Tobacco Documents Online

REVIEW Risk Factors and the Sex Differential in Coronary Artery Disease Jacqueline F. Price and F. Gerald R. Fowkes In industrialized countries, the incidence of coronary artery disease is three to four rimes higher in men than in women. Our review examines whether differences in the prevalence of, or susceptibility to, various coronary risk factors might con- tribute to this sex differential. Cigarette smoking, hyperten- sion, and hypercholesrerolemia are ri~k factors for coronary artery disease in both sexes and are present at higher levels in middle-aged men than women. To date, such differences have failed to explain the coronary sex differential, although higher biological susceptibilities to smoking, blood pressure, and total cholesterol in men may also be important. Mean levels of protective high-density-lipoprotein choles- terol are higher in women than in men throughout life and may also contribute to the sex differential. In addition, men with a female pattern of abdominal fat distribution are at lower risk of coronary artery disease than those with a male pattern. In diabetic populations, the sex differential is greatly reduced, but studies on the effects of hyperinsulinemia and a lower imulirt resistance in women compared w~th men are scarce. Prospective studies on the effect of fibrinogen and other he- mostatic factors in women are also required. (Epidemiology 1997;8:584-591) Keywords: coronary artery disease, sex differential, risk factors. Coronary artery disease is the leading cause of death for both men and women in industrialized countries. At any given age, however, coronary mortality is much higher in men than in women. Estimates vary between coun- tries, but a three- to fourfold excess in men is commonly found) The sex differential narrows with advancing age, mostly owing to a declining rate of increase in male mortality, but the death rate in women never exceeds that in men (Figure 1). A male predominance is also found in the incidence of myocardial infarction and total coronary artery disease,zo and the degree of coro- nary atherosclerosis in women at autopsy lags behind men by 10-15 years: It is generally believed that the sex differential in coronary artery disease results from a combination of genetic and life-style factors that are thought to deter- mine disease via risk factors, such as hypercholesterol- emia and hypertension. The aim of this review is to explore potential causes for the sex differential in coro- nary artery disease in terms of sex differences in several such risk factors, including smoking, hypertension, se- rum lipids and lipoproteins, abdominal fat distribution, From the Wolfson Unit for Prevention of Peripheral Va~:ular Dim, Depart- men¢ of Public Health Sciences, University of F.dinburgh, ~dinburgh, United Kingdom. Address correspondence to: Jacqueline F. Price, Wolfson Unit for Prevention of Peripheral Vascular Diseases, Department of Pubhc Health Sciences, Universit~ of Edinburgh, Teviot i~ace, F.dinburgh, EH8 9AG, United Kingdom. Submitted August 26, 1996; final version accepted Janua~ 27, 1997. © 1997 ~ F.pidemiolo~/Resources Inc. hyperinsu[inemia, and hemostatic factors. These are ei- ther well established coronary risk factors or have been suggested as contributors to the sex differential in the medical literature, but the list is by no means compre- hensive. In particular, steroid sex hormones are not discussed, mainly because they have been reviewed ex- tensively elsewhere,1a-7 but also because much of the epidemiologic evidence suggests that they influence dis- ease by affecting the levels of risk factors that are in- cluded in the review. Our review of the literature involved a comprehensive on-line literature search (including MEDLINE) for epi- demiologic studies on coronary artery disease that in- cluded both men and women. Studies on a single sex have occasionally been cited where no equivalent study on both sexes was available. Wherever possible, studies reviewed were prospective investigations of the general population. We excluded animal studies and primary or secondary prevention trials. We also restricted studies to white populations, since the sex differential is known to be lower in blacks than in whites at all ages, and discus- sion of the possible reasons for this disparity was beyond the scope of the review. Finally, where disease rates are compared between men and women, these are age ad- justed unless otherwise stated. Principles of Prevalence and Susceptibility The sex differential in coronary artery disease might be influenced by two "properties" of risk factors. First, the level (or prevalence) of risk factors in the population may vary between the sexes. Atherogenic risk factors that are more prevalent in men or protectivd factors that 584 This article is for individual use only and m~y not be further reproduced or stored electronically without writ-ten permission I~om the copyright holder. Unauthorfzc~I repro~uctfon may rebait t'a ffnaa~fal ~nd other pcn~flt'~i~. (~) WILLIAM~ d; WILKIN5

Epidemiology September 1997, Volume 8 Number 5 CORONARY SEX DIFFERENTIAL 585 I0000 5O00 1000 500 50 FIGURE 1. Age-specific coronary disease death rates for men and women. (Reprinted with permission from Kalln MF, Zumoff B. Steroids 1990;55:330-352.1) are more prevalent in women would be expected to contribute to the higher incidence of disease in men than in women, whereas lower levels of atherogenic factors in men would be expected to result in a narrow- ing of the sex differential. It is also possible that women could be biologically protected from a particular risk factor compared with men, resulting in greater su~eptibility of men or resis- tance of women to the same level of the risk factor. For example, in a group of men and women with the same lifetime smoking exposure, men might develop more disease ff smoking is especially damaging in males. It is not immediately evident how such differences in suscep- tibility might occur biologically, since the exact patho- physiologic mechanism by which many risk factors pro- mote disease is unknown. One possibility is that interaction with other risk factors results in a relative protective effect in women, but ff this were the case, we would expect higher levels of such protective factors in women. Alternatively, some inherent biological proper- ties may favor women, for example the geometry or vessel wall composition of their blood vessels. Thus, to explain the sex differential in coronary artery disease in terms of recognized risk factors, men must demonstrate a higher prevalence of, or greater suscepti- bility to, atherogenic risk factors. The potential effect of different risk factor prevalences on the sex differential can be demonstrated at a simplistic level by comparing groups of men and women with the same risk factor level. If the sex differential is caused solely by different prevalences of the risk factor in question, then such men and women should have the same incidence of disease. It is unlikely, however, that the sex differential is caused by a single risk factor, and it is more probable that the sex ratio will only narrow in this situation. Two statistical measures of risk associated with a given risk factor may be related to biological suscepti- bility. The first of these is the relative risk of disease in subjects exposed to a given risk factor compared with those not exposed. A higher relative risk of disease in women than in men may simply reflect their very low disease rates in the bottom group for each risk factor.- It is therefore important to consider the absolute risk in- crease, or attributable risk, for subjects exposed vs those not exposed. Smoking, Hypertension and Total Serum Cholesterol More than 10 large-scale, prospective studies throughout Europe and America have consistently found a higher incidence of fatal and nonfatal coronary artery disease in smokers compared with nonsmokers in both men and women.8-z~ In the majority of studies, smoking more than doubled the risk of coronary artery disease, inde- pendent of other coronary risk factors, such as blood pressure, total serum cholesterol, and body mass index. Similarly, hypertension and total serum cholesterol have been found to be independent risk factors for coronary artery disease in both sexes ~ numerous~ prospective studies within a variety of differ~t countries,ts,ls'l°~l'2~-2s PREVALENCE In industrialized countries, both the percentage of smok- ers and the number of cigarettes smoked is higher in men than in women, although this difference has narrowed since the 1920s.z~'3° Mean systolic blood pressure is also higher in young and middle-aged men than in women, although after the menopause the situate.on is rev'etsed.31 Diastolic pressures do not differ substantially between the sexes,3z but the prevalence of hypertension accord- ing to traditional criteria is also higher in men until the age of 50 years, after which it predominates in wom- en.3z-34 Men's mean total cholesterol level~ are higher than women's, but again, this situation is reversed after the menopause.3s The higher levels of smoking, hypertension, and total cholestero| among middle-aged men than women could potentially contribute to the coronary sex differential. Accordingly, in Doll and Pero's survey of British Doc- tors,s'9 the difference in death rates from ischemic heart disease in men compared with women was slightly lower in nonsmokers than in the total population. Neverthe- less, the rate of coronary mortality in lifelong nonsmok- ers was still markedly higher in men than in women, resulting in a male/female ratio in coronary mortality of 5.3 for nonsmokers less than 65 years of age.s'~ Similarly, for any given level of systolic and diastolic blood pres- sure, women still have at least a two to three times lower risk of death from coronary artery disease than men~ (Table I) and, for any given total cholesterol level, men still have coronary mortality rates above those of wom- en.36 For example, in Scotland, the difference in coro- nary mortality among 45- to 64-year-olds was such that women in their top quintile of cholesterol (greater than

586 PRICE AND FOWKES Epidemiology September 1997, Volume 8 Number 5 TABLE 1. Age-Adjusted Rates and Relative Risks of Coronary Artery Disease (CAD) Mortality for Different Levels of Blood Pressure in the Chicago Heart A~sociation Detection Program in Industry Studyzs Systolic/Diastolic Blood Premure CmmHg) CAD Mortality per Relative Risk I0,000 Person-Years* CAD Mortality~" Men Women Men Women <120/<80 7.9 1.8 1.00 1.00 120-139/<80 19.0 8.8 2.40 4.82 120-139/80-89 24.8 12.8 3.15 7.05 140-159/<80 32.1 11.0 4.07 6.04 140-159/80-89 35.0 14.6 4.43 8.02 140-159/90-99 31.8 12.2 4.03 6.68 140-159/100+ 51.1 25.9 6.47 14.23 160+/80-89 40.0 18.6 5.07 10.24 160+/90-99 58.7 22. I 7.43 12.49 160+/100+ 73.4 36.0 9.30 19.79 White men and women ages 35-64 year*. Risk of coronary arter~ dt~ease rel~t'ive to blocd pressure <120/<80. TABLE 2. Male/Female Ratio in Incidence of Coronary Artery Disease before and after Adjustment for Mean Levels of Coronary Risk Factors Male/Female Ratio Age Follow-up Before After Risk Factorst Stud*/ (Yearn) (Years) Outcome* Adjustment Adjustment Adjusted for. Johnson,~ 1977 45-54 18 CAD 55-64 18 CAD 65-74 18 CAD Wingard?9 1982 30-69 9 CAD mortality Wingard et d,~° 1983 30-69 8 CAD mortality Larmon et 0/?7 1992 54 12 Fatal and nonfatal MI Jan~horbani et al,z4 45-64 10-14 CAD mortality 1993 2.2 4.8 3.2 2.8 -21%~: Smoking, sBP, dBP, TC, LVH, GI +31%~: Smoking, sBP, dBP, TC, LVH, GI +25%:1: Smoking, sBP, dBP, TC, LVH, G1 2.3 Age, hypertension, smoking, diabetes) weight§ 2.4 Smoking, sBP, TC, B~, obesity, marital status, education 3.1 Smoking, dBP, TC, BMI 3.4 Age, smoking, sBP, dBP, TC, BMI, LDL, HDL * CAD m cotonar/heart disease; MI z myocardial infurction. ~" sBP ~ systolic blood pressure;' dBP m diastolic blood prmsere; TC m total cholesterol; LVH ~ left ventricuIar hypertrophy; GI = glucose intolerance; BG = blood gtt~ose; BM[ ~ body mass index; HDL = high-demiw-lipoptotein cholasrerol; LDL ~ low-density.|ipoptotein cholesterol. ~: Alteration in ~x difference if meh a~lsned women's mean risk f~ctor levels. | Self-reported risk factor parameten. Aho adjusted for race, income, che~t pain, heart trouble, alcohol intake, physical activity, sleeping panerc~, and l~ychomcial variable~. 7.2 mmol per liter) had lower mortality than men in their bottom quintile (less than 5.0 mmol per liter),t6 The extent to which differences in risk factor preva- lence might influence the sex differential has been in- vestigated in prospective studies by adjusting the male/ female ratio in incidence of coronary artery disease for mean levels of a variety of coronary risk factors, includ- ing smoking, hypertension, and total cholesterol. In general, this adjustment had little effect on the ra- tioz4'37-~° (Table 2). After adjustment, a two- to three- fold sex difference remained, and in at least one study,z4 adjustment increased the sex differential, implying that sex differences in the prevalence of smoking, hyperten- sion, and total cholesterol have little or no effect on the coronary sex differential. The majority of these studies, however, involved predominantly postmeno. pausal subjects, in whom we would expect the levels of these risk factors to be higher in women than in men. If the same adjust- ment were made in younger populations, when they are higher in men, the sex ratio would be more likely to narrow. It is therefore still possible that higher prevalences of smoking, hypertension, and hypercholesterolemia could have a greater im- pact on the sex differential than indicated by these studies, although they are unlikely to explain the en- tire differential. SUSCEPTIBILITY Relative risks of heart dis- ease for smoking (Table 3), hypertension2~,2s (Ta- ble 1), and total cholester- ol~ have all been found to be little different between the sexes, or slightly higher in women. The at- tributable risk of disease associated with each risk factor, however, is gener- ally higher in men than in women. For example, in Doll and Peto's survey of British doctors, for men smoking 25 cigarettes or more per day compared with nonsmokers, the yearly death rate increased by 239 per 100,000, whereas in women, the death rate increased by only 54 per 100,000.s'~ Although this sex difference may be influenced by the coexistence of other coronary risk factors in smokers, a similar sex difference in Scottish men and women was independent of age, total cholesterol, diastolic blood pressure, body mass index, and social chass,ie An increased susceptibility of men to cigarette smoking may be due to their increased propensity to inhale cigarette smoke and to smoke more of each cigarette compared with women~z or to some intrinsic biological difference between the sexes. Differences may also exist between men and women in the increased risk associated with hypertension and total cholesterol. Thus, the increase in rate of coronary

Epidemiology September 1997, Volume 8 Number 5 CORONARY SEX DIFFERENTIAL 587 TABLE 3. Prospective Studies of Smoking and Coronary Mortality in Men and Women Relative Risks of Coronary Mortality* Number and Gender Age Follow-up Study of Subjects (Yca~s) (Years) Men Women Factors'i" Controlled for: United Kingdom"" 30,189 d" <65 20--22 2.1 ?-.5" 5,831 ? 2.5 2.7s 4,251 d >65 20-22 1.2 2.2~ 363 9 1.5 2.8h California~0 1,129 ~ 35-54 11 4.9 5.4' 1,484 ~ Rancho 8emardoLz 1,089 d' 40-79 7 0.4 1.5a sBP, TC, QI, DM 1,358 ~ LRCt3 4,105 d" >30 Mean 8.5 1.8 3.1~ sBP, HDL, LDL, TG, QI 3,356 ? Alameda Countyt4 1,699 d ~40 9 1.6 1.4t Hypenem~on, demographic, 2,052 ~? and behavior factors Tecumseh~s 1,057 8 4.5:-64 18-21 1.6 2.6 sBP, TC, BMI, Ol 1,032 9 Scotland~6 7,137 d' 45-64 15 2.0 3.4* TC, dBP, BMI, SC 8,262 q No~wa~a~ 44,290 ¢~ 35-49 13 1.2 1.8h TC, sBP, BMI 24,535 ~ Finlandts5,641 d 30-49 15 3.0 2.8~ BMI, sBP, TC 5,706 ~ 2,099 ~ " 50-59 15 1.9 3.8~ BMI, sBP, TO 2,667 ~ * Relative risks for smokers us nommokers: • 15-24 cigs/da~/*as never-smoker; b >75 cigs/day ws never-smoker; "current and ex-smoker,~s never.smoker; a Cox regression ¢oefficicnt';" 20 cigs/day vs 0 cigs/day; ¢ current vs never-smoker; ~ > 15 cigs/da¥ ~s never-smoker; h I0 cigs/day increase (Cox model); ' cur~enr-/recenr-smoker vs never-/ex-smoker. ~" sBP = s~totic blood pressure; TC - total chole-steroi; HDL - high-dcnsity-lipoprocein cholesterol; LDL = low-densit~- lipoprotem cholesterol; TG ~ trigl~eride~; DM ~ d~abcres mc[[i~us; BMI - body mass index; Ql ~ Quetdes index; SC - ~ocial class; GI ~ glucose inmlcmnce. mortality was only 34.2 per 10,000 person-gears in women, compared with 65.5 per 10,000 person-years in men with blood pressures of 160+/100+ compared with an optimal blood pressure of <120/<80 (Table 1). The attributable risk of coronary morvality associated with raised cholesterol was also greater in men than in women, even after adjustment for potentially confound- ing risk factors such as body mass index, diastolic blood pressure, smoking, and social class?3 If the sex differential were due to increased male susceptibility to a given risk factor, then comparison of men and women in whom that risk factor is extremely low or absent should indicate that such men and women have similar rates of disease. When men and women with low levels of a variety of risk factors were compared (including nonsmoking, systolic blood pressure less than 105 mmHg, and total cholesterol less than 185 gm per dl), the difference in incidence of coronary heart disease between men and women was low (9 per 10,000).44 Nevertheless, even at the very low disease rates occur- ring at these risk factor levels, a male/female ratio of almost 4:1 persisted (incidence 15 per I0,000 in men and 4 per 10,000 in women), implying that other influ- ences were affecting the sex differential.44 Low.Density and High,Density-Lipoprotein Cholesterol Low-density-lipoprotein (LDL) cholesterol, which is highly correlated with total serum cholesterol, has been related to the risk of coronary artery disease in both sexes, although this rela- tion has been less easy to demonstrate in women than in men. Two major prospective studies have measured LDL cholesterol in both sexes, the Fram- ingham study~-46 and the Lipid Research Clinics (LRC) follow-up study.~3,~¢ Gordon4s calculated a 0.8% risk increment in coronary artery disease per 1% increment in LDL cholesterol in men and a corresponding 1.9% in- crement in women in the Framingham study, whereas in the LRC study, corresponding values were 1.9% and 1.4% in men and women, respectively.4s Four major prospective studies, the . Frarningham study,~9'4s'49 the LRC follow- up study,4¢ The Norway Na- tional Health Screening Service studyz*,ze and the Donolo-Tel Aviv study,~° have demonstrated an inverse relation betweem the risk of coronary artery disease and high-density-lipoprotein (HDL)cholesterol leveLs in both men and women. In Ftamingham, there was a 1.90% decrement in inciqtence of coronary heart disease per 1 mg per dl increment in HDL in men and a 3.24% decrement in women,s~ The LRC study demonstrated a 3.60% decrement in incidence of cardiovascular mortality per 1 mg per dl increment in HDL in men and a 4.72% decrement in women.5~ Despite these higher percentage decrements in women, for a similar drop in HDL cholesterol, the absolute reduction in risk was fairly equal between the sexes, or greater in men. These data were not adjusted for confounding risk factors. LDL cholesterol levels are higher in men than in women but only until the menopause, after which the situation is reversed,sz Mean HDL cholesterol levels, on the other hand, are greater in women throughout adult life.5~ Although this higher level of protective HDL cholesterol in womert may contribute to rkie coronary sex differential, for any given HDL chdlesterol level, women still have considerably less risk of disease than men.z~'~¢'~'5° Framingham data indicate that the total/ HDL or LDL/HDL cholesterol ratio most accurately reflects art individual's risk of coronary artery disease.53 As the LDL/HDL cholesterol ratio is greater in men than in women throughout their lifespan)~ it is conceiv- able that this ratio explains the coronary sex differential. In Framingham, however, women reached the same in- cidence of coronary artery disease as men only at ratios greater than 7.5, an unusually high lipoprotein index.~z

588 PRICE AND FOWKES oWom~n n M~ 0.6 (~7 0.8 0,9 1 1-1 Waist tohipr~tio FIGURE 2. Twelve-year incidence of coronary heart dis- ease by waist/hip ratio and sex (women ages 50, 54, and 60 years at baseline and men ages 54 years at baseline). (Re- printed with permission from Larsson B, et d. Am J Epide- mid 1992;135:266-273J?) At ratios less than 7.5, women still had considerably less risk of coronary heart disease than men.5z Abdominal Fat Distribution It has been known for several years that men have more abdominal fat than women,ss More recently, this so- called android (vs gynoid) fat distribution was shown to be a risk factor for coronary artery disease in both men and women,s6'57 Thus, men with a more "female" distri- bution of adipose tissue had comparatively less coronary artery disease than those with a typical "male" distribu- tion, and v/ce versa.58 Although the waist/hip ratio is often used to assess degrees of abdominal fat, the results of computed tomography indicated that it was more likely to be the amount of intra-abdominal or visceral fat that predicts coronary artery disease.59,~° Adjustment for waist/hip ratio has been shown to reduce the male/female odds ratio for incidence of cor- onary artery disease from 3.2 to 1.4.3r Further adjustment for diastolic blood pressure, total cholesterol, body mass index, and smoking reduced this odds ratio to 1.1. Un- fortunately, there was very little overlap in waist/hip ratios between the male and female populations in this study; therefore, it was impossible to analyze meaning- fully this subsection of the population. Thus, although a difference in waist/hip ratio could statistically explain the coronary sex differential, it may simply have been another (indirect) way of determining sex. This latter theory does not explain the association between waist/ hip ratio and coronary artery disease within each sex group nor the suggestion of a graded association almost independent of sex (Figure 2). Notwithstanding the methodologic problems associ- ated with the above study,37 it is possible that determin- ing the relation between abdominal fat distribution and other coronary risk factors could aid our understanding of the coronary sex differential. In this respect, visceral fat accumulation has been frequently associated with raised LDL cholesterol and triglycerides, reduced HDL cholesterol, hypertension, hyperinsulinemia, and insulin resistance.~t Furthermore, matching men and women Epidemiology September 1997, Volume 8 Number 5 according to waist/hip ratio greatly reduced sex differ- ences in triglycerides, HDL cholesterol, and apolipopro- reins A-I and B.6~ In general, however, it is too early to draw any firm conclusions about the role of body fat distribution in the coronary sex differential. Hyperinsulinemiaflnsulin Resistance Among many diabetic populations, the sex difference in coronary artery disease appears to be grossly reduced or eliminated.Kz'~3 Furthermore, diabetes mellitus has been shown to be a stronger risk factor for fatal coronary artery disease in women than in men, even after con- trolling for hyperglycemia and other coronary risk fac- tors frequently associated with diabetes.~ Such observa- tions have stimulated interest in two factors, intrinsic to diabetic populations, that may contribute to the coro- nary sex differential, namely hyperinsulinemia and insu- lin resistance. HYPERINSULINEMIA In most,6~-¢° but not all,n cross-sectional and prospective studies, raised fasting and/or post-glucose plasma insulin levels were associated with coronary artery disease in men. Of only three major cross-sectional studies to in- clude women, two found fasting~s or post-glucose~ in- sulin levels to be indicators of coronary artery dLsease in both sexes. The third showed an excess coronary risk associated with hyperinsulinemia in men only.~ In the single major prospective study to include both genders, hyperinsulinemia did not affect coronary mortality in either sex.7z A smaller prospective study in Nauru sug- gested an association between insulin and the subse- quent development of ischemic electrocardiogram changes in both sexes)3 In addition to a possible direct atherogenic effect on blood vessel walls,TM insulin levels have been associated with hypertension,~,~.7~ raised total cholesterol, LDL cholesterol and triglycerides and reduced HDL choles- terol?~,7~ as well as android obesityTM in both sexes. In- deed, Reaven7~ suggested that hyperinsulinemia under- lies the clustering of these coronary risk factors in individuals at high risk of atherosclerotic cardiovascular disease (syndrome X). More recently, Fontbonne8° pos- tulated that hyperinsulinemia is a marker of syndrome X and therefore of increased cardiovascular risk in men only, owing to its relation with android obesity. As discussed above, however, the finding on which this theory was based (namely that hyperinsuIinemia is a coronary risk factor in men but not women6;) has since been challenged.6s,~9 Any rote for hyperinsulinemia and/or syndrome X in the cardiovascular sex differential therefore remains speculative. INSULIN R~SISTANCE Although women tend to have equivalent or higher plasma insulin levels than men,69,s~ for a given quartile of insulin, they have lower serum glucose concentrations (Figure 3). This relation is consistent with a greater insulin sensitivity (lower insutin resistance) in women, a

Epidemiology September 1997, Volume 8 Number 5 CORONARY SEX DIFFERENTIAL 589 88- 80- .,~Men Women I I ~ I II Ill N M~ies(n) 23 2/, 20 20 l~enge 2.1-6.3 ~,J,-8.7 ~B-12.112.2-~0.7 Females{n} 18 22 21 22 l~enge 2.0--3.q /,.O-fig Z0-12.012.1-30.0 ~erti[e of inso[in ( oO/m[} FIGL~E 3. Glucose concentration (mg per dl) plotted by quartile of insulin concentration (/~U per m|), by sex, among 170 young adults ages 20-24 years. (Reprinted with permis- sion from Donahue RP, et aL Am J Epidemlol 1987;125: 650-657.s~) finding that has been demonstrated direcdy using the euglycemic clamp,sz Donahue et a/s~ suggested that it is insulin resistance per se, rather than the associated hy- perinsulinemia, that is the important predictor of coro- nary artery disease, and that intersex differences in in- sulin resistance could therefore contribute to the coronary sex differential, lndeed, in a recent study,~4 coronary artery disease was directly associated with in- sulin resistance in men. There is no comparable study in women, however, and litde other evidence supports Donahue's theory at present. -The observation that coronary risk is higher among men and diabetic women compared with nondiabetic women has led to an alternative theory: that nondia- betic women respond differently to insulin resistance (and/or hyperinsulinemia) than do men and diabetic women, for example owing to the android obesity and androgenicity associated with male sex and diabetes mellitus in women.~ Hemostatic Factors Plasma fibrinogen has been found to be an independent risk factor for coronary artery disease in both men and women,s5,~6 The relation between fibrinogen and disease was at least as strong as that between disease and the conventional cardiovascular risk factors (smoking habit, diastolic blood pressure, and LDL cholesterol),a7 In Framingham, both the relative risk of coronary artery disease and the absolute increase in risk of disease, in the highest tertile of plasma fibrinogen compared with the lowest, was higher in men than in women, although a substantial difference in risk of disease persisted within each tertile,s5 Although a variety of other hemostatic and coagula- tion factors have been implicated in the pathogenesis of coronary artery disease in men, including factors VII and VIII, yon Willebrand factor, tissue plasminogen activa- tor, plasminogen activator inhibitor, and platelet aggre- gability,~ very little is known about-the relation be- tween hemostatic factors and heart disease in women. Raised levels of plasminogen activator inhibitor-1s9 and yon Willebrand factor86,s9 have been found in male and female myocardial infarction survivors. Nevertheless, a number of hemostatic variables, including factors VII and VIII and yon Willebrand factor, were not indepen- dently associated with the risk of carotid atherosclerosis in a large case-control study involving both men and women?° Unfortunately, the results of these Studies were not analyzed separately according to gender. In keeping with their higher rate of coronary artery disease, it might be expected that men have higher levels of hemostatic factors than women. In general, however, women have slightly higher levels of fibrino- gen, factors VII and VIII, and platelet aggregability than men.9~-9~.Results for von Willebrand factor are conflict- ing; some investigators have found higher levets93 and others lower levels in women (Lowe GDO, Rum~ey A, Lee AJ, Tunstall-Pedoe H, Hubbard AR, manuscript in preparation). Plasminogen activator inhibitor-1 levels were higher in men up to 50 years of age when the situation was reversed.°4 Thus, information on the relation between hemo- static factors and cardiovascular disease in women is sparse. The data indicate that, if anything, women are likely to be hypercoagulable compared with men. Conclusion There are many potential causes for the sex differential in coronary artery disease, of which only a few have been discussed in this review. No single risk factor, nor com- bination of risk factors, has been shown to explain fuIiy the coronary sex differential. Nevertheless, differences between men and women in their susceptibility to con- ventional coronary risk factors, as well as differences in other, less well established risk factors, may prove to be important. References 1. Kalin MF, Zumoff B. Sex hormones and comna~ disease: a revte~v of the climcal studies. Steroids 1990;55'.330-352. 2. "Wiihelmsen L Epidemiology of ¢o~naw heart disease in young women. In:

590 PRICE AND FOVv~<ES Epidemiology September 1997, Volume 8 Number 5 Oliver MF, ed. Comnaty Heart Disease in Young Women. Edinburgh: C~urchill Livingstone, 1978;3-11. 3. Dawber TR. Incidence of cornnaty heart disease, stroke and peripheral arterial disease. In: The Fmminghara Study; the Epidemiology of Athero- sclerotic Disease. Cambndge, MA: Harvard Universiw Press, 1980;59-75. 4. Sternby NH. Athermcletosis in a defined population: an autopsy study in Matron, Sweden. Acta Microhiol Stand 1968;194(suppl):61-110. 5. Bush TL, Bartert-Connor E. Nonconttaceptive estrogen use and cardiovas- cular disease. Epidemiol Rev 1985;7:80-104. 6. Gods[and IF, W~/nn V, Crook D, Miller NE. S~x, plasma lipoproteins and asherosclerosis: prevailing a.~mmptions and outsmdding questlons. Am Heart ) 1987;114:1467-1503. 7. Stampfer MJ, C~lditz GA. E~trogen mplacemant therapy and coronary heart disease: a quantitative assessment of the epidemiologic ev~dance. Prey Med 1991;20:47-63. 8. Doll R, Grey R, Hafner B, Peru R. Mortality m relation to smoking: 20 years' observation on female British doctors. BMJ 1980;2:967-971. 9. Doll R, Peto IL Mortality in relation to smoking: 20 years' observation on male British doctors. BM.I 1976;2.:I 52.5-1536. 10. Friedman GD, Dales LG, Uty HK. Mortality in middle-aged smokers and nnnsraokets. N Engl J Med 1979;300:213-217. 11. Hammond CE. Smoking in relation to the death rates of one millio~i men and women. In: Haemzel W, ed. Epidemiologic Approaches to the Study of Cancer and Other Chronic Diseases. Natl Cancer Inst Monogt 1966;19: 127-204, 12. Barrert-Comaor F., Wingard DL. Sex differential in isehemic heart disease mortality in diabetes: a prospective popuhtion-based study. Am J Epidemiol 1983;118:489-496. 13. Criqui MH, C~v~m LD, Tyrnler HA, Bengfliwala S, Heiss G, Wallace RB, Cohn R. Lipoptotains as mediators for the effects of alcohol consumption amt c~garette smoking on cardiovascular mortality: results from the L@id Research Clinics Follow-up Study. Am J Epidemiol 1987;126:629-637. 14. Wingard DL, Cohn BA. Corona~ heart disease mortality among women in Alameda County, 1965 to 1973. Im Faker ED, Packard B, Wenger NK, Clarkson TB, T~axfler HA, eds. Coronary lteart Disease in Women. New York: Haymarker-Do~/~,aa, 1987;99-105. 15. HiggL~ M, Keller JB. Cholesterol, coronaw heart disease, and total mortal- ity in middle-aged and elderly men am] women in Tecuruseh. Arm ~pide- mini 1992;2:69-76. 16. Isles C.G, Hole D~, Haw[home VM, Lever AF. Relation between corona~ risk and corona~ mortali~ in women of the Ranfrew and Paisley survey: comparison with mere Lancet 1992-,339:702-706. 17. Tverdal A, Thelle DS, Stansvold I, Leren P, Biartveit K. Mortality in rein[ton to smoking history: 13 years' follow-up of 68000 Norwegian men and women 35-49 years. J Clin Bpidemiol 1993;46:475-487. 18. Jousilahti P, Tuomilehto J, Vartiainen E, Pekkanen J, Puska P. Body weight, cardiovascular risk factors, and coronaw mortality: 15-year follow-up of middle-aged men and women in Eastern Finland. Circulation 1996;93:1372- 1379. 19. Castelli WP, Garrison RJ, Wilson PWF, Abhor[ RD, Kalousdian S, Kannel WB. Incidence of cosonary heart disease and lipopmrein cholesterol levels. JAMA i986;2.56:2835-2838. 20. Harris T, Cook EF, Kannal WB, Go[dma~ L. Proportional hazards analysis of risk factor~ for coronary hair[ disease in individuals aged 65 or okler, the Framingham Heart Study. J Am Geriart Soc 1988;36:1023-102.8. 21. Nj¢lsrad I, Arnesen E, Lond-Larsen PG. Smoking, serum tipids, blood pressure and sex differences in myocardial infarction. Circulation 1996;93: 450-456. 22. Nvhce J, Jet,sen. G, Appleyard M, Schnohr P. Smoking and the risk of first acute myocardial infarction. Am Heart J 1991;122:438-447. 23. Lam.m MG. Assessment of cardiovascular risk ~'acrors in the elderly: the Framingham Heart Stud~/. Star Meal 1995;14:1745-1756. ~4..hrtghorbanl M, Hedley A], .~oaes RB, Zb.ianpoux M, Gi~anour WH. Get~der differential in all-cause and cardiovascular disease mortality. In[ J Epidemiol 1993;22.:1056-1063. 25. Tyroler HA, He,den S, Bartd A, Cassel J, Comoni JC, Homes CG, Kleinbaum D. Blood pressure and cholesterol as coronary heart disea~ risk factors. Arch Intern Meal 1971;128:907-914. 26. Stensvold I, Urdal P, Thtlrmec H, Tverdal A, Lund-Larsen PG, Fuss High-density lipoprutein cholesterol and coronary, cardiovascular and alt came mortality among mlddle.aged Norwegian men and women. Ear Heart J 1992;13:I155-1163. 2.7. Stensvold L Tverdal A, Urdal P, Gmff-lversen S. Non-fasting serum triglyo- eride concentration and mortality from coronary heart disease and any cause in middle aged Norwegian women. BMJ 1993;307:1318-1322. 28. Stamler J, Stamhr R, Neaton JD. Blood pressure, sysrolic and diastolic, and cardiovascular risks: US population data. Arch Intern Med 1993;153:598-~ 615. 29. Tobacco Advisory Council. Smoking staststics ~ited in: Wald N, Ni- cholaides-Bouman A, eds. UK Smoking Statistics. Oxford: Oxford U~ver- sity Pros, I988. 30. Ot:~ce of Population Censuscs and Surveys. General Household Survey, 1992. London: Her Majesty's Stationery Office, 1994. 31. Breeze E, Maidment A, Bennett N, Flatley j', Carey S. Health Survey for England, [992: a Survey Carried Out by the Social Services Division of OPCS on Behalf of the Department of Health. London: Her Majesty's Stationery Office, 1994. 32. gachariah PK, She~ SG, Bailey KR, Wiltgen CM, Moor~ AG. Age-related characteristics of ambulatory blo~xt pressure load and mean blood pressure in normotensive subjects. JAMA 1991;265:1414-1417. 33. Kannel WB, Dawber TR, McGhee DC.. Perspectives on s'~stolic h~erten- sion: the Framingham Study. Circulation 1980;61:1179-1182. 34. Cornoni.Huntley J, Lecroix AZ, Havlick RJ. Race and sex differentials and the impact of hypertension in the United Status: the National Health and Nutrition Survey I Ep~demiologic Follow-Up Study. Arch b.tern Med 1989; 149:780-788. 35. Mann Jl, LeWLS B, Shepherd J, Winder AF, Fenster S, Rose L, Morgan B. Bkmd lipid concentrations and other cardiovascular risk factors: distribution, prevalance and detection in Britain. BMJ I988;296:1702-1706. 36. Kronraal RA, Cain KC, Ye Z, Omerm GS. Total serum cholesterol levels and mortality nsk as a function of age: a report based on the Framingham data. Arch Intern Med 1993;153:1065-1073. 37. Larsson B, Bengtsson C, Bj6rnts~rp P, Lapidus L, Sj6srt6m L, Sv~'~:lsudd K, Tibblin G, Wedal H, Welin L, Wilhelmsen L. ls abdominal fat distribution a maior explanation for the ~ex difference in the iacidance of myocardial infarction? Am J Epidemiol 1992;135:266-?.73. 38. John.son A. Sex differentials in coronary heart disease: the explanatory role of primary risk factors. ] Health ~ Behav 1977;18:46-54. 39. Wingard DL. The sex differential in mortality rates: demographic and behavioral fhcturs. Am .I Epidemiol 1982;115:205--216. 40, Wingard DL, Suarer L, Bartert-Connor E. The sex differential in mortality from all causes and ischemic heart disease. Am J Epidemio[ 1983;117:165- 172. 41. Manolio TA, Pearson TA, Wenger NK, thrtert-Connor E, Pa~e GH, Harlan WR. Cholesterol and heart dtsease in older persons ~ women: review of an NHLBI Wo~hop. Ann Epidemiol 1992;2:161-176. 42. McGill HC, Stern MP. Sex and arheroselero~is. Athetos¢lerosis Rev 1979; 4:.157-?.42.. 43. Isles CG, Hole DJ, Gillis CR, Haw[home VM, Lever AF. Plasma choles- terol, coronaW heart disease, and cancer in the Renfrew and Paisley survey. BMJ 1989;298:920-924. 44. Lemer DJ, Kannel WB. Patterns of coronar,/heart disease morbidity and mortality in the sexes: a 26-year follow-up of the Framingham popolauon. Am Heart J 1986;111:383-390. 45. Castalli WP, Andenon K, Wilson PW, Levy D. Lipids and risk of coronaw heart disease: the Ftamingham Study. Ann Epidemiol 1992;2.'23-28. 46. Gordon T, Castelli WP, Hjorthnd MC, Kannel WB, 12~awber TR. High density lipoptotain cholesterol as a protective factor again.~ coronary heart disease. Am J Med 1977;62:707-714. 47. Jacobs DR, Melbane II. Bangdiwala SI, Criqui MH, Tyroler HA. High density lipoprotein eholesternl as a predictor of cardiovascular disease mor- tality in men and women: the follow-up study of the Lipid Research Clini~ Prevalence Study. Am J Epkhmiol 1990;131".32.-47. 48. Gordon DJ. Role of ciroularing high.dansity lipoprotein and ttiglycerides in coronary artery disease: risk and prevention. Endocrinol Metah Clio North Am 1990;19:299-309. 49. Abbot~ RD, WiLson PWF, Kannel WB, Castelli WP. High density l~popro- rein cholesterol, total cholesterol screening and myocardial infarction: the Ftamingham study. Arteriosclerosis 1988;8:207-211. 50. Bronner D, Weisbort J, Meshulam N, Schwarr~ S, Gross J, Salt~-Rennert H, Airman S, Loehl K. Relation of serum total cholesterol and high-density hpoprotain cholesterol percentage to the incidence of definite coronary events: twenty-year follow-up of the Donolo-Tel Aviv Pro~e~tive Coronary Artery D/sense Stud~'. Am J Cardiol 1987;59:12.71-1.276. 51. Gordor, D|, Prohstfleld JL, Garrison RJ. High denstty lipoprotain cholesterol and cardiovascular disease: four prospective American studies. Circulation I989;79'.8-!.5. 52. Kannel WB. Metabolic risk Factors for coronary heart disease in womera perspectives from the Framingham Study. Am Heart J 1987;114:413-419. 53. Kannel WB. High density hpoproteins: epidemtotogic profile and ~ks of coronary artery d~ase. Am J Cardio[ 1983;52:gb-72b. 54. Hazard WR. Atherogertesis: why women live longer [hart men. Geriatrics 1.985;40:42.-52.. 55. Kvist H, Cb.owdury B, Grangard U. Predictive equasiora of total ~ad visceral adipose tmsue volumes derived ~m measurements w~th computed tomng- raphy in adult men and women. Am J Clio Nutr I988;48:1351-1361. 56. Lapidus L, Bengts~n C.. Regional obesity as a health hazard in women: a prospective study. Acta Med Stand Suppl 1988;723:53-59. 57. Larsson B. Regional obesity as a health hamrd in men: prospective studies. Acra Med Scand Suppl 1988;723:45-51. 58. Thompson CJ, Ryu JE, Craven TE, Kahl FR, Crouse JR. Central adipose distribution is related to coronary atherosclerosis. Artarioscler Thromh 1991; 11-.327-333. 59. Fujioka S, Matsuzawa Y, Tokunaga K, Turui S. Contribution of intra-

Epidemiology September 1997, Volume 8 Number 5 CORONARY SEX DIFFERENTIAl. 591 abdominal far ,accumulation to the impairment of glucose and lipid metab- olism in human obesity. Metabolism 1987;36:54-59. 60. Peiris AN, Hermes MI, Evans D], Wilson CR, Lee MB, Kissebah AH. Reht~onship of anthtopometric measurements of body fat dJztriburion to metabolic profile in premenopausal women. Acta Med Scand Suppl 1988; 723:179-188. 61. Despr~s J.P, Moorjani S, Lupien PJ, Tremblay A, Nadeau A, Bouchard C. Regional distribution of body fat, plasma lipoptoseins, and cardiovascular disease. Arteriosclerosis 1990;,10:.;D7-511. 62. Freedman DS, Jacobsen SJ, Berboriak JJ, Sobocinski KA, Anderson AJ, Kissebah AH, Sasse EA, Gmchow HW. Body fat d~stribution and male/ female differences in lipids and lipoptoreins. Circulation 1990;81:1498- 1506. 63. Karmel WB, McGhee DL. Diabetes and cardiovasoular disease: the Ftam- ingham Study. JAMA 1979;241:2035-2038. 64. Berrert-C.onnor ~ Colin BA, Wingard DL Why it diabetes melhtus a stronger risk factor for fatal ischemi¢ heart disease in women than m men? JAMA 1991;265:627-63 L 65. I'N6rilili K, Savolainen E, Kaukola S, Haapakoski J. Plasma insulin as a coronary heart disease risk factor: relationship to other risk factors and predictive value during 9.5 ,/ear follow up of the He[sinki Policeman Study Population. Acra Med Stand Suppl 1985;701:38-52. 66. Fontbonne A, Charles MA, Thlbult N, Richard JL, CPaXide JR, Warner JM, Rcmdin GF~ F.schwege E. Hyperintulinemia as a predictor of coronary heart disease mortality in a healthy population: the Paria Pr~pective Study, 15-yaar follow-up. Diabetologia 1991',34:356-361. 67. Mudan M, Or J, Karasik A, Droty Y, Fuch* Z, Lnsky A, Chetrit A, Halkin H. Hypetinsulinemia, sex, and risk of' arherosclerotic cardiovascular disease. Circulation 1991"84:1165-1175. 68. g6nnemaa T, Laakso M, P`/6riil~l K, Kallio V, Pukka P. High fatting plasma insulin is an indicator of' coronary' heart disease in non-insulin-dependent diabetic patients and nondiabetic sublects. Arterioscler Thromb 1991;11: 80--90. 69. Mykkanen Io Laakso M, P,/6r~lit IC High plasma insulin levels associated with ¢orortaty heart disease in the elderly. Am J Epidemiol 1993;137:1190- 1202. 70. Despr'e* J-P, Lamarche B, Mauti~ge P, Canrin B, Dagenais GR, Mooriani. S, Lupien P-J. Hyperinsulinemia as an independent risk factor for ischemic heart disease. N Engl J Med 1996~334:952~957. 71. Welin L, Etiksson H, Lart~on B, Oh[son L-O, Sv~irdsudd K, Tibblin G. H'fperinsulinemia is not a ma~or coronary risk factor in eldarl,/men: the stud,/of men born in 1913. Diaberologia 1992;35:766-770. 72. Cullen K, Srenhottse NS, Wearne KL, Welbom TA. Multiple regression anal~ia of risk factors for cardiovascular disease and cancer mortalit~ in Busselton, Western Australia: 13-year study. J Chmn Dis 1983;36:371-377. 73. Collins VR, Dowse GK, Zimmet PZ, Tuomilehto J, Albert~ KG, Gareeboo H, Nan L. Serum insulin aml ECG abnormalities suggesting coronary artery disease in the population of Mauritius and Nauru: cross-sectional and lon- gitudinal associations. J Clin Epidemiol 1993;46:1373-1393. 74. Stout RW. insulin and atheroma: 20-,/ea~ perspective. Diabetes Care I990;, 13:631-654. 75. Foumier AM, Gadia MT, Kuhntsly D8, Sk,/ler JS, So*enko JM. Blood pressure, insulin and gl,/cemia in non-diabetic subj~ts. Am J MUd 1986:80: 861-864, 76. Manolio TA, Savage PJ, Burke GL, Lui KA, Waganknecht LF., Sidney S, Jacuhs DR, Roseman JM, Donahue RP, Oberman A. Association of fasting insulin with bkx~d pressure and liplds in young adult~: the CARDIA Study. Arteriosclerosis 1990;10:430-4.36. 77. Laakso M, Berrett-Cormor F.. A.symptomatic hyperglycemia is associared with Lipid and li~ptotein changes favoring atherosclero~is. Artetio~cletosis 1989;8:665-672. 78. Stem MP, Haffner SM. Body fat distribution and hyperinsulmemia as risk facturs for diabetes and cardiovascular disease. Arteriosclerosis 1986;6:123- 79. Reaven GM. Rote of'insulin resistance in human disease. Diabetes 1988;~7: 1595-1607. 80. Fontbonne A. Insulin: a sex hormone for cardiovascular ~k? Circulation 199l;~:1442-1444. 8L Orchard TJ, Becket DJ, Kullet LH, Waganer DR, LaPorte RE, Drash AL. Age and sex variations in glucose tolerance and insulin responses: parallels w~th cardiuw~seulat risk. J Chron Dis 1982;35:IZ3-[3L 82. Yki-Jasvinen H. Sex and insulin sensitivity. Metabolism 1984;33:1011- 1015. 83. Donahue RP, Orchard TJ, Becket DJ, Kuller LH, Drash AL. Sex differences m the coronary heart disease risk profile: a [x~sible role for insulin: the fieaver County Stud`/. Am .~ Epidemioi 1987;125:650-657. 84.. Bressler P, Belle`/S, Sand ME:. Insulin rasistance and coronas3' artery disease:. the missing [ink (Abstract). Diabetes 1.992;41(suppl):24A. 85. Runnel WB, Wolf PA, C.,a.~relli WP, D'Agmtino RB. Fibtinogen and risk of cardiovascular disease: the Framingham Study. JAMA 1.98"/:Z58:1183-1186. 86. Hamsren A, Blomback M, Wiman B, Svensson J, Sz~mnsi A, de Faire U, Metringer L. Haemo~tatio func~on in myocardial infarction. Br Heart J ~.986;55:58-66. 87. Lowe GDO, Lee AJ, Rumley A, Price .IF, Fowkes FGIL Blood visce~ity and ~k of cardiovascular events: the Edinhngh Artery Study. Br J Haemarol 1997;96:168-173. 88. Meade TW. F.pidemiology of atheroma, thrombosis and ischaemic heart disease. In: Bloom AL, Thomat DF, ed~. Haemostasis and Thtomhnsi~. F.dinburgh: Churchill Livingstone, 1987;651-678. 89. Smith FB. Von Willebrand factor, beta-thromboglobulin and plarelet acti- vation, in: Fowkes FGR, ed. Epidemiolog,/of Peripheral Vascular Disea.,e. Berlin: Springer-Varlag, 1991;271-283, 90. Wu KK, FoLsom AR, Heis~ G, Davis CE, Conlan MG, Barnes R. Association of coagularion factors and inb.ibirors with carotid artery athetosclerosis: early results of the Atherosclerosis Risk in Communities (ARIC) Stud,/. Arm Epldemiol 1992;2:4"/I-480. 9|. Meade TW, Vickers MW, Thompson SG, Stifling Y, Haines AP, Miller GJ. Epidemiolog~cal characteristics of p~areler aggregabitity. BMJ 1985;290:428- 4.32. 92. Folsom A~., Wu KK, Davis CE, Conlan MG, Sorlie PD, SzIdo M. Population correlates of plasma fibrinogen and factor VII, putative cardiovascul#r risk factors. Atherns¢letosis 1991;91:191-205. 93. Conlan MG, Folsom AR, Finch A, Davia C'E, Sorlie P, Marcucci (3, Wu KIC Assu¢iarions of factor VIii and yon Willebtand factor with age, race, sex, and risk factors for atherceclerosis. The Atherosclerosis Risk in Communi- tie* (ARIC) Study. Thtomb Haemost 1993;70:380-385. 94. Sunddl IB, Ni[sson TK, Ranby M, Hallmans G, Heilsren G. Fibrinolyri¢ variables are related to ,age, sex, blood pressure, and bud,/build measure- ments: a ctos*-secrional arod,/in Norsjo, Sweden. J Clin Epidemiol 1989; 42:"/19-723.