By Nadia O’Hara

In the past three years, as the COVID-19 pandemic spread across the world, a long-standing problem was brought to the forefront of public attention: medical devices can be extremely biased. Pulse oximeters, devices that non-invasively estimate blood oxygen saturation levels, have been found to be inaccurate for patients with darker skin tones. A recent cross-hospital study by Fawzy et al found that blood oxygen levels were systematically overestimated in self-identified Black, Asian, and non-Black Hispanic patients, leading to greater levels of occult hypoxemia. This may lead to patients with dark skin tones not receiving the supplemental oxygen they require. Pulse oximeters are crucial for triage and diagnosis for many respiratory illnesses. Hospitals in the U.S. used pulse oximetry readings to determine eligibility for treatment of COVID-19 with remdesivir and dexamethasone. Throughout the study, 23.7% of patients never had their treatment eligibility recognized, and 54.8% of these patients were Black. Suggesting that Black patients were less likely to not receive treatment when they needed it.

Pulse oximeters work by analyzing the relative absorption of two wavelengths of light, red and infrared, by hemoglobin in oxygenated and deoxygenated blood. The ratio is then translated to an estimated blood oxygenation. However, as highlighted by Okunlola et al, melanin in darker skin can also absorb some of the red wavelength of light, leading to falsely high oximetry readings, although it is important to note that this effect has not been thoroughly studied. There is clearly a need for the medical community to invest in further research that considers the effect of racial bias in medicine. The lack of accurate information and research is contributing to existing disparities in the healthcare system. In promising news, the FDA is convening a meeting of the Medical Devices Advisory Committee to discuss the racial bias that is built into pulse oximeters on November 1^st.

Pulse oximeters are not the only example of racial bias within medical devices. The ever popular wearable devices, such as Apple watches and Fitbits, also have greater inaccuracies when used by POC consumers. These devices are designed to track heart rate, arrhythmias, and sleep patterns. Their algorithms are beginning to make a transition to actual health-related research and are becoming FDA approved. This is concerning because their mechanism of function, photoplethysmographic (PPG) green light signaling, is less effective on darker skin tones. Wearable devices rely on a similar principle to pulse oximeters, measuring the relative absorption of green light by the patient’s blood, to identify moments of higher blood volume, and consequently heart rate. However, as mentioned, skin tone absorbs light differently, confusing the device’s algorithm, which is standardized to lighter skin tones.  

If the problem of racial bias within medical devices has been very clearly identified, why is it so widespread, even in new products? One of most prominent causes of the prevalence of racial bias is the alarming lack of ethnic diversity in medical device clinical studies. Clinical trials are the basis for evaluating the safety and efficacy of a new medicine or medical product. However, the sample groups for many trials chronically under-represent racial and ethnic minorities. This leads to a deficit of reliable information on how the devices work on dark skin. According to FDA guidelines, at least 15% or 2 individuals (whichever is larger) in a trial must be “darkly pigmented” for the FDA to approve the product. However, as mentioned by Okunlola, there is no specific information on how to meet this requirement. Many studies use the Fitzpatrick skin type (FST) scale to assess an individual’s skin tone for the trial. However, this is a misuse of FST. According to an article published in Cutis, a peer-review dermatology journal, the FST was developed to evaluate the likelihood of someone’s skin burning during phototherapy and should not be used as a proxy to take note of a patient’s ethnicity or race. The FST is completely based on the subjective assessment of the medical professional, leading to considerable bias and inconsistency. These kinds of scales are not an appropriate measure of sample group diversity in clinical trials. A new clinically relevant process for describing skin tone should be developed.

The procedure behind diagnosing and managing chronic kidney disease (CKD) is a distressing example of poor clinical trial methodology that resulted in intense racial bias. Estimated glomerular filtration rate (eGFR) is a convenient way to assess kidney function and the primary method used today. The equations to calculate eGFR are based on creatinine levels in the blood. Creatinine is a muscle metabolism waste product that does not get removed efficiently when the kidneys are damaged. Hence, poor filtration rates are reflected as high creatine levels in the blood. A study conducted in the US suggested that self-identified Black Americans have on average higher muscle mass and therefore produce more creatinine. Medical professionals subsequently included a “race adjustment” within the equations, to account for the extra creatinine. However, the results of the study are misleading. The trial relied largely on convenience sampling and did not control social factors, such as lifestyle or medical history. This small sample group could hardly be representative of the entire population of Black Americans, let alone the world. As of 2021, the National Institute of Health and Care Excellence in the UK no longer includes the “adjustment” for race in their guidelines for chronic kidney failure due to its inaccuracy. According to a study conducted between 2015 and 2018, it is estimated that if the “race adjustment” were removed from US practice: 3.3 million more Black Americans would be considered to have Stage 3 CKD, 300,000 more would become eligible for certain drug treatments and 31,000 additional Black Americans would pass the threshold for inclusion on the transplant list. In response, a group of researchers have recently developed new equations for calculating eGFR that are based on creatinine and another protein cystatin C but do not include any “adjustments” for race whatsoever. It is hoped that these equations, or others like them, can become standard in clinical practice and help remove this significant barrier that has been a part of the systemic inequality within the healthcare system for decades.

Racial bias in medical devices is only one small part of the large picture of racial bias in medicine, which ranges from a systemic lack of diversity in clinical trials and healthcare staff through to a severe lack of proper education and information for healthcare professionals. Spurred by the biases highlighted during the healthcare response to the COVID-19 pandemic, the healthcare world has been taking some of the first necessary steps to dismantle a systemically biased system. Change is slowly being achieved, starting with conducting more thorough research, but a lot more still must be done before racial bias within medicine has been properly addressed and real progress is made.

The views expressed in this article are the author’s own, and may not reflect the opinions of The St Andrews Economist.

Image Source: NHS England