potential confounders (hazard ratio 3.8 (1.3 to 11.0)).
Conclusion: Antihypertensive treatment may increase the risk of myocardial infarction in elderly men with treated diastolic blood pressures </=90 mm Hg.
In our 10 year follow up of 484 elderly men the risk of an ischaemic cardiac event was higher in men who were taking antihypertensive drugs than in those who were not
Among men with diastolic blood pressure >90 mm Hg, the risk was increased twofold but disappeared when adjustments were made for other cardiovascular risk factors
Among those with diastolic blood pressure </=90 mm Hg, the risk associated with taking antihypertensive drugs was four times higher and remained after adjustment for other cardiovascular risk factors
These findings support the concept of a J shaped curve for risk of myocardial infarction in relation to treated diastolic blood pressure
A major aim of antihypertensive treatment is to reduce the risk of myocardial infarction, but its ability to achieve this in clinical trials is modest,1 2 3 and its effectiveness in reducing the incidence of myocardial infarction in the general population is questionable.4 5 6 Indeed, several investigations indicate that antihypertensive treatment may increase the risk of myocardial infarction,7 8 at least in subjects whose myocardium is hypertrophic or ischaemic.9 This may relate to critical reduction of diastolic blood pressure.7 8 9 Recent studies have also shown increased cardiovascular mortality in elderly men taking antihypertensive drugs who have reduced diastolic blood pressure.10 For these reasons, we investigated the long term incidence of ischaemic cardiac events in relation to antihypertensive treatment and diastolic blood pressure in a population based cohort of elderly men.
Subjects and methods STUDY POPULATION
“Men born in 1914” is a prospective population study with the aim of investigating determinant factors for cardiovascular and respiratory diseases. It was carried out in the city of Malmo, Sweden, and a first examination was performed during 1969-70.11 The cohort was redefined in 1982-3, when the men were 68 years old, and a baseline cross sectional study was carried out. Five hundred of the 621 men (80.5%) born in the even months of 1914 who were living in Malmo during 1982-3 participated in this examination. Information on both blood pressure and antihypertensive treatment was complete in 484 of the participants, and these men constituted the current study population.
The health examination began with a structured interview at home that included questions about medical history and drug treatment. A comprehensive evaluation of cardiovascular history and status was made at the department of community medicine and at the laboratory of clinical physiology at Malmo University Hospital, a week later. The procedure has been described in detail elsewhere.12
Casual systolic and diastolic (phase V) blood pressures were recorded to the nearest 5 mm Hg. The subjects were asked to abstain from smoking on the morning of the test day, and the measurements were made on the right arm, after 15 minutes of supine rest, with a mercury manometer and a rubber cuff (12 × 35 cm). The presence of angina and intermittent claudication was assessed by questionnaire,13 and episodes of ST segment depression were assessed by 24 hour electrocardiographic ambulatory registration (Holter monitoring).14 Ischaemic heart disease was defined as presence of angina or occurrences of ischaemic type ST segment depression.
Blood samples were drawn after a minimum fasting period of 8 hours, and concentrations of serum total cholesterol, serum triglyceride, serum creatinine, and blood glucose were determined at the department of clinical chemistry at Malmo University Hospital. Hypercholesterolaemia was defined as a serum cholesterol concentration >/=6.5 mmol/l, and hypertriglyceridaemia as a serum triglyceride concentration >/=2.3 mmol/l.15 Creatinine concentration was classed as high when it was equal to or higher than the 95th centile of the distribution of the participants.
Body mass index was calculated as weight (kg)/ (height (m))2. Obesity was defined as a body mass index >28.16 Diabetes was diagnosed when subjects were receiving drug treatment for the disease or had a fasting blood glucose concentration >/=7.0 mmol/l.
Data on current drug treatment, years since diagnosis of hypertension, and smoking were assessed by a structured questionnaire, which was filled in during the home interview. Subjects reporting continuous use of antihypertensive drugs were asked to show the packages of prescribed drugs to the research assistant, who recorded the trade names. Reported duration of hypertension was arbitrarily dichotomised into <5 years and >/=5 years. The men were classed as current smokers, former smokers (those who had stopped smoking at least one month before investigation), and those who had never smoked. Survey data on previous admission to hospital for myocardial infarction or cerebrovascular events (transient ischaemic attacks or stroke) were validated against medical records.
The men were followed up from baseline in 1982-3 until 31 December 1992, or until death or the occurrence of an ischaemic cardiac event. Fatal and non-fatal ischaemic cardiac events were identified according to the procedure used for the Malmo heart infarction register.17 All events were validated against medical records or necropsy reports. An ischaemic cardiac event was defined as either a myocardial infarction (ICD-8 code 410 (international classification of diseases, eighth revision)) or death due to chronic ischaemic cardiac disease (ICD-8 code 412). In men with more than one ischaemic cardiac event only the first event was used for analysis. Information on deaths outside hospital was obtained from the national mortality register.
The crude incidence, the rate ratio, and Kaplan-Meier estimates were calculated to compare the incidence of ischaemic cardiac events between the men who took antihypertensive drugs and those who did not. Proportional hazards models were used to estimate the hazard ratio of ischaemic cardiac events among users and non-users of antihypertensive drugs after adjustment for potential confounders—systolic blood pressure, diastolic blood pressure (except when the analysis included stratification on diastolic blood pressure), presence or absence of previous myocardial infarction, other ischaemic heart disease, intermittent claudication, cerebrovascular events, hypercholesterolaemia, hypertriglyceridaemia, elevated serum creatinine concentration, diabetes mellitus, obesity, a history of hypertension of >/=5 years, use of cardiac glycosides, and history of smoking. All confidence intervals were calculated at the 95% level.
Of the 484 men in the present study, 395 (82%) had also participated in the initial examination in 1969-70.11 The prevalence of antihypertensive treatment was then only 3%. Mean blood pressures in 1969-70 were 156/93 mm Hg in those who were taking antihypertensive drugs in 1982-3 and 131/81 mm Hg in those who were not. Men taking antihypertensive drugs in 1982-3 had almost identical mean blood pressures if they had had cardiovascular disease before 1982-3 (157/93 mm Hg) or if they had not (156/92 mm Hg). Cardiovascular disease was uncommon in 1969-70 but seemed to be more common among those who took antihypertensive drugs in 1982-3 (11%) than among those who did not (6%). However, of those men who had had cardiovascular disease in or before 1969-70 and who were taking antihypertensive drugs in 1982-3, only two experienced an ischaemic cardiac event during follow up.
Table 1 shows the baseline characteristics of the 484 men in this study. Of these, 113 (23%) were taking antihypertensive drugs: 68 used one drug, 39 used two, and six used three drugs. The antihypertensive drugs used were diuretics (mainly thiazides, 72 men), β blockers (64 men), hydralazine (18), nifedipine (2), bethanidine (2), prazosin (2), methyldopa (2), and reserpine (1). The men taking antihypertensive drugs had a higher prevalence of hypertension, previous cerebrovascular and cardiovascular events, ischaemic heart disease, diabetes mellitus, hypertriglyceridaemia, raised serum creatinine concentrations, and former smoking. The total follow up time was 3842 person years, and the mean follow up was 8 person years. No subject was lost to follow up.
Table 2 shows the crude incidence rates and rate ratios of ischaemic cardiac events in relation to several potential risk factors. Figure 1 shows the occurrence of ischaemic cardiac events during follow up among men who were taking antihypertensive drugs and among those who were not.
Table 3 shows the rate ratios of ischaemic cardiac events and the hazard ratios after adjustment for differences in risk factors and with stratification by diastolic blood pressure. The crude incidence rate of ischaemic cardiac events was higher in the men taking antihypertensive drugs than in those who were not (rate ratio 2.6 (95% confidence interval 1.7 to 3.9)) and differed only marginally by type of drug taken (rate ratio for use of β blockers v non-use, 2.2 (1.3 to 3.8); for use of diuretics, 2.6 (1.6 to 4.3); for use of other antihypertensive drugs, 2.4 (1.1 to 5.2)). After adjustment for potential confounders (smoking habits, systolic and diastolic blood pressure, years since diagnosis of hypertension, and prevalence of ischaemic or other cardiovascular or cerebrovascular disease, hypercholesterolaemia, hypertriglyceridaemia, diabetes mellitus, obesity, and raised creatinine concentrations) the rate was reduced but still raised (hazard ratio 1.9 (1.0 to 3.7)).
This analysis was then repeated with stratification by diastolic blood pressure (table 3). Among men with diastolic blood pressure >90 mm Hg the use of antihypertensive drugs was associated with a twofold increase in incidence rate (rate ratio 2.0 (1.1 to 3.6)), but this increase vanished after adjustment for potential confounders (hazard ratio 1.1 (0.5 to 2.6)). Among men with a diastolic blood pressure </=90 mm Hg the use of antihypertensive drugs was associated with a fourfold increase in the incidence rate (rate ratio 3.9 (2.1 to 7.1)), which remained after adjustment for potential confounders (hazard ratio 3.8 (1.3 to 11.0)).
In this population based cohort study of elderly men the use of antihypertensive drugs was associated with an increased incidence of ischaemic cardiac events. However, after adjustment for confounders the twofold increase in risk among those with a diastolic pressure above 90 mm Hg disappeared, whereas the fourfold increased risk among those with a pressure of 90 mm Hg or below remained.
RELIABILITY OF STUDY
Observational non-randomised studies usually have various weaknesses, and studies of patients treated for hypertension may have problems in assessing the occurrence of non-fatal events and the extensive follow up necessary for an appropriate evaluation. These arguments are particularly pertinent as our findings are at variance with common opinion on the benefit of antihypertensive treatment. On the other hand, results similar to ours have been obtained in other studies.4 5 6 7 8 9 10
As the study population constituted a random sample from the total population and as the participation rate was high, it is not likely that selection problems invalidate the findings. Furthermore, an investigation on the characteristics of non-participants has confirmed the representativeness of the study population.12
The study was performed in an urban area with only one general hospital, and this facilitated the assessment of occurred non-fatal events. Moreover, computerised systems for registration of myocardial infarctions and causes of death in the population have long been established and the number of necropsies performed during follow up was high. Hence, misclassification of cardiac events is unlikely.
All the men were personally interviewed about exposure to risk factors and drug treatments at baseline, and they were all examined by a physician. The reliability of this information was high when investigated in a separate study.17 Hence, major misclassification of risk factors is unlikely. Also, a large number of potential confounding factors have been accounted for in the analysis, including differences in smoking habits, systolic blood pressure, the duration of hypertension, and the prevalence of cardiovascular and cerebrovascular disease, diabetes mellitus, hyperlipidaemia, and raised creatinine concentrations.
In order to reduce the problems associated with subgroup analysis, the division of diastolic blood pressure at 90 mm Hg was based solely on the World Health Organisation's definition of hypertension.18
EXPLANATION OF RESULTS
The higher incidence of ischaemic cardiac events in men with diastolic blood pressure </=90 mm Hg might reflect severity bias or confounding by indication—that is, that men with previous myocardial infarction or cerebrovascular disease and other high risk subjects were treated more aggressively with antihypertensive drugs and consequently had a lower diastolic blood pressure. This seems unlikely, however, since our adjustments accounted for possible differences in indications for treatment at baseline (these adjustments seem credible as they eliminated the increased risk in men with diastolic blood pressure >90 mm Hg). In addition, the data from the physical examinations of the subjects 13-14 years earlier (in 1969-70) do not suggest that severity bias could explain the findings.11
The main antihypertensive drugs used in this study were diuretics (mainly thiazides) and β blockers. It is often claimed that metabolic side effects may antagonise the cardioprotective purpose of these drugs, but metabolic side effects of thiazides are negligible at the low doses used in Sweden during the time of follow up.19 Furthermore, differences in hypercholesterolaemia, hypertriglyceridaemia, and diabetes mellitus at baseline were accounted for in the adjustments. As the incidence of ischaemic cardiac events was higher in those with lower blood pressure, poor compliance with treatment does not seem to be a reasonable explanation.
A plausible explanation for the increased incidence of ischaemic cardiac events in men taking antihypertensive drugs whose diastolic blood pressure was </=90 mm Hg is that it was confined to a subgroup of subjects in whom lowering of diastolic blood pressure could be detrimental. Patients with compromised coronary flow reserve could be such a group.8 This would be compatible with the idea of a J shaped curve for diastolic blood pressure in relation to risk of myocardial infarction—that is, that lowering blood pressure below an optimal level increases the risk.9
This study was presented in part at the 11th congress of the International Society of Pharmacoepidemiology, Montreal, 27-30 August 1995, and the first congress of the European Association of Clinical Pharmacology and Therapeutics, Paris, 27-30 September 1995.
Funding The study was supported by the NEPI Foundation; by grants from the National Corporation of Pharmacies' (Apoteksbolaget's) fund for research and studies in health economics and social pharmacy, the Bank of Sweden Tercentenary Foundation, the Swedish Heart and Lung Foundation, and the Medical Faculty of Lund University; and by the National Institute for Public Health.
Conflict of interest None.