Elevating diverse voices in research

A guide to inclusive practices in digital health

This article is meant for those who want to influence more equitable outcomes, specifically in the field of digital health, through the implementation of inclusive research practices.

Photo Source: www.pexels.com

Acknowledgements

This work is part of the IBM Watson Health Design Justice Initiative, a grassroots effort that seeks to address racial health disparities through Watson Health’s digital health products, led alongside colleagues Andrea Barbarin, Cherese Cooper, Debi Ndindjock, and Morgan Foreman.

Contributors to the current scope of work include: Frances DiMare Dailey, Adria Spivack, Morgan Foreman, and Laura Chen.

Note: Diversity in the context of this initiative was initially scoped to target underrepresented racial/ethnic minorities with the goal to achieve racial equity by way of overcoming racial disparities in healthcare.

Introduction

Why equity in health? African Americans and other people of color experience significant health disparities as a result of racism embedded in the social, economic, and political forces that impact individual and community health. The health system in particular has been complicit in promoting and supporting health disparities through a range of racist and discriminatory practices including unethical medical experimentation and uneven provision of healthcare services based on race and ethnicity (e.g., Washington, 2006; Bean, 2014).

Design research is an essential practice in uncovering problems to solve and learning from the communities who experience these challenges. Therefore, the context of this guide is to support design researchers and those who practice design research to uncover “wicked problems” related to racism in healthcare, learning from the communities who experience them in order to co-create the future of digital health experiences that are both fair and equitable.

Using this guide

“Designing for inclusion starts with recognizing exclusion.” Kat Holmes, Author of Mismatch: How Inclusion Shapes Design

This is the start to a discussion around implementing more inclusive design research practices in digital health. Iteration is expected as we learn more and listen to the communities we aim to empower and serve in our work. The recommended actions in each stage of research are inspired by principles of equity-centered community design (ECCD) and community-based participatory research (CBPR), as well as racial equity considerations among traditional internal review boards.

Inclusive practices summary

The following practices fall along the main phases of the design research cycle: research planning, participant recruitment, data collection, data analysis.

• Phase 1: Historical empathy and understanding
• Phase 2: Inclusive planning
• Phase 3: Racial diversity benchmarking
• Phase 4: Trust-building via consent, compensation, and data collection
• Phase 5: Participatory analysis

Overall, we believe the practice of Design Research will benefit from implementing the practices above, resulting in socially-aware products that do not widen disparities and are better-suited for the diverse user.

Phase 1: Historical empathy and understanding

Learn about the historical and current context of systemic racial inequity.

Illustration of Dr. J. Marion Sims (“Father of Gynecology”) with Anarcha by Robert Thom. Anarcha was subjected to 30 experimental surgeries.

One of the first steps in the equity-centered community design (ECCD) framework is “History and Healing.” In other words: “The history of the project’s topic, target community, and idea must be remembered, considered, and assessed. To understand the motives behind — and the potential impact — of the project, the personal history and trauma of each actor must be considered while integrating healing practices within the process.”

Similarly, an important first step as design researchers in digital health is to educate ourselves on the racial disparities that exist in healthcare. Learning about these disparities along with their contributing socioeconomic factors, such as poverty, lack of education, and unequal access to health care — all social determinants of health — are important to understanding the context of the challenges in the chosen field of study.

Learning about the historical and current context also helps build humility and empathy, in which researchers “examine how our own identities, values, biases, assumptions, and relationships to power and privilege impact how we engage with ourselves, each other, and the communities we work with. [Furthermore,] it’s not enough to build empathy — we also have to acknowledge what we know we don’t know (and what we don’t know, we don’t know)” (source).

How to implement

• Conduct a literature review or scan to gain insight on racial disparities in healthcare, leveraging examples from peer-reviewed academic journals, reputable news outlets, and authored books or publications.
• Browse public user forums or social media to learn more about the shared personal experiences of Black, Indigenous, and People of Color (BIPOC) healthcare professionals and patients in your respective healthcare domain.

Phase 2: Inclusive planning

Invite Black, Indigenous, and People of Color (BIPOC) participants to shape the goals of the study.

Photo Source: www.pexels.com

It’s important to be intentional when scoping a research study. One of the best ways to prioritize research goals related to reducing racial inequities in healthcare is to invite people who map to the demographic(s) we’re looking to research into the process, working side-by-side with them as co-creators at the onset/inception of a project. This approach provides members of the researched community an opportunity to plan and shape the context of the research upfront, maximizing benefits and minimizing any potential harm that could result from not obtaining their input. This is also important in order to develop and deepen relationships between design and research participants, placing both on an even playing field as experts in their relative domains.

Furthermore, team members should acquire training on how to best communicate with the study population (or consult with trained researchers in the organization) in addition to inviting data collectors who reflect the study population. Both are important in order to foster safety in the research environment.

How to implement

• Build on secondary and/or prior primary research in order to craft the definitions of the community groups we want to address in the research work.
• Schedule an initial planning session to present basic information and the goal of the project, encouraging discussion about the ethical implications and considerations of the research work.
• Invite 1–2 community leaders who represent the target demographic(s) to an initial planning session. Empower these individuals to question and improve the project plan, recruitment methods, and discussion guides accordingly, keeping in mind that all research is an experiment with the goal of improving with every time.
• Obtain skills training to learn how to conduct research with diverse populations and/or invite at least one internal team member who represents the target demographic as a data collector or consultant for the project. Before invitations are sent out, determine if there is availability and willingness to participate. This should always be done when the topic of conversation will purposefully and undoubtedly be connected to someone’s personal health experiences surrounding race/ethnicity.

Phase 3: Racial diversity benchmarking

Use screening criteria in an effort to over-represent individuals from Black, Indigenous, and People of Color (BIPOC) communities.

Photo Source: www.pexels.com

Overall, it’s important to ensure equitable selection of research participants. According to Racial Equity Considerations and the Institutional Review Board, “when identifying potential study participants, researchers should carefully select participants who are appropriate to answer each research question. Individuals or groups should not be selected solely because they are readily available or convenient to reach. This practice ensures justice because it distributes the benefits and burdens for participating in research.”

Furthermore, the screening and recruitment process can be critical to establishing how our participants choose to define themselves, and thus gives design researchers an opportunity to choose well according to the project goals and customize interview environments later in the process.

How to implement

• Identify what demographics are needed to adequately answer each research question. If it is clear that the specific topics or industries have known underrepresented (or even harmed) groups, this should be top of mind when calculating benchmarks. For example, African Americans make up about 13% of the U.S. population, but represent only 5% of clinical trial participants (source). When establishing screening criteria for a research study on patient challenges during clinical trials, it’s important to make sure the research sample adequately represents this demographic.
• Over-represent historically underrepresented voices whenever possible. As a best practice, highlight voices “at the extremes” of the population. This provides the opportunity to serve the “average” better if we focus on the needs of the most in-need.
• When possible, modify screening quotas to have MINIMUM thresholds for under represented groups and MAXIMUM thresholds for people who are adequately and over represented. Example: ≥ 50% more than the existing population of the underrepresented group, ≤ 50% more than the existing population of the over-represented group.
• Lastly, make sure all demographic questions in a screener are OPTIONAL, giving respondents the autonomy to choose whether or not to provide their data.

Phase 4: Trust-building via consent, compensation, and data collection

Build trust upfront and throughout the data collection process by requesting consent continually throughout the study, offering fair and equitable compensation for participation, and recognizing the whole person in interviews.

Photo Source: www.pexels.com

Trust is a critical component of research. Without it, participants are more hesitant to share intimate details about themselves or their work and less willing to dive deep into the challenges and problems they face. Historically, there has been a documentation of mistrust towards health professionals and the healthcare system among Black, Indigenous, and People of Color (BIPOC), which can also translate to professionals working in digital health organizations. Therefore, in the context of a research study, it is the sole duty of the researcher to rebuild and maintain trust with participants, especially in setting where a perceived and/or real imbalance of power exists.

Here are three main areas in which we can start:

• Trust-building via consent
• Trust-building via compensation
• Trust-building via data collection

Trust-building via consent

Recollecting memories of personal experiences with systemic racism and discrimination in healthcare can be traumatic and/or uncomfortable to discuss in a research setting. Therefore, consent is required both upfront and continually throughout the study in order to ensure participants feel safe, empowered, and in control. Furthermore, consent itself is an ethical and legal requirement for research and required to inform voluntary participants on what the study is about, as well as how their data will be used.

How to implement

• Prior to the session, provide a participant “bill of rights” or consent form detailing the study and their role. Ensure the document achieves the following:
  1) Articulates clearly the terms surrounding confidentiality, privacy, and data ownership
  2) Acknowledges historical wrongdoings in research, while being completely open about the research process, emphasizing that participation is voluntary
  3) Describes current safeguards to protect research participants
  4) Contains simplified language to increase accessibility
• During the session, request permission to record audio and/or video, and be prepared if participants decline the recording. Consent must be verbally attained in the conversation for recording. For example: “Is it okay if we record this conversation? It will only be used with our direct research team so that we can be more present with you today and not ever shared in a way that identifies you.” If the participant declines, the recording should default to the note-taker; otherwise, offer to reschedule to a time where adequate note-taking is prepared in lieu of traditional screen or voice recording.
• During the session, request permission to discuss sensitive topic areas, and be prepared if participants are triggered by past or current experiences of trauma. At the top of every call, give a brief but accurate overview of the topics that will be covered, and ask if the participant feels comfortable and prepared to cover those topics. Let them know they are free to ask for clarification, pass on a topic, even leave the conversation at any point. Their privacy, safety, and experience are top priority.

Trust-building via compensation

Another way we can ensure a safe experience for our research participants is to compensate fairly. It is not acceptable to interview people at the benefit of a for-profit company without compensating them. Furthermore, research teams typically offer monetary compensation to make research participation a more revenue-neutral experience and/or recognize participants’ time. Overall, make sure reimbursement is fair and equitable.

How to implement

• Neutralize costs for individuals by providing fair reimbursement, especially when their participation might otherwise be too costly. Try to anticipate what the participant may be forfeiting to participate in the research. For example, are incentive amounts enough to cover costs associated with missing work or arranging childcare to participate in data collection? Also make sure the amount provided is also non-coercive; in other words, not so much that participants feel they’re unable to decline the opportunity.
• Make sure the monetary value of time is equitable across respondent groups. For example, are clinical professionals offered more compensation than other adult participants, such as patients/caregivers? Researchers should avoid inadvertent judgements about the value of people’s time based on their job or socioeconomic status.

Trust-building via data collection

Lastly, it’s critical to build trust during the process of collecting data, especially when gathering stories related to sensitive experiences with the healthcare system. Considering 1–1 interviews as one of the most intimate and commonly used research methods, there are several ways to prepare a safe and trusting environment when moderating dialogue and conversation.

How to implement

• Build trust by showing face on camera. As researchers, this brings an element of safety by showing vulnerability and will be able to convey how we are intently listening and responding to the participant. Also ask participants to share their camera as well. If they decline, assure that this is acceptable and they can turn it on or off at any point in the conversation.
• Build rapport by disclosing first name and organization. It is not expected for the researcher/interviewer to share more than a first name. Removing titles and last names builds a rapport that is about the person, while protecting the researcher’s privacy.
• Recognize the whole person, and respect other priorities in their life. Researchers can do this by following the participant’s line of thought to other areas of importance to them, and demonstrating patience and leniency if there are interruptions to the call or a need to reschedule. It’s important to maintain the participant at the center to show appreciation and value of their time.

Phase 5: Participatory analysis

Use participatory research to bring in multiple perspectives and experts in data analysis that acknowledge how race and larger systems of influence impact outcomes.

Photo Source: www.pexels.com

Inviting diverse collaborators to help analyze the data collected during a study is important for two reasons:
1) It helps the entire team understand what data were collected and empowers them to contribute to the development of insights.
2) It provides more than one perspective and interpretation of what the data are telling us.

Furthermore, incorporating a racial equity lens during this process means exploring the individual, communal, and historical contexts of race to clarify the patterns discovered, as well as informing the conclusions and recommendations. Given that, it’s important to have strong participation from a variety of stakeholders.

How to implement

• Engage diverse domain experts. Consider hosting a participatory workshop, consulting people individually, or a hybrid of both. Overall, inviting multiple perspectives will help eliminate some of the potential bias from being the sole interpreter or analyzer of the stories and data collected.
• Leverage data from mixed methods approaches. While this is important to include earlier in the process during the planning phase, it’s helpful to keep both stories and numbers in the analysis when trying to understand the lived experience of the community.
• Dis-aggregate data, analyzing intersectional experiences. One example is to look for insights across race by gender (e.g. differences between Hispanic women and men) and consider how these identities have changed over time in the interpretation of findings. Without doing this, researchers can unintentionally gloss over inequities that could be connected to factors like gender, age, or region, which can lead to “invisible experiences” going unacknowledged. On the other hand, there are risks and trade-offs in shifting the focus of analysis to a population that is already over-researched, which is why it’s important to have experts involved in the process.

Final thoughts

“Of all the forms of inequality, injustice in health care is the most shocking and inhumane,” stated by Martin Luther King, Jr.

There’s a dark side to medicine designed since its origin, spanning from the history of chattel slavery in North America in which physicians evaluated Black bodies for the sole purpose of demonstrating “soundness” for potential buyers, to the current discriminatory outcomes experienced in digital health technologies (see: “The New Jim Code” by Dr. Ruha Benjamin).

Another powerful quote is by Dr. Donald Berwick, former Administrator of the Centers for Medicare and Medicaid, who stated that “every system is perfectly designed to achieve exactly the results that it gets.” To this point, there’s an incredible need to:

1. Increase awareness of the racial and ethic disparities that exist in healthcare
2. Empower designers to develop more equitable experiences for marginalized communities

Considering the encoding of racial discrimination in the data- or AI-enabled healthcare technologies that have been recently developed (see: Optum example), it’s important to note that exclusion doesn’t have to be explicit, we just have to be indifferent to it. Implementing the inclusive practices above, therefore, is an important first step in rebuilding a system that is fair and just for all.

Raven L. Veal, PhD is a Design Researcher at IBM based in Austin, Texas. The above article is personal and does not necessarily represent IBM’s positions, strategies or opinions.