Prescription digital therapeutics, software authorized by the FDA to treat a medical condition, were supposed to be the category that proved software could carry the same evidentiary weight as a drug. A systematic mapping study published in npj Digital Medicine took FDA-authorized DTx products with decisions between 2019 and 2022 and catalogued every study behind each one, before and after authorization. The results explain why payer coverage for this category has been so uneven.
How much evidence actually backed these authorizations?
The mapping study classified evidence into development and pilot studies, pivotal randomized controlled trials, and post-authorization studies. The consistent finding: the size and rigor of pivotal evidence varied enormously across products, with some DTx authorized on a single small randomized trial and others backed by larger, better-powered programs. Source: Mapping the evidence supporting FDA-authorized digital therapeutics, npj Digital Medicine.
This variation is structural, not accidental. Most authorized DTx went through the De Novo or 510(k) pathway as software devices, not through the drug approval pathway, which means the evidentiary bar was set by device regulation, not by the two-adequate-and-well-controlled-trials standard that governs pharmaceuticals. A device-equivalent standard applied to a behavioral intervention produces exactly the heterogeneity the mapping study found.
What does the condition-specific evidence look like?
Where disease-specific meta-analyses exist, the picture is more encouraging than the authorization-level mapping alone suggests.
A 2025 systematic review and meta-analysis in the Journal of Medical Internet Research pooled randomized controlled trials of digital therapeutics for adults with diabetes and found consistent, if modest, improvements in glycemic control compared to usual care. Source: JMIR meta-analysis, diabetes DTx.
For prescription digital CBT-insomnia, the real-world DREAM study followed patients using a prescribed digital therapeutic outside the trial setting and found reductions in insomnia severity that tracked with the RCT effect sizes, alongside improvements in co-occurring depression and anxiety symptoms. Source: DREAM study, PubMed.
A 2025 sham-controlled multicenter RCT for a digital therapeutic targeting temporomandibular disorders reported statistically significant improvement over a sham digital comparator, which matters because sham-controlled DTx trials are still rare and control for the placebo effect of simply engaging with an app. Source: JMIR sham-controlled TMD trial.
Why build a composite index at all?
Because no single number currently lets a payer or clinician compare DTx products the way a NNT (number needed to treat) or an FDA drug label lets them compare medications. A 2025 proof-of-concept paper proposed a Composite Digital Therapeutic Index (cDTI), a multidimensional framework scoring FDA-authorized DTx on evidence quality, engagement, safety and other axes, applied retrospectively to already-authorized products. Source: cDTI framework, PMC.
The fact that researchers felt compelled to build this index in 2025, years after the first DTx clearances, is itself informative. It signals that the field lacks a standardized way to answer the buyer's most basic question: is this product's evidence good enough for my population.
The three evidence gaps payers keep flagging
Durability. Most pivotal DTx trials run 8 to 12 weeks. Chronic conditions require years of adherence. Very few products have published data past six months.
Engagement decay. Digital interventions depend on sustained use, and open-label, real-world engagement drops well below trial-level engagement, which was often supported by study coordinators nudging participants. The DREAM study is one of few real-world persistence datasets published for this category, and even there, effect sizes assume ongoing use.
Active comparator design. Many pivotal DTx trials compared the app to waitlist or treatment-as-usual, not to an active digital placebo. The TMD sham-controlled trial is closer to the standard payers actually want, and it remains the exception rather than the rule across the category.
What this does not prove
The evidence reviewed here does not prove that digital therapeutics as a class are equivalent to pharmacotherapy in effect size, nor that authorization-level heterogeneity has been resolved. It does not establish long-term adherence outcomes for most authorized products, and it does not mean every FDA-authorized DTx carries the same evidentiary weight, despite carrying the same regulatory label.
The takeaway
"FDA-authorized" tells a payer almost nothing about trial size or rigor for a digital therapeutic, because the category was authorized through a device pathway with a variable bar. The products with sham-controlled trials, real-world persistence data, and published long-term follow-up are pulling ahead commercially for exactly that reason. Anyone evaluating a DTx contract should ask for the pivotal trial's comparator arm and its post-authorization real-world evidence, not just the clearance letter.
