Does diabetes technology have a race problem? Are the latest devices least available to those that need them most?
On the second day of the ADA’s annual Scientific Sessions conference, a panel of experts discussed the topic of systemic bias in diabetes technology access.
The broad consensus from the panelists was that diabetes technologies – namely, the insulin pump, the continuous glucose monitor (CGM), and the closed loop systems linking the two – are far more likely to be recommended to and adopted by people with diabetes that are already well-controlled. And this population is likely to be younger, wealthier, better educated, and whiter than those without access to these devices.
The neglect extends even into the experimental research for technology itself: the overwhelming majority of studies devoted to new devices test a fairly homogeneous population of well-controlled Caucasians.
The highlight of the talk was a searing presentation given by Jill Weissberg-Benchell, a professor of psychiatry and behavioral sciences at Northwestern University. Ms. Weissberg-Benchell described patterns of bias and neglect affecting nearly every aspect of the application of technology in diabetes care.
Many nationwide studies have found that the American medical establishment is riddled with bias, a likely explanation for the deep health disparities between whites and blacks (and other minorities). Black Americans are likely to receive lesser quality healthcare, and to have worse insurance. It is therefore perhaps unsurprising, but no less tolerable, that the same problems would be present in the diabetes world. It has been known for years that African-Americans and other minorities have higher A1c’s and worse diabetic outcomes than do white patients.
Ms. Weissberg-Benchell’s talk focused on the frequency with which diabetes technology is recommended to and adopted by different populations. Several data reviews, for example, have showed that minorities with Type 1 diabetes are less likely to be offered or recommended the use of an insulin pump, even after controlling for socioeconomic status and other demographic factors. CGM use is also much less frequent among minorities with T1D. This is all despite the fact that these populations have higher A1c’s and therefore arguably need these devices more than most.
One particularly troubling study showed that Caucasian and Latinx youth were significantly more likely to be told to check glucose as often as four times per day than African-American youth. As Ms. Weissberg-Benchell put it, “that’s a grave concern, if we’re making different recommendations based on the color of a patient’s skin.”
Lower socioeconomic status also looms large as a trigger for bias. Ms. Weissberg-Benchell pointed to a survey in which doctors admitted that they made assumptions about a family’s ability to pay for various treatments and allowed that to impact their recommendations. Those assumptions may prevent even patients that can afford the latest device from ever getting the option. “Before providers sit down and gather information from their families … they’re already deciding who is appropriate to have a technology recommended to them.”
The bias and neglect actually begins with the studies designed to prove that new diabetes technology is safe and efficacious. Taking the example of automated insulin delivery systems (also known as closed loop systems, or the artificial pancreas), Ms. Weissberg-Benchell raised doubts about impressive study results that “may not represent the experience of perceptions of the broader community of individuals with T1D or their loved ones.”
Ms. Weissberg-Benchell reviewed 81 studies of automated insulin delivery systems, all published in peer reviewed journals since 2014. Seventy-six of those studies either did not consider patient ethnicity or used exclusively Caucasian participants. Of the 5 that reported any diversity in the population, Caucasians represented between 85% and 96% of participants.
“So what we know about automatic insulin delivery systems is almost exclusively from the perspective of Caucasian participants.”
Of those 81 studies, none reported on socioeconomic status. Only 6 reported a mean A1c at baseline greater than 8.5%, and 75 required previous experience with insulin pumps. By recruiting participants with comparatively well-controlled diabetes, researchers can avoid including patients thought to be “poorly compliant.”
“So I have to ask … what is the generalizability [of these studies] to people who struggle the most with daily diabetes demands, or experience the absolute highest levels of diabetes distress?”
These patterns of bias seem almost perversely designed to exacerbate already-existing health disparities. The CGM, insulin pump and artificial pancreases are astounding technological advances: but in practice they appear to have been mostly designed for and are certainly mostly used by the population of people with diabetes that need it least, patients that are affluent, well-educated, overwhelmingly white and already well-controlled. Conversely, the very people that most need significant help with diabetes management – typically marginalized populations – are also those least likely to receive this gamechanging technology. “In some ways, these technologies are ideal for those that are hurting the most, and yet they’re hardly ever used.”
While there has been a paucity of study of the effect of the latest diabetes technology specifically in more vulnerable populations, what evidence we do have suggests that they can have very powerful effects. A 2018 study, for example, found that teens with “very poor glycemic control” enjoyed tremendous improvements in A1c when given an insulin pump. This is exactly the sort of group that most clinicians would generally be hesitant to recommend an insulin pump for.
Another speaker, Dr. Kimberly Spiro was able to speak to potential of technology in the most vulnerable cases. Dr. Spiro is the clinical director of the NICH program at the Oregon Health & Science University, which is dedicated to helping children with complex health conditions whose families cannot provide adequate care.
The Type 1 children and teens that Dr. Spiro works with, all of whom entered the program with high A1c’s and after experiencing multiple episodes of DKA, face immense challenges. These youths are from families dealing with “systemic inequities including poverty, abuse, neglect, mental health issues, substance abuse, and histories of trauma that really limit daily functioning. Many of the families report experiencing discrimination and marginalization of the health care system.” To put it mildly, these children do not have the support needed to thrive.
Before enrolling in the NICH program, only 5.8% of these youths had an insulin pump, and zero had access to a CGM. Dr. Spiro’s work has facilitated technology access for these unfortunates, and she has seen many A1c’s drop and young lives set back on the right track. But while the NICH program has been a success, it’s an exception: “Our clinical care model isn’t designed to provide adequate support to address the social determinants of health of the [patients] with the highest degree of challenges.”
“Really, these patients are slipping through the cracks.”
Ms. Weissberg-Benchell believes that there is a moral imperative for improving access to diabetes devices: “we are obliged to leverage the power of technology to the broader population.”
“I think we have to check our assumptions, as both clinicians and researchers. I think we have to check our biases. I think we have to take the time that is necessary with our [patient] families, and that’s going to be a lot more time than we typically spend. I think we have to listen a lot more. I think we have to talk a lot less. I think we have to judge a lot less.”