Under the inpatient prospective payment system (IPPS), firms that own new and expensive technologies can apply for these technologies to receive new technology add-on payments (NTAPs.) The NTAP program was created in 2001 after Congress, concerned that the existing IPPS reimbursement structure did not incentivize the use of new technologies, passed a law[1] requiring the Centers for Medicare & Medicaid Services (CMS) to design a mechanism to recognize the cost of new medical services and technologies under the IPPS. NTAP applicants must prove that their technology is new, costly, and provides substantial clinical improvements. If the technology has been designated as a breakthrough device by the Food and Drug Administration (FDA), CMS assumes that this technology is new and provides substantial clinical improvements, and the applicant must only meet the cost criterion. All applicants must receive FDA approval by May 1 prior to the fiscal year for which they are applying for NTAP designation. Once a product has been designated for NTAP, hospitals utilizing this product receive an add-on payment outside of the standard IPPS Medicare Severity Diagnosis-Related Group (MS-DRG) payment amount. This add-on payment is equal to 65 percent of the difference between the cost of the case utilizing the technology and the standard MS-DRG payment, capped at 65 percent of the cost of the technology. NTAP designation cannot last more than three years.
As AI has become more prevalent in healthcare, and as the benefits of the technology become more apparent, policymakers have been forced to consider how these technologies can be reimbursed under Medicare. As no dedicated reimbursement mechanism exists for AI in Medicare, NTAP has become the default way to reimburse for these technologies under the IPPS. However, the uptake of AI in healthcare has far exceeded the rate at which these products are approved for NTAP reimbursement. In 2020, Viz ContaCT (Viz LVO), a software that helps with stroke triage, became the first AI software to be reimbursed under NTAP. Only five additional software products have received NTAP designation since. In the FY 2025 IPPS proposed rule, two AI technologies submitted NTAP applications: Quicktome Software Suite, a cloud-based software that uses data points from magnetic resonance imaging (MRI) scans to quickly and accurately map a patient’s brain network, and Annalise Enterprise CTB Triage – Obstructive Hydrocephalus (OH), which helps radiologists triage images by flagging brain scans with features suggestive of OH.
Regardless of whether these products are approved, CMS will need to consider a better way to reimburse hospitals for their use of AI software. NTAP alone does not provide significant incentives for hospitals to use AI. At maximum, the add-on payment covers 65 percent of the cost of the new technology, and the technology cannot be designated for more than three years. Medicare beneficiaries deserve access to the healthcare benefits provided by AI technology, and such access will be difficult to ensure without a dedicated, permanent reimbursement mechanism for AI. It is highly unlikely that such a mechanism will be enacted under any of the major payment rules this year, as attempts to regulate AI are still early in development. The Department of Health and Human Services (HHS) AI taskforce that the Biden Administration tasked with creating a comprehensive plan on responsible deployment of AI in the health and human services sector is not required to release this plan until January 2024. The Medicare Payment Advisory Commission (MedPAC) just began preliminary discussions on this topic in their November 2023 session, and the Commission acknowledged that they are far from establishing any recommendations.
The surge in telehealth use following flexibilities granted by CMS during the Covid-19 pandemic demonstrates the power that CMS has to increase access via technology friendly policies. Prior to the pandemic, telehealth use accounted for less than one percent of physician encounters[2]. Now, it is a critical aspect of the healthcare system, particularly for behavioral health. The technology necessary for widespread telehealth adoption was present long before the pandemic, but it took a supportive policy framework for this adoption to materialize.
Health systems face workforce shortages, lasting burnout from the pandemic, an aging population, and inflation. Responsibly used AI offers potential to mitigate these issues and provide savings for the Medicare program. However, this potential will not be realized until the correct policies and reimbursement are put in place.
[1] Section 533 of the Medicare, Medicaid, and SCHIP [State Children’s Health Insurance Program] Benefits Improvement and Protection Act of 2000