Hypertension, Antihypertensive Treatment, and Sudden Coronary Death. The Framingham Study WILLIAM B. KANNEL, L. ADRIENNE CUPPLES, RALPH B.

Transcription

1 Hypertension, Antihypertensive Treatment, and Sudden Coronary Death The Framingham Study WILLIAM B. KANNEL, L. ADRIENNE CUPPLES, RALPH B. AND JOSEPH STOKES IE D'AGOSTINO, SUMMARY During 30 years of follow-up, there were 183 sadden deaths hi men and 77 hi women ages 35 to 94 years who participated hi the Framingham Study. Risk of sudden death was increased threefold in hypertensive persons but only if there was no previously diagnosed coronary heart disease. Men receiving antihypertensive treatment had more than twice the risk of sudden death compared with those who were untreated, whether or not they had prior manifestations of coronary heart disease. More than twice as many men who died suddenly were receiving antihypertensive therapy compared with those hi the population at risk of the same age. In those with overt coronary heart disease, 34% of those dying suddenly were on antihypertensive treatment compared with 18% of those of the same age in the general population. Multivariate analysis taking into account the level of blood pressure, electrocardiographic abnormalities, and previously diagnosed coronary heart disease and cardiac failure, all of which are predisposing factors for sudden death, indicated a persistent increased risk of sudden death hi association with antihypertensive treatment. Tests of interaction indicate that the excess sudden death risk was not confined to those with electrocardiographic abnormalities. In women, it may be associated with diabetes. These data suggest that some feature of antihypertensive treatment as practiced hi the general population may contribute to sudden death incidence in an ill-defined subgroup of hypertensive persons. (Hypertension 11 [Suppl U]: II-45-II-50, 1988) KEY WORDS multivariate analysis epidemiology factors diabetes HYPERTENSION is a powerful predisposing factor for atherosclerosis, accelerating lipidinduced atherogenesis. Indeed, hypertension may play an essential role since atherosclerosis seldom occurs in low pressure segments of the circulation, such as in the pulmonary arteries or veins. 1 Controlled trials have indicated the efficacy of antihypertensive therapy in prolonging life and delaying onset of strokes, cardiac failure, and renal insufficiency. 2 " 3 Animal experiments have demonstrated effectiveness of blood pressure reduction in slowing lipid-induced atherogenesis. 6 However, the efficacy of antihyperten- From the Section of Preventive Medicine and Epidemiology (W.B. Kannel, J. Stokes W"), the School of Public Health (L.A. Cupples), Boston University Medical Center, and the Department of Mathematics (R.B. D'Agostino), Boston University College of Liberal Arts, Boston, Massachusetts. Address for reprints: William B. Kannel, M.D., M.P.H., Section of Preventive Medicine and Epidemiology, Boston University Medical Center, Doctor's Office Building, Room 1105, 720 Harrison Avenue, Boston, MA cohort study cardiovascular risk sive therapy in avoiding coronary heart disease () has been less well demonstrated. 78 The purpose of this report is to examine the role of hypertension in the development of sudden death in the Framingham Study cohort and the impact of antihypertensive treatment on sudden death incidence. Subjects and Methods The Framingham Study cohort was originally composed of 5209 men and women aged 30 to 62 years on entry into the study between 1948 and This report is based on information derived from subjects aged 35 to 94 years at the biennial examinations. New events were monitored at biennial intervals by reexamination, daily monitoring of hospitalizations, information from personal physicians, medical examiner's reports, and death certificates. 10 ' " The circumstances of each death were carefuly determined from hospital records, medical examiner's reports, and interviews with spouses and personal physicians. Atherogenic traits and life-styles possibly predisposing to were recorded biennially. These in

2 H-46 ANTfflYPERTENSIVE DRUG EFFECTS SUPPL n HYPERTENSION, VOL 11, No 3, MARCH 1988 eluded cigarette habit, serum lipids, blood glucose, and blood pressure. 9 The methodology, sampling procedure, laboratory methods, criteria for disease end points, and completeness of follow-up have been published elsewhere. 9 " 11 The assignment of coronary heart disease as a cause of sudden death is inferential. Coronary cause is assumed when other possible causes such as ruptured or dissecting aneurysm, pulmonary embolism, and trauma are excluded clinically or by means of postmortem examination. Under these circumstances, death within minutes in persons not otherwise ill with a potentially lethal illness was designated as a coronary sudden death. 10 " Subjects aged 35 to 94 years were classified according to level of suspected cardiovascular risk factors, or congestive heart failure status, electrocardiographic (ECG) abnormality, and antihypertensive treatment status at each biennial examination by age and sex. To examine the risk factors related to occurrence of sudden death, person-bienniums were used to form 2-year age-adjusted incidence rates. Subjects at risk were followed for 2 years for occurrence of sudden death in relation to age, sex, and antihypertensive treatment status with and without prior. Tests of significance and estimates of risk ratios for occurrence of sudden death in relation to blood pressure and antecedent antihypertensive treatment were based on logistic regression models. 12 ' l3 Risk factors considered in addition to blood pressure and antihypertensive treatment for assessment of the net effect of treatment included cholesterol, cigarette habit, glucose intolerance, relative weight, hematocrit, and vital capacity. Risk factors and age were redefined for each subject at the beginning of each biennium. The criteria for hypertension are based on two systolic and diastolic blood pressure measurements taken at each biennial examination. A participant was considered to have definite hypertension if both systolic pressures were equal to or exceeded 160 mm Hg or if both diastolic blood pressures were 95 mm Hg or greater or if one systolic and one diastolic pressure exceeded these limits. Participants were classified as having borderline hypertension if they had either one systolic blood pressure between 140 and 159 mm Hg or one diastolic blood pressure between 90 and 94 mm Hg but not meeting the criteria for definite hypertension previously stated. In the early portion of the study, diuretics and reserpine were the predominant drugs used, and /3-blockers were among the drugs used for treatment in the last 10 years of the study. Results Sodden Death Incidence During 30 years of biennial surveillance of the cohort of 5204 subjects, aged 35 to 94 years, there were 183 sudden coronary deaths in men and 77 in women. There were 142 sudden deaths without other evidence of overt and 118 with prior evidence of. This constitutes a biennial overall incidence of 6.3 per 1000 for men and 2.0 per 1000 women. Women TABLE 1. Incidence of Sudden Death by Age, Sex, and Coronary Heart Disease Status Biennial rate per 1000 With Free of prior All subjects Age Men Women Men Women Men Women Total 35^ , Total = coronary heart disease. Data for this and all subsequent tables from a 30-year follow-up of the Framingham Study. Subjects are years of age. lagged behind men in incidence by approximately 20 years. Incidence increased sharply with age in both sexes, with the gap in incidence between the sexes closing with advancing age. Incidence rates in those with prior were six to eight times that of those free of (Table 1). The age trend in those with was markedly blunted, particularly in men. However, rates were lower in women with prior. Hypertension Risk of all initial clinical manifestations of coronary heart disease, including sudden death, increased in relation to the severity of hypertension (Table 2). At all levels of hypertension, women maintained their advantage over men with respect to development of clinical coronary events, including sudden death. In those free of prior, risk of sudden death in both sexes was significantly related to the level of both systolic and diastolic pressures (Table 3). There was no evidence of a closer relationship to diastolic than to systolic pressure, nor was there evidence of any critical value of pressure. Treatment for hypertension is not taken into account in Tables 2 and 3. Hence, treated subjects are classified according to their blood pressure readings. In sharp contrast to those without prior overt, risk of sudden death in those with prior was neither significantly nor strongly related to hypertensive status (Table 4). At all levels of blood pressure, TABLE 2. Clinical Manifestations of Coronary Heart Disease by Hypertensive Status Hypertensive status* Angina pectoris Myocardial infarction Sudden death Total Men Women Men Women Men Women Men Women Normal 7~6 5~8 8~4 3~0 4~2 L3 17A 9~6~ Mild Definite = coronary heart disease. Normal = <140/90 mm Hg; Mild = /90-94 mm Hg; Definite = a 160/95 mm Hg. Data significant at p<0.001.

3 HYPERTENSION AND THE FRAMINGHAM STUDY/Kannel et al TABLE 3. Twenty-Year Risk of Sudden Death Among Persons Free of Coronary Heart Disease by Systolic and Diastolic Blood Pressures Tertiles of blood pressure Lower Middle Upper BP = blood pressure. /><0.001; tp<0.01; Age-adjusted rate Systolic BP Diastolic BP Men* Woment Ment Woment subjects with prior were at a substantially increased risk of sudden death, including normotensive participants, and the risk ratios for sudden death were not significantly greater in hypertensive than in normotensive subjects. Only in those without prior was hypertension a significant contributor to the incidence of sudden death. Antihypertensive Treatment In the early years of the Framingham Study, there was little treatment for hypertension. However, by the third decade, 68% of male hypertensive and 80% of female hypertensive subjects were under treatment. Persons who indicated that they were taking antihypertensive drugs were at approximately twice the risk of sudden death compared with those who were not (Table 5). This is not unexpected since they had higher pressures. What is unexpected is that they exhibited this increased risk even if they had prior (significant for men only). Furthermore, examination of sudden death risk confined to those with hypertension continues to show the same doubling of risk in those receiving antihypertensive therapy (Table 6). Table 7 also indicates that the current use of antihypertensive medication was twice as common among those who experienced sudden death, whether or not they had prior. A multivariate analysis was performed to examine the net effect of antihypertensive treatment, taking into account not only the actual level of blood pressure but all the other cardiovascular risk factors, ECG abnormalities, and cardiac failure as well. This analysis indicates a marginally significant independent contribution of antihypertensive treatment to sudden death incidence in men without prior and a significant effect among men with prior. Because the coeffi- TABLE 4. Risk of Sudden Death by Hypertensive Status Age-adjusted rate per 1000 Hypertensive No Prior status Normal Mild Definite Men* Women* Ment Woment = coronary heart disease. *p<0.0001; tnot significant. TABLE 5. Risk of Sudden Death by Antihypertensive TreatmentStatus Age-adjusted rate per 1000 Anti-HBP treatment No Prior status Men Women Men Women * t t HBP = high blood pressure; = coronary heart disease. *p<0.001; fp<0.05; $p<0.01; not significant. cients for antihypertensive treatment in those with and without prior are similar in magnitude, we combined these two groups in order to increase the likelihood of detecting a significant contribution for antihypertensive treatment (Table 8). From earlier work, 14 we have noted that generally the risk profiles for sudden death are different for those with and without prior. Therefore, we incorporated all the variables given in Table 9 and then ran a stepwise regression on interactions between status and all measures between age and vital capacity in the list. The significant interactions are indicated in Table 8. The results of Table 8 indicate that antihypertensive treatment provides a significant independent contribution to the risk of sudden death among men and a marginally significant contribution among women. Additional analyses were undertaken to exclude possible selection bias by considering the risk profile among treated persons at the exam before treatment. These results confirm those presented in Tables 9 and 10. Evidently, prior risk factors, blood pressure level, ECG findings, and coronary disease and cardiac failure status, which might have influenced the decision to treat, do not account for the increased sudden death risk observed in those receiving treatment, nor apparently do differences in the intensity of treatment (Table 9). A significantly greater reduction in pressure was noted in men without prior who later died suddenly, though in women the opposite is indicated. The average length of treatment was quite similar. Interactions Because the Multiple Risk Factor Intervention Trial 7 data strongly suggested that in high-risk coronary candidates treated for hypertension and free of clinical drug treatment was associated with higher mortality, including sudden death, in the presence of ECG abnormalities but not in their absence, an analysis was carried out to try to detect such an interaction. A high- TABLE 6. Risk of Sudden Death by Antihypertensive Treatment Among Hypertensive Subjects Age-adjusted rate per 1000 Anti-HBP No With treatment status Men Women Men Women * t * t significant.

4 11-48 ANT1HYPERTENSIVE DRUG EFFECTS SUPPL II HYPERTENSION, VOL. 11, No 3, MARCH 1988 TABLE 7. Current Use of Antihypertensive Medication in Subjects Who Died Suddenly Versus General Population Subjects General population Sudden deaths Age-adjusted percent on anti-hbp medication Total Population With O0 34 HBP = high blood pressure; = coronary heart disease. risk subgroup of the cohort, based on one or more of hypertension, glucose intolerance, serum cholesterol over 240 mg/dl, and more than 20 cigarettes per day, was defined and placed at risk of sudden death depending on whether the subjects did or did not have ECG abnormalities (Table 10). This was done for both those with and without prior. Although those receiving antihypertensive treatment had twice as many ECG abnormalities as those not receiving treatment, there was no indication that the excess mortality was confined to those with ECG abnormality. Regression analyses also do not support a significant interaction between treatment and ECG abnormality. In fact, the most striking adverse, though not significant, effect of antihypertensive treatment was noted in men with prior who had no ECG abnormality (Table 10). In those of both sexes who were without ECG abnormality and no prior and who were at high risk, antihypertensive treatment was associated with a significant increased risk of sudden death. Only in men with prior was there a significantly increased risk of sudden death associated with treatment when coexistent ECG abnormality was present. TABLE 8. Net Effect of Antihypertensive Treatment on Sudden Death Incidence, Taking Other Relevant Variables into Account Multivariate regression coefficients Variables Men Women Age Systolic BP Serum cholesterol Cigarettes/day Glucose intolerance ECG abnormality* Relative weight Hematocrit Vital capacity Prior Prior x age Prior x hematocrit Prior CHF Antihypertensive medication.017t.0094: t $ t -.062t -.107*.944t It *.540* *.021* ) BP = blood pressure; ECG = electrocardiogram; = coronary heart disease; CHF = coronary heart failure. *ECG abnormality consists of tp<0.001, *p<0.05, 1, p>0.05<0.10. TABLE 9. Comparison of Treatment Intensity Among Hypertensive Subjects on Treatment in Sudden Death Versus Noncases Men Women Treatment intensity Reduction in systolic blood pressure (average mm Hg) No sudden death Sudden death Time on medication (average years) No sudden death Sudden death *p<0.01; tp<0.05. No * * With t No * With * An attempt was undertaken to detect other subgroups that might be at elevated risk of sudden death when treated for high blood pressure. The analyses were performed by incorporating all the variables in Table 9 and performing stepwiseregressionon interaction terms between antihypertensive treatment status and the other variables in the model. No significant interactions were detected for men. The data for women suggest that in those with prior, antihypertensive treatment strongly predisposes to sudden death if there is coexistent diabetes (Table 11). The risk ratio for sudden death in those receiving antihypertensive therapy who had impaired glucose tolerance was 3.8, whereas sudden death risk in those on treatment who were without diabetic tendencies was only 80% of that of the general population. Discussion The data reported in this study bear a certain resemblance to observations in clinical trials that have suggested that some subjects being treated with drugs for hypertension may actually be at increased risk of sudden death. 7 ' 8 However, the data reported here are derived from an observational study and not from a controlled clinical trial with random assignment to treatment. Although an attempt was made to eliminate TABLE 10. Sudden Death Incidence in High-risk Candidates* by Electrocardiographs and Antihypertensive Status Anti-HBP Treatment No prior Prior Age-adjusted rate per 1000 No ECG abnormal Men * Women ECG Men ( abnormal t Women *High-risk subjects = >20 cigarettes/day, definite high blood pressure, glucose intolerance, cholesterol >240 mg/dl. tecg abormal. Also includes ECG-myocardial infarction as well as left ventricular hypertrophy, intraventricular block, nonspecific abnormality. HBP = high blood pressure; = coronary heart disease; ECG = electrocardiogram. *p<0.01; p<0.05; Up<0.12; Up<0.07.

5 HYPERTENSION AND THE FRAMINGHAM STUDY/Kannel et al. H-49 TABLE 11. Adjusted* Sudden Death Risk in Women with Prior Evidence of Coronary Heart Disease by Antihypertensive Treatment According to Glucose Intolerance Anti-HBP medication Without glucose intolerance With glucose intolerance *For variables listed in Table 8. HBP = high blood pressure. tp<0.05. Risk ratio t selection bias, there is no guarantee that some hidden bias has not been overlooked. It is also possible that findings from a clinical trial in which the subjects were deliberately selected to exclude interfering variables such as diabetes or serious illness and in which a precise protocol was followed and compliance and side effects carefully monitored cannot be generalized to all patients with hypertension. As indicated by experience in the referred care groups in both the Multiple Risk Factor Intervention Trial and the Hypertension Detection and Follow-up Program, clinical practice often necessitates treatment under very different circumstances. Controlled trials indicate that in general, hypertensive persons who receive treatment live longer than those who do not. 2 " 5 However, clinical trials indicate that the benefits of antihypertensive therapy apply mainly to occurrence of stroke, cardiac failure, renal insufficiency, and aortic aneurysm. 23 Data indicating benefits for have been less consistent. 7 ' 8 Trial data on the effects of treatment on sudden death incidence in particular are sparse, indeed. The Multiple Risk Factor Intervention Trial showed an unfavorable mortality experience in hypertensive men with ECG abnormalities at rest. When compared with those in the usual care group, an adjusted relative risk indicating a 67% excess mortality was noted. 78 The excess coronary mortality among the hypertensive men with ECG abnormalities was attributed chiefly to sudden death. Also, this excess mortality was independent of the baseline blood pressure level. An interaction between ECG abnormality and diuretic treatment was noted. The Framingham Study data do not allow an accurate appraisal of the type, amount, and compliance with treatment. However, during the period when most of the sudden deaths occurred, the predominant drug treatments available were diuretics and reserpine. Studies have shown an association of diuretic therapy with ventricular arrhythmias 13 ' l6 and with sudden death in persons with prior. 17 It is not clear whether hypokalemia is responsible for these associations. 17 Diuretics adversely affect blood lipids 2 " 4 but so do other antihypertensive agents. There is evidence to suggest that hypokalemia may contribute to the occurrence of sudden death, especially in the presence of an increase in circulating epinephrine that accompanies a myocardial infarction, and also in persons with hypoglycemia. 13 The efficacy of antihypertensive drug therapy has been well demonstrated in some clinical trials, 2 ' 3i 5 but the benefits for mortality have not been conclusively demonstrated. 2 " 5 A controlled trial involving a modest sample of older men has also suggested an adverse effect of antihypertensive treatment with diuretics on coronary mortality. 18 The remarkable decline in stroke and cardiovascular mortality in the USA and some other countries over the past 2 decades is to an extent very likely a product of the successful detection and treatment of hypertension." Although the decline in stroke mortality preceded the widespread introduction of effective drug therapy, it definitely has accelerated in its rate of decrease since the emphasis on therapy. This indicates that hypertension control is on the whole beneficial. However, more detailed information is required to define better the indications, contraindications, best drugs, hazards, and side effects of antihypertensive treatment for the avoidance of coronary heart disease. Therapy for mild to moderate hypertension could well be nonpharmacological for the coronary candidate. There is growing evidence that nonpharmacological therapy may be a reasonable alternative to drug therapy. Such therapy includes weight reduction, reduced sodium intake, increased potassium, magnesium, and calcium, reduced fat, and reduced alcohol intake. 20 However, such measures are at present difficult to implement effectively except in research settings. As the incidence and mortality from stroke, cardiac failure, and renal failure decline, more emphasis will have to be placed on reducing coronary mortality and sudden death. Antihypertensive agents for this purpose may have to be tailored to a profile that is more effective against atherogenesis and less inclined to increase myocardial irritability. Diuretic therapy tends to worsen the lipid profile and decrease glucose tolerance and serum potassium. The general efficacy of lowering blood pressure seems well established. We now need better information on the efficacy of various drugs and on nonpharmacological therapy for reducing morbidity and mortality and better data on the proper indications and contraindications for this sequela of hypertension. References 1. Dustan HP. Atherosclerosis complicating chronic hypertension. Circulation 1974;50: Veterans Administration Cooperative Study Group on Antihypertensive Agents. Effects of treatment on morbidity in hypertension, n. Results in patients with diastolic blood pressure averaging 90 through 114 mm Hg. JAMA 1970;213:l Report by the Management Committee of the National Heart Foundation of Australia. The Australian therapeutic trial in mild hypertension. Lancet 1980;l: Helgeland A. Treatment of mild hypertension: afive-yearcontrolled drug trial. The Oslo Study. Am J Med 198O;69: Hypertension Detection and Follow-Up Program Cooperative Group. Five-year findings of the Hypertension Detection and Follow-up Program. I. Reduction in mortality of persons with high blood pressure, including mild hypertension. JAMA 1979^242: Deming QB, Mosbach EH, Bevans M, et al. Blood pressure, cholesterol content of serum and tissues and atherogenesis in the rat. J Exp Med 1958;107: Multiple Risk Factor Intervention Trial Research Group. Base-

6 H-50 ANTIHYPERTENSIVE DRUG EFFECTS SUPPL H HYPERTENSION, VOL 11, No 3, MARCH 1988 line rest elcctrocaidiographic abnormalities, antihypeitensive treatment, and mortality in the Multiple Risk Factor Intervention Trial. Am J Cardiol 1985;55: Cutler J A, FurbergLD. Drug treatment trials in hypertension: a review. Prev Med 1985; 14: Dawber TR, Meadors GF, Moore FE Jr. Epidemiological approaches to heart disease: the Framingham Study. Am J Public Health 1981;41: Doyle JT, Kannel WB, McNamaraPM, Quickenton P, Gordon T. Factors related to suddenness of coronary death: combined Albany-Framingham Studies. Am J Cardiol 1976;37: Kannel WB, McGee DL, Schatzkin A. An epidemiological perspective of sudden death: 26-year follow-up in the Framingham Study. Drugs 1984;28(suppl 1): Walker SH, Duncan DB. Estimation of the probability of an event as a function of several independent variables. Biometrika 1967^4: Schlesselman JJ. Case-control studies. New York: Oxford University Press, 1982: Schatzkin A, Cupples LA, Heeren T, Morelock S, Kannel WB. Sudden death in the Framingharn Heart Study. Am J Epidemiol 1984;120: Holland OB, Nixon JV, Kuhnert L. Diuretic-induced ventricular ectopic activity. Am J Med 1981;70: Medical Research Council Working Party on Mild to Moderate Hypertension. Ventricular extrasystoles during thiazide treatment: substudy of the Medical Research Council Mild Hypertensive Trial. Br Med J 1983;287: Duke M. Thiazide-induced hypokalemia: association with acute myocardial infarction and ventricularfibrillation.jama 1978:239:43^5 18. Morgan T, Adam WP, Hodgson M, Gibberd RW. Failure of therapy to improve prognosis in elderly males with hypertension. Med J Aust 1980;2: Kannel WB, Thom TJ. Declining cardiovascular mortality. Circulation 1984;70: Kaplan'NM. Non-drug treatment of hypertension. Ann Intern Med 1985;102: