Chris was my first month at a Catholic high school in North-eastern Wisconsin. There were many kids who smoked cigarettes between the fingers and the life-size Jesus statue. They also skipped Mass to grab fries from the nearby fast-food restaurant. Chris and her group sat somewhere near the school's social hierarchy. She viewed the absurdity and drama of adolescence with cool, heavy-lidded insight. I watched her from afar, and then I moved around her orbit, smiling whenever she invited me to join her gang for lunch.
We lost touch after high school. Chris moved to the Midwest, while I went east. She sold costume jewellery at Daytons department shop to pay for college at the University of Minnesota. She married Adam, a tall classmate. They had two daughters and she merged with the main stream to become a lawyer. She used to run at the YWCA, and would cook oatmeal for breakfast. In 2010, she was 35 years old and began to feel stomach pains. It was a strange feeling, she said. She was told it was indigestion by her doctor and sent home. The symptoms continued to return. Strangely, she was tired and constipated. She went back to the doctor. She said she didn't feel well. He said that you're tired because you are raising children. Youre stressed. You should be exhausted. She was frustrated and saw other doctors. They said that you were a working mother. Relax. You should add fibre to your diet. These problems escalated in frequency. She was always tired and anaemic. When she had coffee with a friend, she would feel tired. She was advised to get some sleep. You might try sleeping pills.
Every Saturday, get the Guardians award winning long reads delivered to your inbox
Chris was unable to walk around the block by 2012 due to fatigue. She would fall asleep around three in the morning. Her skin was becoming pale. She felt pain every time she ate. Adam suggested that she visit his childhood doctor, who was only 40 minutes away. The doctor checked her blood. He thought she was bleeding from the inside because her iron levels were so low. He requested a CT scan as well as a colonoscopy. Chris was relieved when they discovered a tumour the size of a golf ball. She was sick. They had been hearing her tell them since childhood. There was now a problem. However, the relief was only temporary. Six days later, surgery revealed that the tumor had spread to her abdomen. Chris was 37 years old when she was diagnosed with stage four colon cancer.
Research on the causes of health disparities among social groups has always focused on patients. This includes their behavior, their situation, and their circumstances. One theory is that some patients are more reluctant to seek treatment or don't follow orders from doctors.
Perhaps patients prefer to receive fewer interventions. Black Americans have always believed that health disparities are caused by patients. This notion was influenced by racism in the medical profession of 19th century. Numerous articles were published in medical journals detailing the physiological problems of Black Americans. Statistics showing an increase in mortality rates in late 19th century America were not considered evidence of economic and social oppression, but rather of physical inferiority.
Research has shifted to the social and environmental determinants and health in the 21st century. This includes the impact on health outcomes of differences in access and insurance coverage. These factors are clearly illustrated by the devastating and disparate effects of Covid-19 on communities and people of color. The disproportionate burden is due to a network of social inequalities, which includes more dangerous working conditions, a lack of access essential resources and chronic health conditions that result from continued exposure to racism, exclusion, and inequality. The burden of disease for trans people, especially trans women of color, is immense. Trans people, who are often marginalized by society and face high rates of poverty and workplace discrimination, unemployment and severe psychological distress, have higher rates of chronic conditions like asthma, chronic pulmonary obstruction disorder, depression, HIV, and other chronic diseases than the cisgender community. Nearly 28,000 trans people in the US were surveyed in 2015. One-third of them had not sought healthcare because they couldn't afford it.
Researchers have recently begun to look at the differences in the treatment of patients by doctors and other healthcare professionals. Study after study has shown that some groups are treated differently than others.
For instance, black patients are less likely to be prescribed pain medication for the same symptoms than white patients. This disparate treatment pattern is even more prevalent for children. Researchers believe this is due to false stereotypes about Black people not feeling pain in the same way as white people. These stereotypes date back to chattel slavery, and were used to justify cruel treatment. This problem is reflected in medical education where race is often portrayed as a risk factor for many diseases rather than the accumulation and stressors associated with racism. For example, black immigrants from the Caribbean have lower rates for hypertension and cardiovascular disease than US Blacks, but their rates of illness rise over time to match those of Americans born Black. This is generally due to the racism they face in the US.
Even when insurance, illness severity, and hospital type are controlled, black patients receive fewer therapeutic procedures. Black patients are less likely than whites to receive guideline-based treatment for heart attacks. They are also less likely to be seen a cardiologist in ICUs for heart failure. This is linked to survival.
These biases can impact the quality of many interactions in clinics. Obese patients are less likely to be seen by doctors, and they tend to spend less time with them. Transgender people are subject to discrimination and prejudice. In 2015, the survey found that a third of respondents had experienced a negative encounter at a healthcare provider in the previous year. This included being denied treatment. Nearly a quarter of respondents were so worried about being mistreated that they avoided seeking necessary healthcare. Transgender people have two options: either disclose their transgender status and face discrimination or hide it to avoid inappropriate treatment.
Unexamined bias, even though doctors aren't generally trying to give better care to certain people than others, can cause devastating harm.
Chris was informed that her increasingly difficult to manage symptoms were not serious. As a group, women receive fewer timely interventions, less pain treatment, and are less often referred to specialists. A 2008 study that included nearly 80,000 patients from more than 400 hospitals revealed that heart attack victims often have to wait longer for treatment and are more likely to die once they reach the hospital. Women are less likely than men to be referred for cardiac rehabilitation or prescribed the correct medication after a heart attack. Women over 50 years old are less likely than men to receive lifesaving intervention. Women with knee pain are 22 percent less likely than men to be referred to a knee replacement. Canadian researchers found that women were more likely to die from critical illnesses after spending less time in the ICU, and after accounting for severity, were significantly more likely not to be given life support.
Poor treatment is more common for women of color. An analysis of childbirth experiences in 2019 revealed that many women experienced disrespectful, ineffective communication and condescending treatment from their providers. Some were even bullied into having C section. The story of Serena Williams' childbirth is well-known. Serena Williams has a history with blood clots. However, when she recognized the symptoms and requested immediate scans and treatment, both the nurse and doctor dismissed her request. Williams got the help she needed. However, Williams's inability to recognize and treat women's distress and symptoms contributed to higher maternal deaths among Native American, Alaska Native, and Black women. In the US, Black women are three to four times as likely to succumb to complications during childbirth than white women.
A structural reason is also behind inferior care. Women have been historically excluded from most medical research. There are many reasons for inferior care. They could be anything from the desire to protect women who are pregnant from drug abuse, or to make research more efficient, or even to judge that men's lives are better than women's. Many of the most important studies on heart disease and ageing have not included women. The all-men study of cardiovascular diseases, MRFIT, was born out of the belief that heart attacks in male breadwinners was a national emergency. However, cardiovascular disease is the leading cause for death for women. One particularly shameful example is the 1980s study that examined the effects of obesity on breast and uterine cancer. Women were excluded because men's hormones were easier and more affordable to study.
These practices were fundamentally based on the assumption that men are the default human. Women were therefore a subcategory that could be safely left out of studies. There is a problem. The assertion that women are complex and unique and therefore cannot be included in research and that all findings should apply to them is logically flawed. The US Congress demanded that all medical research funded by the National Institutes of Health include women. In fact, earlier drug studies excluded women. This may explain why women are 50-75% more likely than men to suffer adverse side effects from drugs.
Steven Epstein, a sociologist, points out that medicine is often founded on categories that are politically and socially relevant. However these categories are not always medically applicable. Relying on race could lead to the exclusion of social causes that cause health disparities, and it may also perpetuate harmful and false ideas in medical practice. However, it is dangerous to ignore differences like sex. Women's symptoms are not medically understood because of this exclusion. Doctors were informed, for instance, that heart attacks can be seen in women with unusual symptoms. These atypical symptoms are quite common for women. These symptoms were not unusual because they had not been studied. There are also differences in how susceptible women and men are to various diseases and the symptoms they present. Some drugs may not be well-received by both genders. Some medications take longer to get out of the body because women's kidneys filter waste slower.
Doctors have been unable to recognize the differences between women's bodies, and instead focus on what is not there. In her book Doing Harm Maya Dusenbery points out that this ignorance can also be harmful to historical stereotypes.
Doctors label women's symptoms medically unexplained if they don't match the textbooks. These symptoms could be classified as psychological or physical. Many women's symptoms can be medically unexplained. This reinforces the notion that women's symptoms are unrelated to medical conditions and casts doubt on womens stories. A study showed that men with irritable bowel syndrome tend to have more scans than women, but women are often offered lifestyle advice and tranquilisers. Chris advised my friend to get enough sleep because of her fatigue and pain.
Chris's doctor finally ordered the correct tests and informed her that he had seen many young women in his clinic whose symptoms were due to stress. Studies show that many diseases are delayed in diagnosing women, including Crohns, Ehlers Danlos syndrome, coeliac disease, and tuberculosis. In 2015, more than 16,000 patients in the UK were surveyed and found that delayed diagnosis for various types of cancers, including bladder, gastric and head, neck and lung cancers and lymphoma. Dusenbery claims that doctors are rarely given feedback on misdiagnoses. They don't know where they went wrong.
It is estimated that diagnostic errors cause around 80,000 deaths each year in the US. In 75% of these cases, cognitive factors play a part. What could Chriss' case have looked like? It is essential that doctors become more aware of their biases and the motivation to overcome them. Biases are more likely when people are mentally ill. It can be helpful to have meaningful, collaborative contact with people from other social groups. There is another way to reduce bias. This can be used in conjunction with all of these efforts and provide additional protection from interpersonal bias.
Elliott Haut is a trauma doctor at Johns Hopkins hospital in Baltimore. He is affable and sweet, but he seems most happy when he talks about safety. His office desk is cluttered with books on preventable deaths. An over-the-computer note reads:
The trauma unit may also see victims of farm accidents or motorcycle crashes in other parts of the country. Many trauma patients at Hopkins are victims of stabbings and gunshots. One patient arrived at Hopkins with a broken beer bottle lodged in his neck. The entire Budweiser word was clearly legible on the jagged glass.
Haut was appointed to lead efforts to improve Hopkins' trauma department 15 years ago. Haut was asked to oversee efforts to improve the performance of doctors and create better outcomes for patients. Haut began to investigate the hospital data and discovered that patients were developing bloodclots at an alarmingly high rate.
An electron microscope shows a blood clot. Photograph by Science Photo Library/Alamy
Blood clots, the condition that endangered Serena Williams' life while she was giving birth, are gelatinous globs made of blood cells that can travel through blood vessels to block blood flow to the lungs. In the US, they kill approximately 100,000 people each year. This is more than all of breast cancer, Aids, and car crashes combined. If doctors recommend the best clot prevention, many of these clots can be prevented. This could be blood thinners in some cases. In others, it might be mechanical squeezy boot that inflates and deflates around the legs to move the blood. Haut discovered that only 33% of high-risk patients at Hopkins were receiving the correct blood-clot prevention. As we sat in Haut's office in East Baltimore, next to a pile of wooden puzzles, he explained that we would get a patient in for a routine procedure and then they would die from a pulmonary embolism. This problem was not limited to Hopkins. Patients at all hospitals across the country were receiving proper clot prevention about 40% of the times, a problem the American Public Health Association called a crisis.
Haut was unsure why doctors failed to recommend the correct interventions. Haut thought that doctors overestimated blood-clot prevention risks because patients who developed complications from blood thinners remembered things better than those who were successfully treated. Haut was not concerned about the disparities. His goal was to make clot prevention more accessible for all.
Haut and his team searched for an approach by Peter Pronovost (another Hopkins doctor who had lost his father to cancer) in order to do this. Pronovost had developed a method for improving medical care using an approach that is used in aviation. This was called the "simple checklist". A checklist is exactly what it sounds like: a reminder of all the steps that a clinician must take. It helps to fill in the gaps and prevents human mistakes from adding up. For example, ICU care requires almost 200 actions per day. Missing even one of these actions can lead to complications.
Pronovost demonstrated that a checklist can reduce infections in intensive care units by making sure that doctors follow a set of predetermined tasks each time. A five-step checklist was used to remind ICU workers to wash their hands after each procedure. This resulted in a 66% decrease in catheter-related bloodstream infection rates. Over the course of the 18-month study, the drop remained steady.
Haut and his team set out to create a checklist to prevent blood-clots. A computerised checklist would appear on the screen whenever a healthcare provider admitted someone to the hospital. The checklist would guide the doctor through the risk factors for blood clots as well as bleeding from blood-thinning medications. Once the checklist was completed, the system would recommend a treatment, such as a blood thinner or a mechanical squeezyboot to move the blood. Doctors who didn't select the recommended treatment had to explain why.
It worked. The checklist was a success. Patients were able to get the best clot prevention and preventable clots in internal medicine and trauma increased. A study that covered a month of hospital admissions showed that two out of every ten internal medicine patients returned to the hospital with blood-clots within the first 90 days. The rate of fatal pulmonary embolism decreased by half after the checklist was introduced.
This could have been the end. Hauts office was located two doors from Adil Haider's office, which studies gender and racial disparities within health care. Haut was intrigued by their conversations and wondered if there were disparities in blood-clot prevention. Although the team had not sliced the data in that manner, they discovered a disturbing pattern when they looked at the numbers again. The rate of treatment failure for male trauma patients was 31%, but it was only 45% for females. This means that women were almost 50% more likely than men to not receive blood-clot prevention and are at greater risk of dying from this cause.
Other factors than gender could have played a role. For example, most patients with gunshot wounds are men. Perhaps doctors recommended more preventive measures for more severe injuries. The researcher who analysed the data stated that the disparities in treatment were consistent with a large, well-established pattern of women receiving suboptimal medical care.
Haut and his team looked at the numbers following the introduction of the checklist to see that there were no gender inequalities. Both men and women received the right clot protection at the same rate. The gap was gone.
The University of Chicago economist Richard Thaler and Cass Sunstein, a legal scholar, created the term choice architecture in 2008 to describe a powerful phenomenon: how the environment we choose has a significant impact on our choices. The design of a process can have a significant impact on our behavior, just as a physical environment can affect our behavior (e.g. coffee shops with no outlets for laptops). It can also be considered a type of architecture.
Researchers at the University of Minnesota found that students could be coaxed to eat more vegetables by simply changing their lunchtime routine. Students are often presented with carrots alongside more appealing options like pizza and fries in a typical lunch line. Researchers gave children a cup of carrots right when they arrived at the cafeteria. This was when they were most hungry. The experiment worked. Kids ate more carrots. It was important to make carrots a part of a competition where they could win. This contest wasn't against fries but against really hungry people. It wasn't necessary to convince students about the benefits of vitamin A in order to change their eating habits. The change was in the architecture.
The Hopkins checklist is a form of choice architecture. It helps doctors to shape their behaviour through design, not persuasion. The checklist doesn't force doctors to consider their biases more; it just interrupts the way they make decisions. The Hopkins checklist requires doctors to separate the thinking involved in a medical decision. It acts in a manner similar to a prism. It reverse-engineers a holistic judgement into its constituent parts the same way a prism separates rainbow colours from white light.
This checklist supports human judgement. Although it is intended to remind doctors about steps they may have forgotten, bias is not about forgetting. It's about making assumptions in order to judge and assess, even if you are not aware. Some doctors are resistant to the intrusion and point out that mandated checklists may not be perfect. One hospitalist explained to me that they might not consider all the factors that a doctor may be considering. A clot checklist may ask questions that assess risk at a particular moment, but an experienced doctor might point out that a patient in pain could be having a procedure tomorrow that could alter the risk profile. This nuance is not covered by the checklist. The checklist is not meant to be a replacement for medical decision-making.
However, checklists have been proven to reduce bias in other areas. The disparities in psychiatric hospitalisations of young, low-risk Hispanic, and Black patients was reduced after the introduction of a structured decision-making process in Illinois. The Mayo Clinic introduced a system that automatically refers patients for cardiac rehabilitation following heart attacks. This eliminated the gender gap in referral rates between women and men.
The Boston Symphony Orchestra, in 1952, began using behavioural design principles to reduce bias. Instead of musicians performing in front of a panel, a screen was used to separate them. To make it more clear, women musicians were asked not to wear high heels and had to take off their shoes. Instead, a man stood on stage and made fake footsteps with his shoes.
American orchestras adopted curtained auditions in the next decades. A heavy cloth was hung from a ceiling or a room divider was stretched across the stage like an accordion. Most people had adopted this practice by the 1990s. Claudia Goldin and Cecilia Rouse, economists, studied the differences in orchestras using this approach and found that it affected judges' assessment of womens skill. The concealment of musicians' identities increased women's chances of being selected for the next round of auditions 50%. Today, almost 40% of orchestras are composed of women.
It is not a good idea to rely on a blunt tool such as blurring out someone's social identity. This can lead to erasure which is a form discrimination. However, it can protect a person's evaluation from unfair advantage or harmful stereotyping in the case of a hiring determination.
The masked approach to medicine is not possible. Medicine relies on face-toface interactions between patient and doctor. Checklists are a close relative. They help people in power to stop making assumptions and relying on preconceived notions. This alone can bring about powerful changes.
This is an edited extract from The End of Bias: How We Change Our Minds by Jessica Nordell, published by Granta on 23 September and available at guardianbookshop.co.uk
