Digital Therapeutics Study Shows How FDA Authorizes, and Medicare Pays for, Therapy Software

Half of the digital therapeutics authorized by the Food and Drug Administration in a recent four-year period reached the market without a pivotal randomized controlled trial, according to a review published May 20 in npj Digital Medicine. Melissa Amyx, Viet-Thi Tran and colleagues identified 45 digital therapeutics, software products that generate and deliver an intervention for a disease or condition, with an FDA decision from 2019 to 2022, and traced 436 studies of them.

The evidence varied widely. For 31.1% of the products the authors found studies in every category they tracked, from pilot work to post-market research; for 11.1% they found none. Some 46.7% had a pivotal RCT. For 13.3%, more than 10 years passed between the first study or publication and FDA authorization, while 53.3% had no results available before they were authorized. Products first authorized through the 510(k) pathway were more likely to lack a pivotal RCT than those authorized through premarket approval or De Novo, 65.6% versus 23.1%.

The authors concluded that "more transparent, timely dissemination of results and consistent evidence standards, appropriate to the level of risk of the individual DTx, are needed." Their findings follow from how such software is regulated and paid for.

Software as a device

FDA treats software that performs a medical purpose on its own as a medical device. Its Software as a Medical Device page uses the International Medical Device Regulators Forum definition, "software intended to be used for one or more medical purposes that perform these purposes without being part of a hardware medical device," and separates it from software inside a hardware device. In the study, most of the products (71.1%) combined hardware and software, and 73.3% required a prescription.

There is no separate pathway for digital therapeutics. The study notes that most are regulated through the same routes as other devices, ranging from the 510(k), in which a lower-risk device can claim substantial equivalence to a predicate and "thus may not require clinical evidence," to premarket approval, which requires valid proof of safety and effectiveness from clinical investigations for higher-risk devices. Because clearance turns on comparison with an earlier device, a product can reach the market through a 510(k) without a new randomized trial of its own. FDA has also issued guidance on software, mobile apps and artificial intelligence, and ran a software precertification pilot, the authors note.

Updates are common. More than half of the products (57.8%) had at least one further FDA clearance for a modification after the first authorization, for 136 authorizations in total.

How Medicare pays

Authorization does not settle payment. Starting Jan. 1, 2025, Medicare created three HCPCS codes for digital mental health treatment, according to a CMS Medicare Learning Network booklet: G0552 for supplying the device with initial education and onboarding, and G0553 and G0554 for the first 20 minutes and each additional 20 minutes of monthly treatment management tied to the patient's use of the device. CMS says the device must be “FDA-approved” to bill G0552, in the booklet's wording, the billing practitioner must bear the cost of supplying it as part of an ongoing treatment plan, and the management codes may be billed only while the patient is actively using it.

For 2026, CMS expanded the policy to devices cleared under a 510(k) or granted De Novo authorization and classified under 21 CFR 882.5803, digital therapy devices for attention deficit hyperactivity disorder, when used as an adjunct to clinician-supervised behavioral health care, according to its summary of the 2026 physician fee schedule final rule. G0552 remains contractor-priced, while G0553 and G0554 carry national pricing, and CMS says it will pay only for uses indicated under each device's FDA classification.

The codes tie Medicare payment to the FDA classification, not to the type or amount of clinical evidence behind a product. The npj Digital Medicine authors reviewed evidence only, not payment, and covered products with FDA decisions through 2022.