Family history of heart attack: a modifiable risk

Transcription

1 PATHOPHYSIOLOGY AND NATURAL HISTORY CORONARY ARTERY DISEASE Family history of heart attack: a modifiable risk factor? KAY-TEE KHAW, M.Sc., M.R.C.P., AND ELIZABETH BARRETT-CONNOR, M.D. ABSTRACT A family history of heart attack is reported to be an independent predictor of cardiovascular death in men. In a 9 year follow-up of 4014 adults from 40 to 79 years old in the Rancho Bemardo Study, men under 60 years of age with a family history of heart attack were at fivefold increased risk. In this study, we sought to determine whether modifiable risk factors, i.e., blood pressure, plasma cholesterol, obesity, and cigarette smoking, have a differential effect on cardiovascular risk in those with and without a family history of heart attack. For both sexes, cigarette smoking was a stronger predictor of cardiovascular disease in those with a family history of heart attack (relative risk of smokers vs nonsmokers was 2.5 for men and 4.0 for women) than in those with no such family history (relative risk of smokers vs nonsmokers was 1.1 for men and 1.7 for women). Conversely, an increased risk of cardiovascular mortality in men with a family history of heart attack was present predominantly in smokers (relative risk related to positive family history was 1.2 in nonsmokers, and 3.3 in smokers). An estimated 68% of the excess deaths in men with a family history of heart attack were attributable solely to the interaction of family history with smoking habit and were therefore potentially avoidable. The risk of cardiovascular disease associated with an apparently inherited predisposition appears to be profoundly affected by modifiable behavior. Circulation 74, No. 2, , A FAMILY HISTORY of heart disease has been reported to increase risk of coronary heart disease by two- to sevenfold in numerous studies.1-' There is also familial aggregation, genetic or environmental, of known cardiovascular risk factors such as raised blood pressure, hyperlipidemia, and cigarette smoking.8'-3 The increased risk of coronary heart disease in those with a positive family history has been partly attributed to the familial occurrence of these risk factors. However, several studies have suggested that family history might itself predict cardiovascular disease independent of the effects of these risk factors.7 We have previously reported that a family history of heart attack was significantly predictive of cardiovascular mortality in men from 40 to 79 years old and that this association persisted after adjusting for level of blood cholesterol, blood pressure, and cigarette smoking.7 In this cohort, cardiovascular risk was increased to a similar From the Department of Community and Family Medicine, School of Medicine, University of California San Diego, La Jolla. Supported by NIH contracts NIH-NHLBI-HV L and NIH- HIADDK-AM K-T. Khaw is a Daland Fellow of the American Philosophical Society. Address for correspondence: Dr. E. Barrett-Connor, Department of Community and Family Medicine, University of California San Diego, La Jolla, CA Received March 11, 1986; revision accepted May 8, extent in men who reported a family history of heart attack whether the relative's attack occurred before or after he was 50 years of age. No association was seen in women. In this study, we assessed whether modifiable risk factors contribute differently to the risk of death attributable to cardiovascular disease in those with and without a family history of heart attack. Methods Eighty-two percent of all residents of Rancho Bemardo, CA, were studied between 1972 and 1974 for the prevalence of heart disease risk factors.'4 All participants completed a standard questionnaire that included questions about cigarette smoking, use of exogenous estrogens, personal history of heart attack, heart failure, stroke, or diabetes, and family history of heart attack (before or after age 50) in first-degree relatives, that is, parents, siblings, or children. No attempt was made to determine which first-degree relative experienced a heart attack or to validate the diagnosis of reported heart attack in family members. Height and weight were measured in subjects in light clothing without shoes; obesity was estimated as body mass index (weight/height2). Blood pressure was measured with a standard sphygmomanometer after the participant had been seated at least 5 min and was repeated in those whose initial readings exceeded 160/90 mm Hg, the lower of two readings being recorded. Total plasma cholesterol level was determined in a standardized Lipid Research Clinic laboratory with an AutoAnalyzer. This group was followed for vital status for an average of 9 years and follow-up was 99.8% complete. Death certificates Vol. 74, No. 2, August

2 KHAW and BARRETT-CONNOR were obtained for all decedents and coded by a certified nosologist who used the eighth revision of the International Classification of Disease. Codes 400 to 438 encompass cardiovascular death. In 30% of the cohort validation of data on the death certificate was attempted by interview of next of kin and/or physician, and by review of hospital records whenever cardiovascular disease was mentioned anywhere on the death certificate. A panel of cardiologists determined that validation data agreed with the death certificate in approximately 85% of cases. The present analysis included all subjects who were 40 to 79 years old and who had no personal history of heart attack, heart failure, or stroke at the baseline examination. For univariate analysis of relative risk, continuous variables (cholesterol, blood pressure, and obesity) were dichotomized into high or low categories with the use of preassigned clinically accepted cutpoint criteria (shown in table 3). Age-adjusted cardiovascular death rates associated with high and low cardiovascular risk factor categories were calculated for men and women with and without a family history of heart attack. The Mantel-Haenszel summary chi square was used to assess statistical significance. The Cox proportional-hazards model'5 was used to determine the independent contributions of age, systolic blood pressure, blood cholesterol, obesity, cigarette smoking, personal history of diabetes, and estrogen use (in women) to cardiovascular mortality in men and women with and without a family history of heart attack. Results Table 1 shows independent relative risks of cardiovascular death associated with a positive family history of heart attack in men and women after adjusting for age, systolic blood pressure, cholesterol, obesity, cigarette smoking, personal history of diabetes, and (in women only) estrogen use. As previously reported,7 a positive family history of heart attack was a significant independent predictor of cardiovascular death in men, an effect seen predominantly in younger men (less than 60 years of age). There was no significant effect of family history of heart attack in women. Risk factor distribution by family history of heart attack is shown in table 2. Men with a family history of heart attack had significantly higher mean blood cholesterol levels and were more likely to have a personal history of diabetes than men without such family histo- TABLE 1 Independent relative risks of family history of heart attack for cardiovascular mortality in men and women 40 to 79 years old after adjusting for age, systolic blood pressure, blood cholesterol, body mass index, cigarette smoking, and personal history of diabetes (Cox regression model) Relative risk (95% confidence limits) Age (yr) Men Women OA ( ) insufficient numbers ( ) 0.9 ( ) A ( ) 0.8 ( ) 240 Ap <.05. TABLE 2 Risk factor distribution by family history of heart attack in men and women 40 to 79 years old No family history of heart attack Positive family history of heart attack Men n Age (yr) 62.2 (10.5) 61.5 (9.9) Systolic BP (mm Hg) (22.8) (22.7) Diastolic BP (mm Hg) 81.8 (11.3) 82.2 (11.3) Obesity index (pounds/inch2 x 102) 3.66 (0.43) 3.68 (0.39) Cholesterol (mg%) (33.8) (36.1)B Current smokers (%/n) 22.0/ /128 Personal history of diabetes (%/n) 4.9/58 81/488 Women n Age (yr) 61.5 (9.8) 60.8 (9.2) Systolic BP (mm Hg) (23.3) (22.6) Diastolic BP (mm Hg) 79.9 (11.4) 79.9 (10.9) Obesity index (pounds/inch2 x 102) 3.38 (0.50) 3.40 (0.53) Cholesterol (mg%) (38.1) (38.0) Current smokers (%/n) 27.1/ /250 Personal history of diabetes (%/n) 2.9/38 3.1/29 Estrogen users (%/n) 38.4/ /366 Values are mean (±+-SD) unless stated otherwise. BP = blood pressure. Ap <.05; Bp <.01, difference between no and positive family history of heart attack. ry; there were no significant differences among the women. Table 3 shows age-adjusted cardiovascular death rates by risk factor category in men and women with and without a family history of heart attack. In men, high plasma cholesterol was a stronger predictor of cardiovascular death in those without than in those with a positive family history of heart attack. Estrogen use appeared to be protective only in women with a family history of heart attack. In individuals of both sexes, the most striking difference in mortality between those with and without a positive family history was seen in the group of current cigarette smokers. Although in women family history of heart attack was not overall associated with higher cardiovascular death rates, current cigarette smoking in both men and women was associated with a greater increase in cardiovascular mortality in those with a positive family history of heart attack. Former smokers had rates similar to those who never smoked. Table 4 shows the independent relative risk of each CIRCULATION

3 PATHOPHYSIOLOGY AND NATURAL HISTORY-CORONARY ARTERY DISEASE TABLE 3 Age-adjusted 9 year cardiovascular disease mortality rates by family history of heart attack and risk factor category in men and women 40 to 79 years old Women Men No family history Positive family history No family history Positive family history of heart attack of heart attack of heart attack of heart attack Rela- Relative Relative tive Relative Risk factor n Rate/l00 risk n Rate/100 risk n Rate/100 risk n Rate/100 risk Systolic blood pressure <160 mm Hg >160 mm Hg A Cholesterol <240 mg% mg% A Obesity <3.8 pounds/inch2 x >3.8 pounds/inch2 x Smoking Never Ex Current B B Personal history of diabetes No Yes Estrogen use No Yes Ap <.05. BRelative risk for smoking is current smokers vs current nonsmokers. TABLE 4 Independent relative risks of cardiovascular death in men and women 40 to 79 years old by family history of heart attack (Cox regression model) Men Relative risks Women No Positive No Positive family family family family Risk factor history history history history Age (5 yr)a 2 0D 1.4D 2.2c 2.8D SBP (20 mm Hg)A l.3c 1j3B 1.3B 0.9 Obesity (0.5 x pounds/inch X 100)A Cholesterol (40 mg%)a 1.5C Personal history diabetes (yes vs no)a c Smoking (yes vs no) B Oc Estrogen (yes vs no) AFor continuous variables except age, intervals represent approximately 1 standard deviation. Bp <.05; Cp <.01; Dp <.001. Vol. 74, No. 2, August 1986 risk factor for cardiovascular death in those with and without a family history of heart attack. Cigarette smoking was highly predictive of mortality in both men and women with a family history of heart attack. We compared risks by testing for significant differences between the regression coefficients for cardiovascular death associated with smoking in those with and without family history of heart attack: in men and in women the relative risks associated with current cigarette smoking habit differed (p =.06) in those with and those without family history of heart attack. In women with a family history of heart attack a personal history of diabetes was a stronger predictor of cardiovascular death than it was in women without such a history; the protective effect of current estrogen use was seen only in women with a positive family history. Finally, smokers and nonsmokers were examined separately (data not shown). Although overall in the group of men 40 to 79 years old independent relative risk of cardiovascular mortality related to a posi- 241

4 K:HAW and BARRETT-CONNOR tive family history of heart attack was 1.5 (p <.05), the risk in nonsmokers was 1.2 (p = NS) and the risk in smokers was 3.3 (p <.01), a significant difference between smokers and nonsmokers (p <.05). Risk of cardiovascular disease associated with a family history of heart attack in women was slightly higher in smokers than nonsmokers, although still not significantly different from 1. The observed cardiovascular death rates in smokers with a family history of heart attack for both sexes were much greater than expected from either an additive or multiplicative model for the separate risks of smoking or family history of heart attack, suggesting a marked interaction; we therefore calculated the excess deaths attributable to the observed interaction between family history and smoking. We used Walker's test16 to estimate the fraction of those with cardiovascular disease among the group of individuals with a positive family history who smoked in whom disease was attributable to the interaction of the factors; this is referred to as the etiologic fraction due to interaction. Using Walker's assumptions in applying the age-specific death rates in nonsmokers without a family history of heart attack we would expect, in the 468 nonsmokers with a family history of heart attack, 31.7 deaths if family history of heart attack conferred no excess risk. Thirty-six deaths were observed, so approximately four ( ) deaths in nonsmokers with a family history of heart attack were attributable to the excess risk conferred by the positive family history. In the 133 smokers with a positive family history, we would expect 5.1 deaths if smoking or family history conferred no excess risk. We would expect 6.2 deaths with a relative risk of 1. 1 for smoking if family history of heart attack conferred no excess risk, and 7.4 deaths with a relative risk of 1.2 for a family history of heart attack, i.e., an expected excess of 1.2 ( ) deaths attributable to family history in smokers. We in fact observed 19 deaths so there was a real excess of 12.8 deaths. Therefore, in the total group with a family history of heart attack, if there were no smoking-family history interaction, we would expect = 5.5 excess deaths due to family history of heart attack; we observed = 17.1 excess deaths. An estimated ( )/17.1 or 68% of the excess deaths due to family history of heart attack could be attributed to the interaction between smoking and family history of heart attack. Discussion Men, particularly those under 60 years of age, with a family history of heart attack are at increased risk of cardiovascular death.'-7 The familial aggregation of risk factors, notably raised blood pressure and hypercholesterolemia, which could be due to both shared environmental and genetic factors, is well recognized. 1-3 However, family history appears to predict cardiovascular mortality independent of the level of these risk factors. In this analysis, the effects of several risk factors differed in those with and those without a family history of heart attack. Although smoking habits also tend to be similar in families, the increased cardiovascular risk of a family history of heart attack in men was unlikely to be due just to familial smoking habits per se. Cigarette smoking was associated with markedly increased risk of cardiovascular mortality in men and women with a family history of heart attack, even though family history of heart attack per se was not associated with increased risk of cardiovascular disease in women. In men, but not women, the independent relative risk of cardiovascular death due to family history of heart attack was significantly greater in smokers than nonsmokers. A possible explanation for the apparent increased risk conferred by smoking in those with a positive family history is that more subjects without a family history of heart attack subsequently stopped smoking. Although we have no follow-up data on this, the similar percentage of exsmokers in the groups of men without (51%) and with a positive family history (54%) makes this explanation unlikely. These results support findings of a study in Utah that reported magnified risks of cigarette smoking in coronary-prone families. 17 However, this was based on telephone and mail survey data and there were no direct measurements of the cardiovascular risk factors. Estrogen use, previously reported in this cohort of women to be associated with more favorable heart disease risk factor status,'" showed a protective effect mainly in women who were current smokers or who had a family history of heart attack. Serum cholesterol also appeared to be a stronger predictor of cardiovascular mortality in men with no family history of heart attack than in those with a positive family history, although differences in relative risk were not statistically significant. At the time of the survey, resources were not available to validate the diagnosis of heart attack in family members, so misclassification of subjects would have occurred. This measurement imprecision would tend to attenuate our results. Some studies have reported differences in risk factors depending on the sex and age of the relative who experienced the heart attack8: we did not have information on which first-degree rela- 242 CIRCULATION

5 PATHOPHYSIOLOGY AND NATURAL HISTORY-CORONARY ARTERY DISEASE tive(s) experienced heart attack, although most would have been in the subject's parents rather than siblings or children. In this cohort, cardiovascular risk was similarly increased in men who reported a family history of heart attack whether the relative's attack occurred before or after he was 50 years of age.7 The known risk factors explain only a proportion of the incidence of cardiovascular disease. The interaction of family history of heart attack with other risk factors may explain discrepancies in predictive strength of various risk factors for cardiovascular disease reported in different studies. For example, since the effect of family history of heart attack is seen predominantly in smokers, one might predict that in populations in which the prevalence of smoking is low, family history of heart attack might not be a strong independent predictor of mortality. Similarly, the protective effect of estrogen was seen mainly in women with a positive family history of heart attack. These results lead to speculation about possible mediating mechanisms. Cigarette smoking severely depresses high-density lipoprotein (HDL) cholesterol and could thus potentiate in a major way an inherited predisposition to low HDL cholesterol, perhaps the most common dyslipoproteinemia associated with early coronary heart disease.'9 This might also explain some of the sex differences: exogenous sex hormones such as estrogen may reduce cardiovascular risk in women, possibly by increasing HDL cholesterol.20 This could counteract the HDL-lowering effect of cigarette smoking in predisposed individuals. We obtained no data on HDL cholesterol at the baseline visit and so could not examine this hypothesis further. Cigarette smoking is also recognized to affect hemostatic mechanisms,21 which are independently associated with cardiovascular disease risk22 and could potentiate an inherited predisposition. These findings have several clinical implications. They confirm previous reports that a family history of heart attack may be a useful marker for identifying men who are more likely to have high levels of blood cholesterol for possible treatment. Second, because family history of heart attack also predicts male cardiovascular mortality independent of known risk factors, men with a positive family history might be more likely to benefit from intervention for any given risk factor of any severity (e.g., mild hypertension). As shown in table 1, a family history of heart attack in men 40 to 59 years old in this population conferred a fivefold increased risk of cardiovascular mortality independent of other risk factors, such as high blood pressure, and thus should identify groups in which the risk-benefit balance might favor intervention. Most important, those with a positive family history appear much more susceptible to the detrimental cardiovascular effects of smoking than those without such history. Conversely, the increase in cardiovascular risk in men associated with a family history of heart attack predominantly affects smokers. An estimated 68% of the excess deaths due to family history of heart attack could be attributed solely to the interaction between smoking and family history of heart attack and they were therefore potentially avoidable. This study suggests that risk of cardiovascular disease associated with an apparently inherited predisposition appears to be profoundly affected by modifiable behavior. This information may be helpful in persuading men and women with a positive family history of heart attack to stop smoking. References 1. Neufeld HN, Goldbourt U: Coronary heart disease: genetic aspects. Circulation 67: 943, Rose G: Familial patterns in ischemic heart disease. Br J Prev Soc Med 18: 75, Phillips RL, Lilienfeld AM, Diamond EL, Kagan A: Frequency of coronary heart disease and cerebrovascular accidents in parents and sons of coronary heart disease index cases and controls. Am J Epidemiol 100: 87, Sholtz RI, Rosenman RH, Brand RJ: The relationship of reported parental history to the incidence of coronary heart disease in the Western Collaborative Group Study. Am J Epidemiol 102: 350, Cambien F, Richard JL, Ducimetiere P: Familial history of coronary heart diseases and high blood pressure in relation to the prevalence of risk factors, and the incidence of coronary heart disease. Rev Epidemiol Sante Publ 28: 21, Snowden CB, McNamara PM, Garrison RJ, Feinleib M, Kannel WB, Epstein FH: Predicting coronary heart disease in siblings- a multivariate assessment. Am J Epidemiol 115: 217, Barrett-Connor E, Khaw KT: Family history of heart attack as an independent predictor of death due to cardiovascular disease. Circulation 69: 1065, Morrison JA, Horvitz R, Khoury P, Laskarzewski P, Gartside PS, Kelly K, Mellies M, Glueck CJ: Parental history of coronary heart disease, hypertension, diabetes and stroke: relationship to coronary risk factor variables in their adult children. Prev Med 9: 773, Deutscher S, Epstein FH, Kjelsberg MO: Familial aggregation of factors associated with coronary heart disease. Circulation 33: 91 1, Nikkila EA, Aro A: Family study of serum lipids and lipoproteins in coronary heart disease. Lancet 1: 954, Garrison RJ, Castelli W, Feinleib M, Kannel WB, Havlik RJ, Padgett SJ, McNamara PM: The association of total cholesterol, triglyceride and plasma lipoprotein cholesterol levels in first degree relatives and spouse pairs. Am J Epidemiol 119: 313, Sosenko JM, Breslow JL, Ellison RC, Miettinen OS: Familial aggregation of total cholesterol, high density lipoprotein cholesterol and total triglyceride levels in plasma. Am J Epidemiol 112: 656, Glueck CJ, Laskarzewski PM, Suchindran CM, Chambless LE, Barrett-Connor E, Stewart P, Heiss G, Tyroler HA: Progeny's lipid and lipoprotein levels by parental mortality. The Lipid Research Clinics Program Prevalence Study. Circulation 73(suppl I): 1-51, Criqui MH, Barrett-Connor E, Austin M: Differences between respondents and non-respondents in a population based cardiovascular disease study. Am J Epidemiol 108: 367, 1978 Vol. 74, No. 2, August

6 KHAW and BARRETT-CONNOR 15. Lee ET: Statistical methods for survival data analysis. Belmont, CA, 1980, Wadsworth, ch 10, p Walker AM: Proportion of disease attributable to the combined effect of two factors. Int J Epidemiol 10: 81, Hopkins PN, Williams RR, Hunt SC: Magnified risks from cigarette smoking for coronary prone families in Utah. West J Med 141: 196, Barrett-Connor E, Brown WV, Turner J, Austin M: Heart disease risk factors and hormone use in post-menopausal women. JAMA 24: 2162, Ordovas JM, Schaefer EJ, Salem D, Ward RH, Glueck CJ, Vergani C, Wilson PWF, Karathanasis SK: Apolipoprotein A-I gene polymorphism associated with premature coronary artery disease and familial hypoalphalipoproteinemia. N Engl J Med 314: 671, Bush TL, Barrett-Connor E: Non-contraceptive estrogen use and cardiovascular disease. Epidemiol Rev 7: 80, The Health Consequences of Smoking the changing cigarette: a report of the Surgeon General. Washington, DC, 1981, Department of Health and Human Services 22. Meade TW, North WRS, Chakrabarti R, Stirling Y, Haines AP, Thompson SG: Haemostatic function and cardiovascular death: early results of a prospective study. Lancet 1: 1050, 1980 Erratum In an article by Gibson et al. appearing in the June issue of Circulation (73: 1186, 1986), several units of measure were in error. Under the heading "Electrocardiographic evaluation" in the Methods section, in the sentence beginning "Patients with minimal-amplitude R waves (i.e., 0.25 mv or less)." the value should have been 0.25 mm. Also, in tables 2 and 4, all of the millivolt (mv) units should have been millimeters (mm). 244 CIRCULATION