The lack of diversity in healthcare is a topic that’s been written about for at least the last two decades. And yet there are still groups that are systematically overlooked and excluded during the design of medical devices. So, whilst the medical field evolves with new technology that helps to improve lives, not all patients may profit equally from these developments.
Failing to account for the diversity of a medical device user population may cause the masking of adverse effects which were not considered during device development and can lead to inappropriate re-purposing of technology developed for one patient group, for another.
Furthermore, without robust consideration of the gender, physicality, mental ability, and skin colour of a patient population, medical device manufacturers risk losing the full effectiveness of their technology. As Poppy Noor writes in her BMJ article “Racism in Medicine”, a cancer AI software intended to detect melanoma that’s only trained using white skin will not be able to perform as well on darker skin, potentially leading to a higher likeliness of death from skin cancer amongst people with darker skin.
In clinical trials, participants are often still homogenous – with no consideration of minorities, or the more diverse spectrum of actual patient characteristics. Historically, this lack of diversity in trials was often intended – used as a method to achieve more consistent short-term results and to avoid trial subject variability. But with today’s level of sophistication in technology, together with the availability of increased funding and broad knowledge, the industry is left with no reason not to work towards improved inclusivity.
That said, there are still challenges to overcome on the path towards medical device development for all. In our topical article below, our Head of Clinical Corinne Larke shares an overview of the important factors to consider when striving for more inclusive medical device development.
Recognising overlooked patient populations
Determining the relevant characteristics
The exclusion of certain patient groups when developing medical devices can lead to very direct consequences for those groups. Take for example, the measurements of pulse oximeters developed considering only white-skinned participants. Without taking into account a diverse patient population, the devices have been shown to produce incorrect results for patients with darker skin tones. Indeed, the darker the skin, the more incorrect the measurement produced by the devices. Gender has also been found to influence the accuracy of pulse oximeters, meaning that female patients with darker skin tones became the most likely group to receive incorrect measurements. Whilst this bias became known in 2005, it wasn’t until 2020 when a study finally prompted the FDA, medical device manufacturers, and independent testing laboratories to perform considerable investigation into the issue.
In the increasing discussion about diversity, characteristics such as ethnicity, location, socioeconomic status, gender, and sexual orientation are frequently cited as factors that influence disparities regarding access to health care. However, factors such as patient age, handedness, body size, skin darkness, mobility, and eyesight can also influence the safety of patient care.
As such, it’s crucial that medical device manufacturers consider that the characteristics of a patient population are seldom universal. On the contrary, they may depend on what technology a device uses, and how a patient with certain characteristics can interact with that particular technology. Where gender may be a bigger factor for one type of technology, body size or education capability could be key for another.
Understanding the barriers
Recognising and understanding that diverse patient groups need to be included when developing a technology and collecting clinical data on how different characteristics impact the effectiveness and safety of a device is only one step towards being more inclusive. In reality, the next challenge to overcome is achieving access to a diverse patient population.
For patient groups that are less considered in medical research, even today various barriers exist that prevent them from entering the research space. To ensure the development of more inclusive solutions, developers and researchers must first understand those barriers that keep certain groups from wanting to, and being able to participate in studies that facilitate health care technology development.
Reluctance to participate in clinical trials can often stem from mistrust related to the research agenda due to a history of abuse or mistreatment of minorities – for example, sterilization experiments in Namibia in the late 1800s, the Tuskegee syphilis study in the US from 1932-1972, or even more recently the harmful / forced drug trials in Nigeria and Zimbabwe in the 1990s.
Other factors can also diminish diverse participation. Greater time and financial constraints, or lack of accessibility to transportation can prevent involvement from more socioeconomically challenged participant groups. Lack of access to information regarding participation opportunities can pose another barrier, for example if recruitment information is not available in a potential participant’s language or communicated via a channel through which they can be reached.
Achieving more diverse device development
Considering diversity on many levels
The Design Council UK has developed a framework for co-design that places greater emphasis on closely involving more diverse users, patients, and idea sources in the design process. The framework’s people- centred approach encourages designers to set their assumptions aside and openly discover what challenges the users face.
Furthermore, Sharma and Palaniappan propose a tiered socioecological framework to increase diversity in clinical research. In this model, diversity in research is ensured through changes on five different levels:
- Public policy
- Community
- Institution
- Interpersonal
- Intrapersonal
Public policy level
On the public policy level, representation of diverse patient populations can be adopted by regulators as an approval requirement for bringing new devices to the market. This also serves as means to unify approaches on how to work with data from diverse trial populations.
This may be a measure that works in some cases, but it may not guarantee inclusiveness. For example, changes in regulations intended to support paediatric device development have not yet led to an increase in devices for this patient group – as the regulations still require further extension to guarantee the safety of more diverse patient populations.
Some authorities such as the FDA have published guidance specifically regarding the inclusion of diverse patients in clinical studies for medical devices. For example, in the case of the pulse oximeter measurements mentioned earlier, FDA issued a safety communication on how to interpret the biased readings from these devices for patients with darker skin, and for device clearances, it now requires medical device manufacturers to test the device on patients with different skin tones. But whilst measures like these have a direct impact on the consideration of diversity, they are not yet commonplace, with authorities like EMA remaining more vague on diversity requirements.
Community level
On this level, the needs of patients from different communities must be considered in order to reach diverse participants. In the example of paediatric patients, this could mean selecting participants from the different community settings of the target patients, e.g., from varied families and educational systems, and considering the changing needs of those paediatric patients who are likely to gain more independence from parental caregivers over time.
“Community Based Participatory Research” (CBPR), which originated in the public health field, is an approach that’s intended to improve patient engagement in healthcare while also increasing representation of underserved groups in research. It’s a collaborative approach with the goal of generating partnerships between academic researchers and community based organisations as well as community members. It’s an approach where communities are involved in technology research at all stages of a research project, not just during data collection.
Institution level
At the level of institutions, Sharma and Palaniappan’s framework outlines that transparency can be improved to highlight areas in which data for different patient populations is lacking. According to Dimitri et al. in their 2021 publication Medical Device Development for Children and Young People, research infrastructures and networks with a focus on fostering deeper knowledge of specific patient populations and more in-depth collaboration across the life sciences sector can also help address diversity. To this end, the FDA proposes that medical device clinical studies be designed with input from diverse patient advisors, in order to create more relevant devices for a broader group of patients.
Interpersonal level
On an interpersonal level, more diverse research and development teams themselves can increase the consideration of more diverse patient needs. Inevitably, researchers with more sophisticated knowledge about minority communities will be able to more effectively communicate with, recruit, and retain more diverse participants.
Intrapersonal level
And finally, on an intrapersonal level, understanding why certain populations are more reluctant or less able to be participants in clinical research can help increase participant diversity.
Ensuring trials yield representative patient data
In their 2021 publication “Addressing Demographic Disparities in Clinical Trials”, Giusti et al. propose creating a framework to ensure patient data gathered in clinical trials is actually representative of the patient populations receiving the care.
The first step is to determine what is representative of the patient population for the device that is being designed. Not every medical condition affects the same slate of patients, therefore aspects such as gender, ethnic background, age, or physical abilities etc. may need to be represented differently in patient populations considered for different devices.
The second step is to use disease registries, which contain longitudinal real-world data that’s broad ranging and includes robust personal, demographic, and clinical patient information.
The third step is to access more diverse patients for clinical trials by looking beyond traditional academic medical centres, to which not all patients may have access. Health Systems, community networks which work with underserved communities, or engaging patients directly via channels like social media can also broaden patient diversity.
While a diverse patient population can present challenges for the evaluation of the data, and a more homogenous trial population for a new technology may have helped decrease the variability in results in the past, there are ways to account for the diversity in data – such as adapting the statistical analysis method and using novel study designs which account for this diversity in results.
Capitalising on technology
Maximising technological advancement is crucial to achieving more diverse medical device development.
Technologies that allow for device customisation on an individual patient level can facilitate adaptation to more diverse patient populations. For example, 3D scanning and 3D printing allow medical devices to be adapted to anatomical differences.
Developments in materials can also address different patient populations. As an example, drug delivery through dissolving patches, microneedles, or needle-free injectors could enable easier use of drug administration devices by patient populations such as paediatrics, or needle-phobic patients.
And thirdly, devices that can help patients monitor their health at home and transfer data electronically to their healthcare provider can address the inequality of location for patients that are not able to physically reach a place of care.
Digital health: new opportunities for diverse healthcare
As the use and capabilities of artificial intelligence technology grow, so does its capacity for bias. If AI is fed data from a non-diverse patient group, it will learn and exasperate the biases. But digital technologies in healthcare also have the potential to be more inclusive, offering solutions for more diverse patient populations. For example, artificial intelligence can be used in imaging technology for more accurate diagnosis across a more diverse user landscape.
In their study of inclusion in digital health technology, Unertl et al. found that there are several benefits to including a diverse community of users early and throughout the development of the technology. They cite advantages such as more relevant research, a wider impact of the technology, a better fit for the users, more effective recruitment and retention of diverse patient populations, better internal validity, and faster translation of research into action.
McKinsey Digital Insights proposes a four-step approach to ensuring diversity in digital health technology:
Step 1 – Create a diverse design team
Ensuring your design team is diverse, including diverse leadership, will not only make your organisation more successful, but will also ensure diversity of perspectives and ideas, leading to more inclusive products. Non-diverse organisations cannot successfully design devices for diverse patient populations and will be less aware of bias programmed into their devices.
Step 2 – Systematically identify sources of bias
Using a systematic process to identify potential sources of bias will enable you to plan for mitigation strategies during the development process. This can be done by building questions into every step of the design process to assess who might be excluded from a certain feature of the digital device, and what functions may increase the bias. The different dimensions of a user base, such as age, gender, physical abilities, socioeconomic situation etc. can be systematically built into the design process, to ensure they are considered.
Step 3 – Utilise a root cause analysis
When biases on a digital device are discovered, a root cause analysis can be used to determine how the biases were created, and which variables were not considered. This can help designers expand their thinking, ask more broad questions, and start considering more diverse user or patient profiles.
Step 4 – Address the causes of bias
Once the causes of the bias have been identified, they can be addressed. For this, it’s most effective to utilise a multidisciplinary approach comprised of different experts in the design teams, marketing, and leadership of the medical device organisation.
In conclusion…
By considering the relevant characteristics of patient groups and understanding the barriers to trial participation, medical device manufacturers can facilitate the active involvement of diverse patient populations during device development, which can help to identify unmet needs, advance technologies, and produce safer medical devices.
Understanding how bias during development can skew the performance of a medical device and therefore impact the patients it’s being used on, and taking action to decrease this bias, is an effort that must be jointly undertaken by clinicians, regulators, and medical device manufacturers.
Purposeful consideration of the diversity that exists amongst patient populations enables the gathering of truly representative patient data and the subsequent development of safer devices – whilst limiting side effects and complications.
And by combining the opportunities presented by technological advancement with the capabilities of diverse design teams to systematically identify and overcome sources of bias, the medical device industry will ultimately work towards more positive patient outcomes for all.
Should you have a question related to medical device development for diverse patient populations, or clinical affairs more generally, feel free to get in touch – our team is ready and happy to help.
References
Racism in Medicine: Can we trust AI not to further embed racial bias and prejudice? | Noor
The Double Diamond | Design Council UK
Patient Engagement in the Design and Conduct of Medical Device Clinical Studies | FDA
Addressing Demographic Disparities in Clinical Trials | Giusti, Hamermesh, & Krasnow
Why Diversity, Equity, and Inclusion Matter for Patient Safety | Lane-Fall
Digital health: An opportunity to advance health equity | McKinsey
Inclusive Tech? It Starts with Design | McKinsey Digital Insights
Improving diversity in medical research | Sharma & Palaniappan
Disparities in Health and Health Care: Five key questions and answers | Orgera & Artiga
