This is part two of a three-part series on what the breakthrough device coverage problem looks like from inside CMS, in conversation with Dr. Lee Fleisher of Rubrum Advising. Read Part 1 here.
This is part 2 of a three-part series on what the breakthrough device coverage problem looks like from inside CMS, in conversation with Dr. Lee Fleisher of Rubrum Advising. Read Part 1 here.
The first article in this series started with meeting Lee Fleisher at MedTech Malta. This one starts with another conversation, this time at the LSI Conference a couple of months ago. Lee and I were catching up at the conference and ended up talking with another former CMS official (I’ll leave unnamed) who had a direct hand in shaping the Transitional Coverage for Emerging Technologies (TCET) program.
I learned two fascinating things during that conversation:
Turns out TCET wasn’t designed for all BDD devices, only select PMA devices
Man, all those CMS guys seem to have the same mindset
I have written about TCET before, so I won’t say more than the fact that it was created as a replacement to MCIT, another program designed to help Breakthrough Device Designation (BDD) medical devices get to CMS coverage sooner than the average 5.7 years.
That conversation, combined with what we now know about how the program has actually performed since its launch, made me think differently about a question I have been chewing on since I wrote my original article. What was TCET actually built for?
So, back to this LSI conversation. Our mysterious second CMS man, who is awesome, just not someone I specifically got approval to quote, told Lee and me that TCET was scoped and implemented by CMS specifically for Pre-Market Approval (PMA) class devices by design. Both Lee and I were surprised by this information, which had me secondarily surprised that this was news to Lee!
Turns out Lee was at CMS during publication of the TCET pathway and knew that resources would be limited to five or so approvals a year, but the implementation of TCET was after Lee’s departure from CMS. Evidently the decision was made after his time that - because the CMS coverage group did not have the staffing to review the full breakthrough device pipeline coming out of the Food and Drug Administration (FDA) - TCET would prioritize PMAs.
Before I get to why I think that prioritization choice deserves a closer look, the staffing piece is worth discussing for a second. Lee walked me through some of this in our recent interview:
The entire CMS coverage group has 35 to 37 people, in his estimate, possibly fewer now after departures earlier in the year. They are responsible for novel coverage decisions across a Medicare program that handles close to a trillion dollars in spending annually. As Director of CCSQ, Lee oversaw a $1.7 billion budget but couldn’t just shift $10 million from one program into the coverage group. The structure of CMS appropriations does not permit that kind of internal reallocation. Expanding the coverage group requires an act of Congress.
So, given that constraint, CMS decided they could not have reviewed every breakthrough designation coming through the FDA with the team they had. They had to prioritize.
I do not know exactly how the prioritization decision got made internally, and I am not going to pretend to. What I do think is worth asking is whether the prioritization that resulted actually tracks with the stated purpose of the program? Or if those constraints were even a real issue? OR, if they were supposed to give coverage to BDDs, then what review is even needed? (Well, that argument could have been made during MCIT but TCET now existed in a post-Trump executive order world where they no longer had that requirement.)
When I wrote my original article on how CMS is killing breakthrough device innovation, I was not focused on PMAs. I did not call that out explicitly in the piece, but in my head, the segment of the breakthrough pipeline I was worried about were the De Novos and 510(k) products.
Here is my reasoning: PMAs are already devices that take significant investment, and the investors who fund PMA-class products are comfortable with the long pathway to coverage because there is real return on investment at the end of it. Most PMAs are implanted devices used during expensive surgeries, which is a generalization but a useful one. Some PMAs are diagnostics, but most companies operating in the diagnostic space end up trying to position their claims as decision support, often pursuing 510(k) or de novo classification instead, because the PMA pathway is too costly and time-consuming to justify for the return. PMAs only really makes economic sense for implanted devices that can be used during expensive surgeries with reimbursement that supports the development cost.
So, when I think about which devices are actually struggling under the current Medicare coverage framework, PMAs are not the segment that comes to mind. The original Trump executive order that kicked all of this off was not trying to fix coverage for PMAs. It was trying to fix coverage for all the other devices that are struggling, the ones where breakthrough designation does not translate into a coverage pathway because the reimbursement amounts do not justify the trial costs and economics do not justify the investment.
I think most of those struggling devices fall in the De Novo bucket. A lot of them may be 510(k) as well. Honestly, I cannot prove this out cleanly because FDA does not publish a breakdown of all 1,246 breakthrough designations by intended pathway.1 The market-authorized breakthrough devices come out roughly evenly split (32 510(k), 30 De Novo, and 31 PMA through end of 20232), but that could reflect the real distribution of designations, or it could reflect that De Novos without existing pathways are the ones really feeling the crunch and not getting enough capital to reach market in the first place. What we do know is that the denominator we cannot see is full of designations that died before they got anywhere.
So, when I hear that TCET focused on PMAs, which to me is the lowest-hanging fruit, I am confused and frustrated. The PMAs already have investors who understand the pathway. The De Novos do not.
There is also a question I want to raise here, which I am not going to be able to answer. Why do any of these programs, MCIT or TCET, require so much effort? Aren’t they just meant to establish codes and coverage without reviewing the evidence until later? They’ve already proven safe and effective to the FDA and somehow demonstrated their ability to provide better outcomes than the current standard of care. The benefit category question is a yes-or-no statutory test. Why does CMS need to layer this much process on top of an FDA designation that was specifically intended to fast-track promising technology?
I do not have a clean answer. What I do know is that the low-risk, high-reward devices would probably be cheaper for CMS to cover than the implanted PMA devices the agency is prioritizing.
And here’s one of those tidbits I love to drop in: Lee told me in our interview that, “CMS, believe it or not, I can attest, does not take cost into consideration when making a coverage decision. They really just use the evidence of clinical benefit.”
That was news to me.
Most people assume the agency is at least partly weighing what something costs Medicare. It isn’t, by design. Which leaves me wondering, frankly, if cost were on the table, would the De Novo and Class II breakthroughs fare a little better in the prioritization? They would be cheaper to cover than the implanted PMA devices CMS keeps prioritizing. But that math does not live inside the coverage decision.
The scoping decision was one half of the conversation at LSI. The other half was about whether the breakthrough device coverage push is justified at all.
The argument from Lee goes something like this: The evidence generated by breakthrough-designated devices has not matched the clinical performance FDA’s designation criteria implied. Therefore the devices have not been delivering on what the BDD designation promised. From CMS’s vantage point, that suggests the agency was right to be cautious about extending coverage to a broader set of breakthrough devices on the strength of designation alone.
I hear the same data and see a different problem. It is a survivorship bias problem.
As of late 2024, FDA had granted 1,041 breakthrough device designations and 128 of them had reached market authorization.3 That is a 12.3% conversion rate. Any data the analysis is built on, the real-world performance of breakthrough devices, can only be drawn from those 128. The other 87.7 percent are absent from the analysis entirely. Those are the breakthrough designations that died somewhere between designation and market, and a meaningful share of them died because they could not raise the capital to fund the pivotal trials that would have generated the data we now wish we had.
There is also a circularity to it. The coverage support that would have funded the trials is the same support being withheld because the trials don’t exist. You cannot fault a cohort for not generating evidence when you have structured the system to prevent the evidence from ever being generated.
(Yeah my head is spinning a bit too after that.)
Lee coauthored a paper that makes the published version of this argument, so I went and read it.
The paper is “FDA Breakthrough Device Designation: Clinical Evidence And Medicare Payment Policies,” published in Health Affairs Forefront on December 13, 2024.4 The authors reviewed the 81 breakthrough-designated devices that had received FDA marketing authorization as of July 2023. They then narrowed to the 26 cleared through the 510(k) pathway, and narrowed again to the 16 of those that were therapeutic in modality. Within that group of 16, they found that 6 had no clinical studies referenced in the FDA decision letter and only 4 included data from randomized controlled trials. They concluded that most 510(k) breakthrough devices do not demonstrate, or even attempt to demonstrate, robust evidence of clinical utility, and that this should disqualify them from automatic coverage or alternative reimbursement programs like NTAP and TPT.
I have three problems with this analysis (respectfully).
The slicing and dicing introduces bias. Starting from 26 510(k) devices and ending up with a clinical evidence finding on 16 of them is a significant narrowing. I get the argument for focusing on therapeutic devices only since they have more direct impact on patient outcomes than diagnostics, but was it cherry picking to exclude the diagnostic devices in the analysis?
Ultimately, the fact that this conclusion is based on 6 out of 839 BDD devices from that time frame makes it a bit hard to think it has any statistical significance.
The evidence reviewed was the 510(k) submission, not the BDD application. The paper draws its conclusion about clinical evidence from what was referenced in the FDA decision letter for the 510(k) clearance. The problem is that a decent number of 510(k) product codes do not require clinical evidence to get your 510(k) - so OF COURSE the summaries don’t include any clinical data in them. That doesn’t mean the companies have zero clinical data and it doesn’t mean they never presented the FDA with clinical data ever. In fact, to get BDD you have to demonstrate with some level of clinical evidence that your product can provide better outcomes than the current standard of care - so they must have pointed to something during that process. Its just there’s no requirement for that to be disclosed.
This is exactly what these programs were designed for. The whole reason this breakthrough device coverage push exists is that 510(k) and De Novo breakthrough devices don’t have the funding to generate CMS-level clinical evidence. The whole point of giving these devices coverage support is to get them into the real world where they can actually generate the evidence CMS wants to see, in the patient population CMS actually serves.
Faulting these devices for not having the right level of pre-clearance clinical evidence to support CMS reimbursement decisions and then using that finding to argue against the programs designed to help them due to the burden of building out that level of evidence is the same circular argument we keep hitting.
By my count, one device has been selected and made it through TCET to national coverage since the program launched in August 2024. The cap is supposed to be five per year so, in over roughly two years of operation, we should see many more companies that have made it through. CMS set its own ceiling and still missed it by roughly 90 percent which makes me think this staffing argument might not be the issue.
It is also worth zooming out on the timeline. President Trump’s Executive Order 13890, the order that originally directed CMS to create a breakthrough device coverage pathway, was signed on October 3, 2019.6 I’ll spare you all of the details on legislative and bureaucratic moves, but that is six and a half years between an executive order telling CMS to build a breakthrough device coverage pathway and the agency moving one device through the program it built. At some point a pattern of agencies missing their own targets stops looking like capacity constraints and starts looking like an agency that does not want to do the thing it keeps announcing it is going to do. The recent announcement that they will be ending TCET seems in line with that logic.
Maybe RAPID is going to be different. The structural cooperation between FDA and CMS that RAPID contemplates is something the prior programs did not have, and it could turn out to be the version that actually moves devices. More on that next week.
Continued in Part 3: RAPID and What CMS Is Building Next.
Blythe Karow is a strategic management consultant and founder of The Karow Advisory Group. She writes The Device Files, a Substack publication on MedTech strategy, market access, and commercialization.
FDA, “Breakthrough Devices Program,” designations granted as of December 31, 2025. https://www.fda.gov/medical-devices/how-study-and-market-your-device/breakthrough-devices-program ↩
Medical Device Academy, Breakthrough Device Designation tracker, data through December 31, 2023. https://medicaldeviceacademy.com/breakthrough-device-designation/ ↩
Prasad NK, Fleisher MA, Karamardian M, Fleisher LA. “FDA Breakthrough Device Designation: Clinical Evidence And Medicare Payment Policies.” Health Affairs Forefront, December 13, 2024. DOI: 10.1377/forefront.20241211.800115. Funded by a grant from Arnold Ventures to Rubrum Advising, LLC. ↩
Gupte T, Nitave T, Gobburu J. “Regulatory landscape of accelerated approval pathways for medical devices in the United States and the European Union.” Frontiers in Medical Technology, May 2025. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12119601/ ↩
Executive Order 13890, “Protecting and Improving Medicare for Our Nation’s Seniors,” October 3, 2019. https://trumpwhitehouse.archives.gov/presidential-actions/executive-order-protecting-improving-medicare-nations-seniors/ ↩
No posts
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.