Digital Health · us
Enter First, Gather Evidence Later: Dexcom Becomes First Participant in FDA Digital Health Evidence Pilot
The FDA is allowing Dexcom’s glucose health program to collect real-world data in specific healthcare payment settings, in an attempt to connect patient use, reimbursement, and premarket review into a single evidence chain; however, selection for the pilot does not mean the product’s effectiveness has been recognized.
Digital tools for chronic diseases have long faced a dilemma: without access to routine care, it is difficult to accumulate sufficient data to demonstrate benefits; without such data, it is in turn difficult to obtain regulatory authorization and stable reimbursement. The U.S. Food and Drug Administration (FDA) is now attempting to loosen this knot by selecting Dexcom as the first participant in the TEMPO Digital Health Device Pilot, allowing the product to undergo real-world evaluation first within a limited healthcare system.
The Dexcom Glucose Health Program will operate in conjunction with the ACCESS chronic disease care model of the Centers for Medicare & Medicaid Services (CMS) Innovation Center and will be used in care related to prediabetes and type 2 diabetes. According to the intended use published by the FDA, patients, healthcare professionals, or caregivers may use it to monitor metabolic and nutritional status and obtain personalized guidance, real-time data, and AI-generated insights to help adjust behavior and make care decisions.
The program will also evaluate whether integrated digital health metrics can support screening for prediabetes and type 2 diabetes, and whether they can help improve glycemic control. Potential uses listed by the FDA also include lowering glycated hemoglobin, but the current announcement does not specify the input data used by the algorithm, the training and validation populations, the comparator-group design, or the intended effect size. Clinical benefit therefore cannot yet be inferred from selection alone.
What distinguishes TEMPO is that the FDA may, at its discretion, decline to enforce certain premarket authorization, investigational device, or other regulatory requirements for a device in specific use settings within the ACCESS model. This is not blanket clearance: the same use outside the ACCESS system is not covered by the pilot. The FDA has also stated explicitly that it has not yet evaluated the effectiveness of participating devices for these purposes.
In exchange, manufacturers must propose risk-mitigation plans; collect, monitor, and report real-world performance; establish patient outcome measures and statistical analysis plans; and may also be required to periodically report adverse events and emerging risks. The FDA expects participants ultimately to submit a 510(k) or other applicable marketing application using pilot data together with other supporting evidence. In other words, enforcement discretion provides an opportunity to gather evidence, not advance approval.
The real question this experiment must answer is not only whether continuous data and AI prompts can help users identify metabolic risks earlier, but whether they are reliable and interpretable in routine care settings and can deliver measurable health improvements. The announcement has not yet provided details on participant numbers, follow-up duration, data quality controls, population representativeness, or privacy governance. If the pilot results are later to be translated into broadly applicable indications for use, these gaps will all be critical to assessing the strength of the evidence.