Designing Technology and Telehealth Systems for Diabetes Care and Health Equity
One in three Americans has prediabetes, putting them at higher risk of developing diabetes, heart disease, and stroke. Too often, the quality of diabetes treatment, care, and outcomes are tied to socioeconomic status, race, and where patients live. Meanwhile, new technology exists that can continuously monitor a patient’s glucose and activity levels to provide better care, including with remote care that doesn’t require in-person visits.
Ramesh Johari is a 2020-2021 Stanford Impact Labs Design Fellow exploring ways telehealth and new technologies can improve diabetes care and make it accessible for more patients regardless of race or wealth. He is a professor in Stanford’s Department of Management Science and Engineering and brings his background in operations research, data science and experimentation, and optimizing online platforms to this work. He is working closely with colleagues at Stanford Children’s Health who are piloting remote monitoring and care for pediatric patients with type 1 diabetes, with a view towards developing a generalizable care model to bring high quality diabetes care to a much broader population.
Q. Why did you apply for Stanford Impact Labs’ Design Fellowship?
I saw the call for the impact labs design fellowship and for quite a long time, I’ve wanted to initiate a research agenda that allowed me to work on diabetes care, partly due to some personal and family experience. I wanted to build on significant technological advances in the care of diabetes—continuous glucose monitoring sensors, activity trackers, and telemedicine—and combine those with my own skills in data science and operations to try to contribute to making these kinds of advances accessible to a much wider population, especially knowing that diabetes is something underserved populations really have been hit hard by.
The typical person with diabetes is not going to get the level of care they should based on the technology we have available. We have technology that can work. What we need to develop is the user-friendly digital interface for health care.
When I learned I was mildly prediabetic, I figured out how to manage the situation because I had access to a whole host of tools including continuous glucose monitoring and activity trackers and I understood how to connect all the dots behind all the data and what it meant for my own health. I look back and try to imagine an individual being able to execute that on their own—it’s totally untenable. The typical person with diabetes is not going to get the level of care they should based on the technology we have available. We have technology that can work. What we need to develop is the user-friendly digital interface for health care so that people have the same experience I had in my own care.
Q. How can technology and telehealth improve diabetes care and reduce racial and economic disparities in treatment and outcomes?
Chronic disease burden in the United States is worse for those with lower socioeconomic status and is intermingled with racial disparities. Among chronic diseases, diabetes really stands out. If you’re a typical individual coping with diabetes, you’ll be lucky if you are seen four times a year, have your data reviewed four times a year, and get feedback from your health care team four times a year. Management of diabetes requires recognizing that blood glucose changes throughout the day in very fine-grained ways based on what activities you’re doing, what you’re eating, your sleep habits, your stress levels, etc. Those four visits a year—which is often the best case scenario—is such a mismatch.
The real vision here and the promise is to get to a point where patients can feel supported and have their needs addressed in an ongoing way with their care team through remote monitoring that doesn’t require them to come into the clinic all the time and that takes advantage of technology—like continuous glucose monitoring—together with telemedicine to be able to help them manage their condition at a distance. In doing so, we can start to close the gap in quality of care and outcomes that you see right now that are tied to socioeconomic status, race, and where patients live.
Q. What kind of research and partnerships are you designing?
I have been incredibly lucky to find a team of colleagues and collaborators across the university, bringing a wide range of backgrounds and expertise to push this work forward. David Scheinker, Clinical Associate Professor of Pediatrics who focuses on endocrinology and diabetes at the Stanford Lucile Packard Children’s Hospital, has been a key collaborator. David Scheinker and I see eye to eye on not overstating the role of technology, being humble about what data and scientific models can accomplish in the absence of working with humans, and the importance of explaining things well.
David Scheinker and I see eye to eye on not overstating the role of technology, being humble about what data and scientific models can accomplish in the absence of working with humans, and the importance of explaining things well.
David is a mathematician by training who has spent his career focused on health care operations challenges; he also holds an adjunct appointment in my department (Management Science and Engineering). He runs Systems Utilization Research for Stanford Medicine (SURF), a group that serves as a bridge from engineering to Stanford Medicine. Through this role, David has been a crucial guide for me to bring my expertise into the world of health care. David and I have been co-advising a fantastic management science and engineering graduate student, Johannes Ferstad, who has been invaluable in helping to bring this research to life.
David, Johannes, and I have been collaborating closely with David Maahs, Professor of Pediatrics and Division Chief of Pediatric Endocrinology at Stanford Lucile Packard Children’s Hospital, and Associate Director of the Stanford Diabetes Research Center, and Priya Prahalad, Clinical Associate Professor of Pediatrics focused on endocrinology and diabetes; both have a deep interest in the exact issues I’m investigating. We’ve been lucky to have the support of Rick Mazjun (Chief Operating Officer), Natalie Pageler (Chief Medical Information Officer), and the leadership team of Stanford Children’s Health. I’ve also been fortunate to be able to collaborate more broadly in this work with a wide range of talented researchers, postdocs, graduate students, and undergraduates who are invested in the same issues.
I have interests in four levels of challenges. The first is technology: how do we build a technology platform that leverages continuous glucose monitor data for remote monitoring and management of patients with diabetes? Second is clinic workflow: we have to find ways to ensure telemedicine doesn’t create additional staff or financial burden in clinics, especially for clinics in under-resourced areas. In an ideal world, the combination of telemedicine and remote patient monitoring has the potential to both increase clinic efficiency and improve patient outcomes at the same time. Third is economics, cost, and scaling: what combination of insurance reimbursement models for remote care, human resources (e.g., diabetes educators) per patient, and technology innovation can lead to wide-scale deployment of these advances for underserved patient populations? I’ve been collaborating closely with David Maahs, Priya Prahalad, and David Scheinker on these issues.
We have to find ways to ensure telemedicine doesn’t create additional staff or financial burden in clinics, especially for clinics in under-resourced areas. In an ideal world, the combination of telemedicine and remote patient monitoring has the potential to both increase clinic efficiency and improve patient outcomes at the same time.
The fourth overarching challenge is equity in access to care. You can develop all the technology you want, but if patients aren’t getting through, you’re not solving the problem of who has access to health care, where, and when. We’re also looking at different kinds of telehealth, which in some cases could be a phone call, and in others, a fully video visit, or more. I’ve been fortunate to have begun a nascent collaboration with Lisa Chamberlain, who leads a health equity research group at Stanford Medicine; together with Brendan Watkins (Chief Analytics Officer at Stanford Children’s Health) and a broader research team, we’re aiming to use granular data on patient pathways to describe variation in how patients of different backgrounds and technological means receive care.
More broadly, we’re seeing opportunities to engage with people facing similar challenges using telehealth in their work and who care about delivering equitable care, including those working to improve equity in pediatric digital health. I’ve also been fortunate to receive support from others who have expertise in this area. For example,Lisa Goldman Rosas, from the Department of Medicine and who leads a start-up lab working to increase access to healthy food to reduce chronic disease, connected me with collaborators who helped me learn about the technology barriers that face underserved populations, and the subsequent disparities in telehealth-based care that these barriers create. We’re starting to connect with a network of funders and policymakers to support this type of work in the future, too.
Q. How does your background in operations research and optimizing online platforms for organizations like Upwork, Airbnb, and Stitch Fix shape your approach?
One of the lessons I took away from my experiences with online platforms and the technology space is that the best work happens in a very tightly integrated loop between the technological system and the social system. In those settings, on one hand, we’re iterating on the technological system rapidly. On the other hand, we’re measuring and monitoring its response in the social system on a similar cadence. We’re feeding what we learn back to iterate the right version of the tool. It makes it clear that there’s a really symbiotic relationship between progress on the technological front and serving a social need.
One of the lessons I took away from my experiences with online platforms and the technology space is that the best work happens in a very tightly integrated loop between the technological system and the social system.
As a technologist, the point that often gets missed is that social systems are much slower to change than technological systems. The organization of healthcare is in part a social system. One of the challenges we have is to figure out models or approaches that could be workable—say, the right kind of billing structure with offices and insurance companies for telehealth visits—but we have to see how it plays it out in clinics and with providers and patients and iterate. We have to tighten the loop between the technological system and the social system.
Q. You started this work just as the coronavirus pandemic began. How did that affect your work, especially the focus on telehealth which many of us learned to use this past year when we couldn’t visit our doctors and clinics in person?
The pandemic really accelerated telehealth and created a moment in time to make telehealth part of the research and learn from what happened. At the start of the pandemic, uptake of telehealth was really high because people didn’t have much of a choice. Many providers and insurance companies and patients found ways to make it work. But we absolutely saw disparities, too, for a whole host of reasons from connectivity, to work flow and billing problems, and more.
It’s important to note that connectivity alone won’t solve all the problems. Esli Osmanlliu, David Scheinker, and I did an analysis, and it even shocked me, to see that Americans living in counties with the highest prevalence of diabetes and heart disease have a six times higher prevalence of poor broadband access compared to counties with low prevalence of these chronic diseases. Counties in the Southern United States with the highest proportion of Black people were particularly affected. And that’s based on the Federal Communications Commission’s really broad definition of broadband access; an important question is whether the situation is even worse in reality, if we had a more accurate definition of broadband access.
It even shocked me, to see that Americans living in counties with the highest prevalence of diabetes and heart disease have a six times higher prevalence of poor broadband access compared to counties with low prevalence of these chronic diseases.
It’s reasonable to expect a 30 to 40 percent steady rate of telehealth post-COVID. We need to make sure we don’t just revert to the pre-2020 era, but stop first to use a feedback loop to capture what we learned with providers, patients, and policymakers, too, so we can continue to find ways to use telemedicine to reach patients where they are.
Q. What has surprised you during the fellowship and your work so far?
What really surprised me was how quickly I was able to start engaging with the problem I care about. It’s the main advice I’d have for anyone thinking of applying. Our working model as academics is that we do what we do, and there’s another thing we might want to do, but the barriers to entry are so high that you don’t know where to get started. That ranges from “I don’t have the right expertise for that,” to “I don’t have the right network,” to “I don’t think I’ll be able to do cutting edge work there because it will take me too long to get where I want to go.” The fellowship structures your thinking so you realize as quickly as possible that you don’t have to be stuck on the how do I get started question.
What really surprised me was how quickly I was able to start engaging with the problem I care about. The fellowship structures your thinking so you realize as quickly as possible that you don’t have to be stuck on the how do I get started question.
What I found striking about this class of problems is that they are so under-resourced from a social standpoint and that’s an incredible arbitrage opportunity. If someone is drawn to a problem with a strong social impact element, it’s almost certainly going to be the case that there are far more problems than there are people who can address them. That lowers the barrier to entry to rapidly build up collaborations and a network and reaching a point where you can make a difference. It’s pretty exciting for students and faculty to know that what we perceive as a barrier to entry is actually not so large.
By far the single most important thing for me was the commitment device of the fellowship. We meet three hours a month. It’s structured enough that you have to do something before every session and you have to take something away from every session. And actually, that’s enough to get someone to work on a continuous basis every month. It works and was really rewarding, especially this year.
The challenge I’m grappling with now is being more efficient with my time. There’s more that can be done than I have time for on my own, and I didn’t anticipate that being the problem by the end of this academic year. I started the year wondering how do I get started, where do I go? To end the year thinking I can’t do every one of these things and to need to focus where it can have the most impact is a great feeling. That says there’s so much there.