A digital therapeutic that clears US review through the FDA's De Novo or 510(k) pathway has, in the United States, answered the regulatory question and inherited the reimbursement one: coverage is negotiated payer by payer, plan by plan, with no single national decision. Developers reading Germany's DiGA pathway translate it as the inverse, a single national reimbursement decision that rewards the regulatory work already done. That translation is where the difficulty begins. The German regime is not a national coverage switch. It is a price negotiation the developer does not control, running on an evidence clock it cannot stop, and from 2026 it pays in part for results rather than for the prescription.
1. The Reimbursement Pathway US Developers Misread
The DiGA pathway was built by the Digitale-Versorgung-Gesetz of 2019 (DVG), which inserted the insured person's entitlement into § 33a SGB V and the directory into § 139e SGB V.1Sozialgesetzbuch Fünftes Buch (SGB V), § 33a (entitlement to digital health applications) and § 139e (BfArM directory; Fast-Track within three months of a complete application, extendable by up to three more in justified individual cases; provisional listing for trial up to 24 months).2Digitale-Versorgung-Gesetz (DVG) vom 9. Dezember 2019 (BGBl. I S. 2562); created the DiGA pathway by inserting § 33a and § 139e into the SGB V. The BfArM runs a "Fast-Track" assessment, deciding within three months of receiving a complete application, a period extendable by up to three further months in justified individual cases, and a product that reaches the directory is reimbursable across statutory health insurance, which covers roughly 74 million people. To a US reader, that architecture looks like a single national formulary decision, the thing the fragmented US payer landscape never delivers.
The entitlement is the part that does not translate. Under § 33a SGB V a listed DiGA is a statutory benefit, prescribed by the treating physician or psychotherapist or approved by the fund on the insured person's application, and the fund must provide it; it is not a coverage decision the fund takes product by product. To a US developer accustomed to negotiating each plan, the absence of that negotiation reads as a feature. What it does instead is concentrate the evidence and the pricing leverage at two national chokepoints, the BfArM listing and the GKV-Spitzenverband price, where a refusal cannot be routed around and where there is no second bite.
The misreading is in the word "listing." A DiGA can enter the directory in one of two states. It can be listed permanently, on proof of a positiver Versorgungseffekt, a positive healthcare effect that is either a medical benefit or a patient-relevant improvement in the structure and processes of care. Or it can be listed provisionally, "for trial" (Erprobung), for up to twelve months and extendable by up to twelve more, while the manufacturer generates that proof.3Digitale Gesundheitsanwendungen-Verordnung (DiGAV) vom 8. April 2020 (BGBl. I S. 768); as amended by the 2. DiGAV-Änderungsverordnung (BGBl. 2026 I Nr. 22), in force 1 February 2026. The provisional door is the one most US developers walk through, and it is a door that closes on a schedule.
The price architecture compounds the misreading. For the first twelve months after listing, the manufacturer sets its own price, subject to the Höchstbeträge, maximum amounts for groups of comparable applications that cap even that interim price where the framework agreement has set one. Read against a US launch, that looks like pricing freedom; it is a one-year window under a group ceiling. From the thirteenth month the price is the Vergütungsbetrag negotiated with the GKV-Spitzenverband, the single federal association of the statutory insurers, and if no agreement is reached within nine months of the listing itself, months before that price takes effect, a Schiedsstelle, an arbitration board, has three months to set it.4SGB V (n 1), § 134 Abs. 1 Satz 2 (the negotiated Vergütungsbetrag applies after the first year), § 134 Abs. 5 (manufacturer's interim price; Höchstbeträge; thresholds), § 134 Abs. 2 Satz 1 (Schiedsstelle sets the price within three months where none is agreed within nine months of listing), § 134 Abs. 4 (Rahmenvereinbarung). The counterparty is effectively the only buyer, negotiating once for the whole statutory market and publishing what it pays. The distance between the manufacturer's first-year price and the later negotiated figure is a matter of record, and a recurring subject of policy criticism.5GKV-Spitzenverband, 'DiGA-Bericht des GKV-Spitzenverbandes 2025' (published 1 April 2026, § 33a Abs. 6 SGB V), on directory removals for unproven benefit, restricted conversions, and the gap between first-year and negotiated prices.
A German listing is less a destination than a negotiation with the only buyer: a price the developer does not set, against a benefit it must keep proving.
2. Outcome-Based Pricing and the Twenty-Percent Rule
That negotiation has changed character. The Digital-Gesetz (DigiG), in force 26 March 2024, rewrote § 134 SGB V to require that, from 1 January 2026, every DiGA price agreement fix a success-dependent component of at least 20 percent of the Vergütungsbetrag; agreements concluded before that date without such a component had to be brought into line by the same deadline.6Digital-Gesetz (DigiG) vom 22. März 2024 (BGBl. 2024 I Nr. 101), in force 26 March 2024; opened class IIb eligibility, added the AbEM (§ 139e Abs. 13 SGB V), and the success-pricing rule in § 134 Abs. 1 Satz 3 SGB V (at least 20 percent from 1 January 2026). From 2026, at least a fifth of the negotiated price turns on something other than the prescription being written.
The interaction with the first-year price sharpens the effect. The price a manufacturer sets for its opening twelve months is not only the revenue it collects before the negotiation; it is the anchor the GKV-Spitzenverband discounts from, and the negotiated price it anchors must, under the 2026 rule, decompose into a fixed part and a contingent part of at least a fifth. A developer that sets an opening price on US launch logic, list price first and evidence of value later, is fixing the reference point for a negotiation whose rules require part of that price to ride on a metric the developer does not yet control.
Stated that way, the rule reads like a familiar arrangement: put a portion of price at risk against performance. Applied, it is far less settled. The statute fixes the floor, at least 20 percent, but not the metric. What "success" measures, which endpoint defines it, over what horizon, and against what baseline are questions for the price agreement and for the framework agreement (Rahmenvereinbarung) that standardizes those agreements, not for the manufacturer's listing dossier. A developer can satisfy every BfArM requirement for the directory and still arrive at a pricing conversation in which the definition of the outcome its revenue depends on is shaped by the counterparty that pays for it.
The success component does not float free. It is tethered to data the manufacturer must generate continuously. DigiG added the anwendungsbegleitende Erfolgsmessung, an application-accompanying performance measurement under § 139e Abs. 13 SGB V, whose methodology the second amendment to the DiGAV concretized.7SGB V (n 1), § 139e Abs. 13 (AbEM, continuous performance measurement published by the BfArM); methodology per DiGAV (n 3); quarterly recording from the third quarter of 2026, first transmission due by 15 April 2027. A "success-dependent price component" is a pricing mechanism, not a clinical-outcome finding, and the two are easy to conflate. A product can demonstrate a positive Versorgungseffekt sufficient for listing and still see at least a fifth of its price exposed to performance measured after launch, on a population and in conditions that the controlled study did not capture.
3. Class IIb and the Vanishing Trial Route
The same reform widened the front door. DiGA had been confined to lower-risk devices, the risk classes I and IIa of the EU Medical Device Regulation (MDR).8Regulation (EU) 2017/745 [2017] OJ L117/1 (MDR), Art. 51(1) (risk classes I, IIa, IIb, III, classified in accordance with Annex VIII); DiGA were originally confined to classes I and IIa. DigiG opened eligibility to class IIb, the band that captures more consequential software such as telemonitoring and AI-assisted diagnostic tools. The change is often described as a 2026 development. The eligibility took effect with DigiG in 2024, which at the same time wrote the evidence standard into the DiGAV: a class IIb application demonstrates its benefit through a prospective comparative study. What arrived later is operational detail, the BfArM's Fast-Track guidance folding class IIb into the application procedure in December 2025, and the second DiGAV amendment, in force 1 February 2026, concretizing the performance measurement every permanently listed product must feed.
The widening is narrower than it looks. The provisional Erprobung on-ramp that lower-risk DiGA use to enter the directory before proving their effect is not available to class IIb. A class IIb application reaches the directory only by permanent listing, with the benefit demonstrated up front, by completed study, at the time of application, and the benefit that counts is itself narrower: where a class I or IIa application may rest on either limb of the positive Versorgungseffekt, a class IIb application must show a medical benefit; the patient-relevant improvement in the structure and processes of care that carries many lower-risk listings does not suffice.9BfArM, 'Das Fast-Track-Verfahren für digitale Gesundheitsanwendungen (DiGA) nach § 139e SGB V' (Leitfaden, Version 3.6, 10 December 2025); class IIb by permanent listing only; medical benefit required (§ 139e Abs. 2 Satz 4 SGB V), shown by prospective comparative study (§ 11a DiGAV). The on-ramp that most developers plan around disappears precisely for the products the reform was meant to admit.
The evidence bar is not the only gate that rises with the risk class. A class IIb device requires a conformity assessment involving a notified body under the MDR, where many class I DiGA reach the market on self-certification, and the two, the notified body's and the BfArM's, do not run in parallel: the conformity assessment, CE marking included, has to be complete before the listing application is filed, and the BfArM allows no exception for a notified body without capacity, so the queue at the notified body sits in front of the three-month listing clock rather than beside it. Where the class IIb application is an AI-assisted diagnostic or monitoring tool, a third regime, the EU AI Act's high-risk tier, can apply on top, a convergence taken up separately in the analysis of AI-enabled medical devices. The reform admitted the higher-risk products; it did not shorten the path they travel.
A US developer that reads "Class IIb is eligible" as an invitation to bring a telemonitoring or AI-diagnostic product into the German market on the same provisional terms its competitors used has mis-scoped its evidence burden by a full controlled study, and with it its timeline and its capital plan. As of publication, the practical pipeline reflects the gap: the directory remained overwhelmingly populated by class I applications, and a class IIb listing had yet to materialize.10BfArM, DiGA-Verzeichnis (as of publication); no risk-class IIb DiGA had been listed, the directory remaining overwhelmingly class I. Whether that is a transitional lag or a structural deterrent is exactly the question the regulation does not answer.
4. Continuous Monitoring and the US Feedback Loop
Listing is not the end of the evidence obligation. Under the success-measurement regime it is the start of a continuous one. Manufacturers of permanently listed DiGA must collect and transmit performance data to the BfArM, which publishes results in the directory; recording begins with the third calendar quarter of 2026, the first transmission is due by 15 April 2027, and indication-specific patient-reported measures follow from 2028. For products that use the trial route, the period ends in a binary outcome: at re-evaluation the manufacturer either proves the positive Versorgungseffekt or leaves the directory. That outcome is not hypothetical. The GKV-Spitzenverband's own reporting records DiGA removed from reimbursement for failing to demonstrate a benefit, and conversions from provisional to permanent listing that arrived only with restrictions on the original scope.
The combination, a negotiated price with a mandatory success component and a regulatory duty to generate the data that measures it, reaches into documents US developers rarely draft for the German market. Revenue that is at least a fifth contingent on post-launch performance is not the fixed list price a US launch models; it behaves like variable consideration: estimated, constrained, and re-measured rather than booked at the invoiced price, and the estimate moves each time the success measurement reports. A finance function that modeled the German launch at list price has mis-stated not only the amount of the revenue but its pattern, and the correction tends to surface in exactly the periods when the outcome data arrive. The data-collection and outcome-attribution arrangements that feed the success measurement are not produced by US-standard commercial templates, and the allocation of who bears the loss when the measured outcome undershoots is rarely where a US agreement puts it.
The data the success measurement generates carries its own freight. What reaches the BfArM is anonymized and aggregated, but the continuous collection that produces it runs through the EU's regime for health information, where such data is a special category subject to heightened conditions, and that regime does not map onto the HIPAA-shaped posture a US developer keeps at home.11Regulation (EU) 2016/679 [2016] OJ L119/1 (GDPR), Art. 9 (processing of special categories of personal data, including data concerning health). A data architecture adequate for a US launch is not, by that fact, adequate for a German listing whose continuous measurement depends on processing the very category of data the GDPR most tightly controls.
This is where a German listing becomes a US problem. A success-contingent price feeds US revenue recognition and, where material, investor disclosure; a re-evaluation that strikes a product from the directory is a market-access event whose consequences do not stay in Germany. The commercial decision to enter, frequently taken on the strength of the regulatory clearance already in hand, is often made before anyone has modeled the price as variable or the listing as revocable. By the time the success measurement reports, those have stopped being German questions and become entries in a filing drafted to US standards.
5. Strategic Considerations
None of what follows resolves on the face of the statute. Whether an evidence package built to FDA expectations supports a positiver Versorgungseffekt in the German sense is one question; whether, for a class IIb product, it does so completely enough to carry a permanent listing without the trial period as a buffer is a harder one. Both turn on study-design choices made long before the German market was in view, against a standard the BfArM applies rather than the FDA.
Valuation is no more tractable. A product whose negotiated price is at least a fifth contingent on outcomes the counterparty helps define, measured continuously after launch, resists a US list-price model, and the difficulty compounds in any transaction, because the success component and the re-evaluation cliff sit directly on the cash flows a buyer is pricing. Whether to enter at all by direct listing, or instead to acquire an existing DiGA and inherit both its listing and its negotiated price, turns on facts the regulation never sees: the internal evidence, the capital plan, the deal pipeline.
Then there is the question of who, internally, owns any of this. The success measurement generates a continuous stream of patient-related data under a German regime, with implications for the data posture the company maintains elsewhere; the success-component price feeds financial statements drafted to US standards; and the entity that negotiates the Vergütungsbetrag, the one that runs the study, and the one that signs the US disclosures are rarely the same, and rarely working from a shared account of what has been promised. Whether those functions have been coordinated, or have each assumed another holds the problem, is the kind of question that surfaces after a re-evaluation rather than before.
Timing overlays all of it. The success-pricing rule binds from 2026 while the metric it measures is still being settled in the framework agreement, and the class IIb route is open in law while, as of publication, no product had yet traveled it. Whether to move into a standard that is unsettled, or to wait for the first listings and price agreements to show where it lands, is a wager on regulatory direction that the text of the reform does not resolve. These questions require analysis tailored to specific facts and commercial context.