4.2. Professional Health Care
As with the general public, medical practitioners are prone to the misconception of CVD being a predominantly male problem [9
]. They also may have an unconscious sex bias. A survey of American cardiologists, using simulated patients, found that they were less likely to rate the usefulness of angiography as “high” for women than for men. This implicit bias was related to their perception that women were less risk-tolerant than men [10
Such inherent bias may manifest in the underuse of screening for CVD risk, as was found in a large study of Australian general practice in which the adjusted odds of a woman being screened for CVD risk factors were 12% lower than for a man [11
]. In secondary prevention, Zhao et al. (2017) presented results from patients with a history of CHD recruited from routine outpatient cardiology clinics in 11 countries across Europe, Asia and the Middle East [12
]. After adjusting for age, they compared the achievement of CVD risk factor and lifestyle targets between the sexes and found that women did worse than men in achieving guideline-based targets for high-density lipoprotein cholesterol, low-density cholesterol, total cholesterol, glucose, physical activity, obesity and cardiac rehabilitation. Men did worse only for blood pressure control and for achieving the target of not smoking (however, to put the latter result in context, female smoking is unusual in Asian and middle-eastern countries). Overall, only 8% of men reached all treatment targets, but the percentage was even lower for women, at 6%; lifestyle targets were more often met but, still, were achieved by relatively few people: 34% of men and 32% of women. Although individual patients have a great deal of responsibility for their own health behaviour, it is reasonable to conclude that there is more that their carers can do to improve risk factor awareness and action, both overall and in relation to sex disparities.
There is also evidence that the treatment given after a CVD event to a woman may be less adequate than that given to a man, which will contribute to risk factor differences in secondary prevention. For example, US guidelines say that patients who survive an MI should be given high-intensity statins (cholesterol-lowering drugs). Yet, Peters et al. (2018), using data from two large health insurance systems, found that, amongst all patients given statins, American women have been less likely than American men to receive high-intensity statins since 2007 [13
]. In 2014/5, 9% (95% confidence interval 8 to 10%) fewer women than men filled a prescription for a high-intensity statin within 30 days, in this study population.
In the acute setting, physicians may miss heart attacks in women because they often experience attacks in a different way compared to men, classic medical text-books having been written according to the male model of the disease. Specifically, women seem to be more prone than men to experience shortness of breath, nausea, vomiting, back, shoulder or jaw pain, and anxiety as symptoms of MI. They are also less likely to experience physical exercise as the trigger to their MI, instead being more likely to experience emotional distress. A UK study found that the odds of a woman having an incorrect initial diagnosis on admission to hospital were 37% higher for women than men, amongst all patients who had a ST-elevation myocardial infarction (STEMI), and 29% higher for patients who had a non-ST-elevated MI (NSTEMI) [14
]. Since those with an incorrect initial diagnosis were more likely to die within 12 months than were those with a correct initial diagnosis (time to death was 10% shorter for misdiagnosed STEMI cases and 14% shorter for misdiagnosed NSTEMI, after adjustment), this suggests that misdiagnosis is an extremely important problem, more so in women. Figure 4
shows data from another study, involving 582,000 admissions to emergency departments in Florida [15
]. Regardless of the sex of the cardiac physician attending, after allowing for several confounding factors, survival was worse for women than for men. A striking feature of these data is that, although female physicians appear to have similar results for female and male patients, amongst patients treated by men, female patients survive their treatment less often than do male patients. Although the differences in probabilities in Figure 4
are small, this issue, if widespread, is of immense importance in terms of female lives that might be saved and is also compounded by the relative lack of female cardiologists in many settings. To address such issues, women’s cardiology clinics have recently been founded [9
] but are yet to have a wide distribution.
For those who survive a MI and leave hospital, women tend to have worse survival and be more likely to have a recurrent event. Figure 5
shows an example, although, in this case, sex differences disappeared after adjusting for age [16
]. This is explained by women being generally older when they suffer MI, which shows the importance of controlling for age in such sex comparisons. All the same, given the female advantage in age-specific CHD rates before a CHD event, similar age-adjusted survival after an MI suggests that women have lost some of their natural advantage, which might be explained by women receiving worse secondary prevention care, as discussed above.
The Food and Drug Administration of the USA has a history of more than 20 years of promoting reports of clinical trial results by sex, because experience has shown that some drugs differ in efficacy or safety between the sexes [17
], as acknowledged by the American Heart Association in 2016 [20
]. However, such policies are not universal, and there is no sign that cardiovascular trials are yet recruiting as many women as men, even in the USA [19
4.3. Risk Factors for both Sexes
Sex differences in the prevalence of classical modifiable CVD risk factors—high levels of blood pressure and cholesterol, smoking, diabetes and overweight/obesity—can be expected to have a substantial effect on sex differences in CVD (as well as other clinical) outcomes. These prevalences differ greatly across the world, and it is impossible to draw generalisations. Instead, one contemporary example will be mentioned. This is a recent analysis of national surveys amongst adults in the USA between 2001 and 2016 [23
] which found that trends in reductions in systolic blood pressure and smoking and increasing prevalence of diabetes mellitus were similar between women and men. However, reductions in total cholesterol were greater in men, and increases in body mass index were greater in women. As in a global study of obesity [24
], women were more likely than men to be obese.
In a systematic series of large-scale meta-analyses of various sets of cohort studies from general populations, several commonly recognised cardiovascular risk factors (smoking [25
], diabetes [26
], atrial fibrillation [27
] and low socioeconomic status [28
]) were shown to have a stronger relative effect on CHD in women than in men, whilst only total cholesterol had a stronger effect on CHD in men than women [29
]. There was no evidence of a sex difference associated with higher blood pressure [30
] or body mass index [31
] (Figure 6
). Therefore, although all these factors increase the risk of CHD in both sexes, some are associated with additional excess risk in one sex compared to the other.
Parallel meta-analyses, in an overlapping but distinct set of cohort studies, found somewhat similar results for stroke (Figure 7
). Again, there was a clear female disadvantage for atrial fibrillation [27
] and diabetes [32
], but the excess female relative risk for smoking was small (6%) and marginally non-significant [33
]. There was insufficient evidence to reliably estimate the effects of social deprivation on stroke [28
], as shown by the corresponding wide confidence interval in Figure 7
. Both total cholesterol [29
] and blood pressure [30
] showed no evidence of a sex difference. These analyses of stroke are, however, limited in that subtypes of stroke were not always distinguishable, and thus Figure 7
shows the results for all types of stroke together. Also, this systematic series of meta-analyses has yet to investigate sex differences in the effects of body mass index on stroke.
The greatest weight of evidence of a female disadvantage is for diabetes [34
]. According to the above meta-analyses, having this risk factor more than doubles the risk of CHD but confers an additional 44% risk to women compared to men (see Figure 6
). Results for stroke are similar but somewhat less extreme: the excess female risk is 27% (see Figure 7
). This could be caused by relatively worse care to prevent the sequelae of diabetes received by women compared to men [35
] and by women tending to have greater adiposity when diagnosed as positive for diabetes [36
A general limitation with these meta-analyses is that most of the results that were pooled came from published studies which employed various methods and covariate adjustments. Although all adjusted for age, age-specific sex differences were not routinely reported and could not be reliably summarised. Another limitation is that it was only possible to pool relative risks. Avoiding these issues, whilst still using a large set of data, Millett et al. (2018) analysed sex differences in risk factor associations with MI in the UK Biobank (n
= 471,988) [3
]. The female disadvantage for diabetes and smoking, on the relative scale, seen in Figure 6
, was confirmed, but there was also evidence of such a disadvantage from high blood pressure. Results for atrial fibrillation and low (versus high) social status went in the same direction (to women’s relative disadvantage) as in Figure 6
but were not statistically significant. Sex differences, where they occurred, in relative risks were retained with ageing. On the absolute scale, men had higher risk differences than women for hypertension, smoking, overweight/obesity but not diabetes.
4.4. Female Risk Factors
Factors concerned with reproduction of the species are clearly specific to women: several have been shown to be associated with future CVD. Peters et al. (2018) found each of the following to increase the risk of CVD in the UK Biobank study population [38
History of hysterectomy
Amongst those who gave birth:
Early age at first birth
History of miscarriage
History of stillbirth.
In this analysis, and another using the similar China Kadoorie Biobank [39
], there was also evidence of an increasing risk of both CHD and stroke with an increasing number of children, at least in those women who had any children. Although this phenomenon had formerly been thought to be caused by female biological factors, in both biobanks the association between CVD and number of children was found to be very similar in men, leading to a conclusion that the causal factors were most likely social in origin. This shows the great value of including male controls, whenever possible, when researching women’s health. In particular, it illustrates a limitation in drawing inferences from the Canadian survey of women’s cardiovascular knowledge discussed in Section 4.2
Adverse pregnancies, involving gestational diabetes (GD) or pre-eclampsia (PE), are also risk factors for future CVD in the mother. For example, in the second Nurses Heath Study, after adjusting for age, pre-pregnancy body mass index and other covariates, GD was associated with subsequent CVD, with a hazard ratio (95% confidence interval) for GD versus no GD of 1.43 (1.12 to 1.81) [40
]. Also, in a meta-analysis [41
], relative risks (95% confidence intervals) for PE versus no PE were: 2.33 (1.95 to 2.78) for CHD; 2.03 (1.54 to 2.67) for stroke; and 2.29 (1.73 to 3.04) for cardiovascular mortality.
A drawback with most studies of pregnancies is that the data used typically derive from maternal registries which lack data on classic CVD risk factors prior to, and after, the pregnancy. Thus, for example, PE could simply reflect a woman’s own natural high blood pressure. This not only requires a caveat to statements about the additional (or “independent”) nature of PE and GD over and above classical CVD risk factors, but also undermines the ability to reliably include PE and GD in a CVD risk score for general use in women currently free of CVD. This presents a fundamental limitation when considering the true CVD risk of women who have a history of adverse health in pregnancy and could thus mean that many fail to receive life-saving treatments.