The past weeks have been extremely exciting for AI in general and for us at Aidoc specifically (and for me personally).
After 2 years of commercial availability, CMS have finally approved the first reimbursement opportunity for AI – specifically for computer-assisted triage and notification for patients with suspicion of stroke .
This has created A LOT of interest in the market and we have been diligently communicating this news to our customers and partners. It is exciting to see how many of them have already introduced us to their finance/billing peers to start working on the coding/billing process.
I joined RSNA’s TweetChat #radAIChat on the topic of “who pays for AI” three weeks ago – just prior to the CMS announcement – and the discussion around “who pays for AI” was definitely a heated one. The main driver for CMS coding – my esteemed panel peers and I agreed – was value. Can an AI solution show measurable value within a clinical setting? How do you define value and who benefits from the use of the solution? The hospital financially upon reimbursement? The radiologist from an always-on AI safety net or , and maybe most importantly, the patient – from expedited and improved treatment paths.
I would like to share 2 recent data points – one was presented at this year’s ASNR and the other at RSNA 2019 – both show value; the first to work with radiologists to expedite formal diagnosis and improve workflow and the second as a tool that expedites treatment and reduces length of stay (in the ED by an hour).
It’s empowering to see that the CMS now recognizes this value and I look forward to additional reimbursement in the future as AI based triage becomes standard of care for additional pathologies (e.g. PE, ICH, Intra Abdominal free air).
[October 12th, 2020]
Over the last week as we continue to introduce this to our customers and sites a frequent topic that comes up is the applicability of this code to other vendor solutions in this space.
Let us explain this further; according to CMS policy, NTAP (New Technology Add-On Payment) status applies not only to the technology of the applicant but also to other technologies that are “substantially similar” to the product that received the NTAP status. This means that all solutions which are a radiological computer-assisted triage and notification software system that analyzes CTA images of the brain acquired in the acute setting, that sends notifications when a suspected large vessel occlusion (LVO) has been identified, and recommends review of those images are covered by the new code which qualifies for reimbursement.
There are three criteria which determine whether the technology is “substantially similar”. The first is whether the technology is described as having the mechanism of action to achieve a therapeutic outcome, i.e. the identification of LVO and shorten time to notification. The second criterion for substantial similarity is whether the technology is assigned to the same or a different MS-DRG, i.e. Medicare Severity Related Groups. And lastly, is whether the technology involves the treatment of the same or similar type of disease and the same or similar patient population. When we read that the US is below global average in health tech adoption, it’s encouraging that NTAP is utilized for AI and I am confident that we are taking a step in the right direction to increase both the adoption and the value of AI in clinical settings. I will be sharing some of our deployment stories at the ACR-DSI meeting next week and will talk more on this topic following this event.
[October 20th, 2020]
Another frequent topic of interest is that of payment amount and more specifically, payment amount per patient. As per CMS reimbursement protocol for NTAP calculation is as follows, NTAP will be paid if the NTAP amount is equal to the lesser of (i) 65% of which the total amount of the case exceeds the MS-DRG payment or (2) 65% of the cost of the new technology. To illustrate this, I will give three examples of the NTAP policy for cases involving a technology with an estimated cost of $1,600 in an MS-DRG that reimburses $30,000.
Example where no NTAP is Paid
Assume that the charges for a particular case are $20,000. After applying the hospital’s cost-to-charge ratio (example: 0.5), the cost of the case is estimated at $10,000. Since the cost ($10,000) is less than the MS-DRG payment ($30,000), no NTAP is provided.
Example where the maximum NTAP amount is paid
Assume that the cost of the case is estimated to be $50,000. Since the cost of the case is greater than the MS-DRG payment ($30,000), an NTAP is made. To determine the amount of the NTAP, compare 65 percent of the excess costs not covered by the MS-DRG payment ($13,000) to 65 percent of the cost of the new technology ($1,040). Since 65 percent of the cost of the new technology is less than 65 percent of the excess costs, the NTAP amount would equal $1,040. Therefore, the total payment to the hospital would equal $31,040.
Example where the NTAP equals 65% of the excess cost not covered by the MS-DRG Assume that the total cost of the case is $31,000. The NTAP amount would be equal to 65 percent of the excess cost not covered by the MS-DRG payment ($650) since it is less than 65 percent of the new technology cost ($1,040). Therefore, the total payment to the hospital would equal $30,650.
In the case of the reimbursement code granted to a computer-assisted triage and notification for patients with suspicion of stroke, CMS determined the technology at a value $1,600 per patient which means potentially a $1,040 reimbursement for each qualified use of such technology, including Aidoc’s LVO AI and this can be well over $100,000 per year for a single stroke center.
[October 28th, 2020]
The current code only applies to solutions enabling computer-assisted triage and notification for patients with suspicion of stroke. This applies only to Medicare inpatients whose CTA was analyzed by such solution, in Aidoc’s case its solution for LVO triage and prioritization. Now that the CMS has granted one code for AI, the big question is what is now to follow? Aidoc offers the most comprehensive AI solution in the market and will be actively working on extending the CMS code to our additional triage and notification solutions.We see this just as the beginning of the road. Indeed, this decision could open the door to gain reimbursement for other potential applications for radiology AI technology.