Disparities and Digital Health
As we are in the midst of Black History month, this is a good time to reflect on whether digital health interventions are being developed with and for a wide range of diverse populations. The burden of health problems in the U.S. is largely borne by people from low income and racial and ethnically diverse backgrounds. According to Kaiser Family Foundation data, African Americans fare worse than whites on 24 of 29 health measures, Native Americans fare worse on 20 of 29 health measures, and Latinos have poorer outcomes on 13 of 29 measures when compared to whites.1 These 3 groups also have less access to care and insurance coverage relative to whites. These data highlight a huge unmet need for digital health interventions and cause us to question whether we are adequately addressing that need. If we are serious about bending the healthcare cost curve, digital health interventions must be a part of the solution to health disparities in this country.
Another reason that ethnic minority populations in the U.S. are ideal targets for digital health interventions are that they use smartphones at the same or even higher rates than their white counterparts. In addition, Latinos and African-Americans are more likely to have access to the internet solely via their smartphones, making them more dependent on the devices.2 However, attempts at developing interventions with these groups in mind have been lacking. For example, a recent search for depression apps in Spanish yielded no results. Spanish language articles on the internet reviewing apps for mental health primarily referred to English language apps.3,4 When apps are developed for conditions such as depression, diabetes, and caregiving, they can be challenging to use by affected individuals who are less tech savvy. Fellow colleagues from UCSF found that a small group of diverse and low-income patients were only able to complete 43% of basic tasks such as enter a blood glucose value or mood rating.5
An additional problem with a lack of diversity in who engages in digital health is that the “big data” sets that are being created are not representative of those who need and could benefit the most. Artificial intelligence and machine learning hold great promise in personalizing interventions. The underlying algorithms rely on data from large numbers of people for determining the appropriate interventions for each individual. However, the bulk of data is coming from earlier adopters who tend to skew white and higher income relative to those that are most impacted by health problems. This is a significant issue that needs to be addressed by increasing diversity as machine learning and artificial intelligence play more central roles in digital health interventions.
If we truly want to make public health impacts with digital health interventions, we must develop with and for those experiencing the greatest burden of health and mental health problems. For those of us who develop apps, we must ask ourselves, “who are we developing our apps for?” Do we have ourselves, our friends, our families in mind as we develop these tools or are we considering people with the highest need? For example, we can make a much broader public health impact helping someone go from no physical activity to light or moderate activity than we can helping already active individuals to optimize workouts are push themselves further. Both approaches can obviously exist but addressing the person that has difficulty functioning will yield in more significant public health improvements. Participatory design is crucial in addressing these barriers to impact. For those apps that are already available, we should aim to disseminate them among populations with the highest need and determine the fit of interventions taking into account the impact of cultural context and educational levels which may be incompatible with existing apps.
Real results from digital health interventions have had difficulty matching early enthusiasm for the innovation. One reason that we have not achieved promise is that we may not be focused on populations that are in the most need and when we do focus on them, our tools do not match needs and abilities. Addressing these issues is not easy by any means but it is worthwhile if we want to make real public health impact. Let’s step up to the challenge!
- Sarkar, U., Gourley, G. I., Lyles, C. R., Tieu, L., Clarity, C., Newmark, L., … & Bates, D. W. (2016). Usability of commercially available mobile applications for diverse patients. Journal of general internal medicine, 31(12), 1417-1426.