FDA Workshop Serves as Inspired Think-tank for the Future of Digital Diabetes Technology
Wednesday, February 04, 2015
The recent flood of diabetes-related apps do everything from track hypoglycemic events to instantly calculate a patient’s insulin doses. But what if a bolus calculator tells a patient to dose 15 units when it should really by five? And what about device compatibility with all these new software options? As apps incorporate more medical recommendations and diabetes devices become more sophisticated data-tracking machines, the FDA grows more concerned over their safety and usability.
To examine these issues, the FDA held a public workshop this past November entitled Regulatory Science Considerations for Software Used in Diabetes Management. The broad scope of opinions discussed— patients, caretakers, clinicians, and industry representatives were all in attendance—will help the FDA construct a more fully informed action plan for regulating these sorts of apps and software.
The daylong conference focused on two main topics: first, Interoperability Between Diabetes Devices and Management Systems, and second, Insulin Bolus Calculators.
Insulin Bolus Calculators
Most of the day was devoted to discussions on bolus calculators. Howard Wolpert, M.D., Director of the Insulin Pump and Continuous Glucose Monitoring Programs at Joslin and Associate Professor of Medicine at Harvard Medical School, contributed heavily to the discussion by offering his clinical expertise. “These apps are extremely valuable to people who are using them,” he says in an interview after the conference. “But because there is an unregulated realm, there is no guarantee that they’re operating the way people think they ought to be.”
In his presentation, Dr. Wolpert noted that all bolus dose recommendations – whether from a calculator or other means – are inherently inaccurate, since they don’t adjust for other factors affecting insulin requirements including activity and stress. He outlined the numerous limitations in insulin bolus calculators such as inaccurate carb counting, as well as exclusion of other factors like dietary fat that can markedly affect insulin requirements from consideration in the dose calculation.
Howard Look, CEO of Tidepool, Palo Alto, CA, and parent of a type 1 patient, pointed out that even an unapproved bolus calculator app from the App store may be better than no bolus calculator at all; patients make significant errors in their dosing math. He feels that these mistakes might carry much more risk than using an unregulated app.
Despite Look’s opinions, there was clear consensus among speakers that software and apps that provide bolus calculators should have some basic validation, quality control, and human factors testing.
Dr. Wolpert emphasized that there are two distinct types of bolus calculators that require different levels of oversight. In one type, insulin dosing algorithms are automatic, only taking a patient’s raw data into account. These types of apps are not reviewed or altered by a clinician to see if the insulin dose is appropriate, making them more dangerous. In Dr. Wolpert’s opinion, for those types of apps where the clinician is out of the loop, the FDA should have clinical trials to test for safety and effectiveness.
The other type of app operationalizes paper-based dosing recommendations that are currently recommended by clinicians in routine care, basically doing math that patients would be doing themselves. With that type of app, the FDA has an important role in ensuring that there is appropriate bench validation that the calculator does the math correctly, as well as evaluation of the user interface.
Dr. Wolpert explains that going forward, the FDA must strike a delicate balance between allowing progressive ideas to make it to the public and testing a new apps’ safety. “You need innovation,” says Dr. Wolpert, “but you need oversight to make sure the apps that become available are safe and meet certain quality standards.”
In a shorter series of presentations, various speakers discussed the current state of diabetes device data and how it is collected and utilized by patients. FDA scientific reviewer and type 1 patient Alain Silk, Ph.D., addressed the downsides to diabetes systems that don’t work together, including inefficient and ineffective data review with physicians (multiple devices and summary reports), as well as frustration, hassle, and burden for patients.
The answer, it seems, is for the FDA to introduce industry standards. Howard Look argued for standard data. He compared the current problems with diabetes devices to the evolution of the digital camera. At one point, camera companies each used their own software to download photo files. Eventually, all camera companies settled on a standard .jpg file which improved the user experience and allowed for more innovative software platforms.
If diabetes devices moved away from proprietary software and allowed users to view their data from anywhere, such as a phone or remote computer software, it would vastly improve user capabilities. “Digital cameras got it right,” said Look. “Access: your choice. View: your choice.”
FDA representative Dr. Courtney Lias, Director, Division of Chemistry and Toxicology devices, Office of In Vitro Diagnostic Device and Radiological Health, CDRH, seemed optimistic about the potential of these new diabetes device standards as well as app testing and regulation. She suggesting the FDA would be open to officially recognizing them, but it was unclear what the process would be to create standards and length of time to do so.
To view a recorded video of the workshop, visit the Webcast section of the workshop registration page.
Page last updated: May 21, 2019