Invisible Women: Exposing Data Bias in a World Designed for Men

also not work as well in women. Acetylsalicylic acid (aspirin) has been found to be effective in preventing a first heart attack in men, but a 2005 paper found that it had a ‘nonsignificant’ effect in women aged between forty-five and sixty-five.⁵ Prior to this study, the authors noted, there had been ‘few similar data in women’. A more recent study from 2011 found that not only was aspirin ineffective for women, it was potentially harmful ‘in the majority of patients’.⁶ Similarly, a 2015 study found that taking a low dose of aspirin every other day ‘is ineffective or harmful in the majority of women in primary prevention’ of cancer or heart disease.⁷

Perhaps the greatest contributor to the numbers of women dying following a heart attack, however, is that their heart attacks are simply being missed by their doctors. Research from the UK has found that women are 50% more likely to be misdiagnosed following a heart attack (rising to almost 60% for some types of heart attack⁸). This is partly because women often don’t have the ‘Hollywood heart attack’ as it’s known in medical circles (chest and left-arm pains).⁹ Women (particularly young women) may in fact present without any chest pain at all, but rather with stomach pain, breathlessness, nausea and fatigue.¹⁰ These symptoms are often referred to as ‘atypical’, a designation to which the British Medical Journal took exception in a 2016 article, saying that the term ‘may lead to the under-appreciation of risk associated with this presentation’.¹¹ And under appreciation of the risk may in turn explain why a 2005 US study found that ‘only one in five physicians across multiple specialties was aware that more women than men die from cardiovascular disease each year, and most of these physicians did not rate themselves as effective in treating sex-tailored cardiovascular disease’.¹²

Atypical or not, for certain types of heart attacks, women (and again especially young women) who present without chest pain are at particular risk of death¹³ – which makes it extremely concerning that current NHS England guidelines specify ‘acute cardiac sounding chest pain’ as part of the criteria for a patient being referred for primary percutaneous coronary interventions (PPCI) at one of the country’s specialist twenty-four-hour heart-attack centres.¹⁴ PPCI is an emergency treatment that restores blood flow during a heart attack, and which according to one doctor I spoke to has ‘massively improved survival and outcome’. But this treatment is only carried out at the twenty-four-hour heart-attack centres and, perhaps as a result, 75% of those who receive this treatment are men.¹⁵

The tests doctors use to determine what’s wrong with a patient are also likely contributing to women’s higher death rates following a heart attack. Standard tests like the electrocardiogram or the physical stress test have been found to be less conclusive in women.¹⁶ A 2016 BMJ paper refers to recent work from Edinburgh which showed that the ‘normal’ diagnostic threshold for troponin (a protein released into the blood during heart damage) may be too high for women.¹⁷ And it’s not just about ‘standard’ levels for biomarkers being incorrect in women, we also need to establish new female-specific biomarkers.¹⁸ A biomarker is a biological characteristic (like troponin) whose presence can act as a diagnostic criteria for a specific disease, and a 2014 literature review of sex difference studies suggests that this may be a fruitful area to research.¹⁹ Unfortunately, it concludes that the work done so far is too limited to be able to say whether or not female-specific biomarkers will be found.

Because women’s heart attacks may not only present differently, but may in fact be mechanically different, the technology we’ve developed to search for problems may not be suitable for female hearts.²⁰ For example, a heart attack is traditionally diagnosed with an angiogram, which will show where there are obstructed arteries.²¹ But women often don’t have obstructed arteries, meaning that the scan won’t show up any abnormalities,²² and women who turn up at hospital with angina (chest pain) may simply be discharged with a diagnosis of ‘non-specific chest pain’ and told they have no significant disease.²³ Except they do: women with ‘normal’ angiograms have gone on to suffer a heart attack or stroke shortly after being discharged from hospital.²⁴

Assuming a woman gets lucky and has her heart disease diagnosed, she must then navigate the obstacle course of male-biased treatment: sex differences have not generally been integrated either into ‘received medical wisdom’ or even clinical guidelines.²⁵ For example, say a man and a woman are both diagnosed with a swollen aorta (the aorta is the main blood vessel that runs from the heart down through the chest and stomach). They are both suffering from an equal level of swelling – but their risk is not the same: the woman has a higher risk of rupture, which carries with it a 65% chance of death.²⁶ And yet, in Dutch clinical guidelines, the thresholds for surgery don’t differ for each sex.²⁷

Diagnostic tests developed around male bodies are also a problem in other medical disciplines, even those where women are more at risk. Women have a higher risk than men of developing right-sided colon cancer, which often develops more aggressively,²⁸ but the faecal blood test commonly used to detect colon cancer is less sensitive in women than in men.²⁹ Meanwhile, because women have on average a longer and narrower colon than men, colonoscopies in women may be incomplete.³⁰ Then there’s what the WHO calls the ‘frequent mistake’ of underestimating the importance of symptoms that can only occur in one sex, such as vaginal bleeding in dengue fever.³¹ When symptoms are listed in order of frequency for all patients rather than separated by sex, female-specific symptoms can be presented as less significant than they are in reality.

The impact of such data gaps can snowball. When it comes to tuberculosis (TB), for example, a failure to account for how female social roles could make the disease more dangerous for women combines with a failure to collect sex-disaggregated data, leading to potentially deadly