Tackling the Economics of Artificial Intelligence
A key ingredient in moving Artificial Intelligence (AI) algorithms into routine clinical practice will be ensuring our healthcare system supports fair compensation for their development. However, figuring out how that will happen may not be as simple as it might seem. With the ongoing escalation of costs in the U.S. healthcare system, the medical community will have to demonstrate the value and cost savings that each AI algorithm brings to our patients and our delivery system before reimbursement from payers will be considered:
- The value to patients may be earlier and more accurate diagnoses and treatments.
- The value to physicians may be improved efficiency in data management and integration.
- The value to our health system may be improved quality of care, overall efficiency, and decreased length of stay.
An understanding of the alphabet soup of current and future payment models will be needed — and it must begin with the current fee-for-service (FFS) model. In this system, specific medical services, procedures, and supplies are reimbursed using the CMS Healthcare Common Procedure Coding System (HCPCS)1. Level I of the HCPCS system is based on Current Procedural TerminologyTM (CPT), which is a numeric coding system developed and maintained by the American Medical Association.The CPT system identifies and describes medical services and procedures commonly furnished and billed by physicians and other healthcare professionals. However, CPT does not include the codes needed to separately report medical items or services for patients that are provided outside of the physician office setting (e.g., durable medical equipment and supplies). The Level II HCPCS was established to provide codes for non-physician providers to submit claims for these items to Medicare and private health insurance programs.
Each HCPCS code is assigned a value by Medicare and other payers, and claims are submitted by providers based on these codes. When medical equipment and supplies are used in the physician office setting, the reimbursement for these items is included in the CPT code payment to the physician as “Direct Practice Expense.” However, when the same services are performed by physicians in the hospital or a site other than a physician office, the payments for equipment and supplies are made directly to the facility. As such, each CPT code in the Medicare Physician Fee Schedule (PFS) has different payments to physicians based on whether the service was provided in a physician’s office (non-facility) or hospital (facility) setting2.
Finally, a portion of the payment for each physician service (“Indirect Practice Expense”) is designed to cover the costs of operating a practice, including office rent, utilities, computers, and billing costs. The Medicare Physician Fee Schedule uses the Resource-based relative value scale (RBRVS) to assign relative value units (RVUs) for each physician service, so all of the practice expenses are then converted to RVUs. RVUs for physician work and compensation for professional liability insurance are added to the direct and indirect practice expense RVUs to comprise the total RVUs for each physician service in the Medicare PFS, which is then multiplied by a conversion factor set by CMS to give the dollar payment to physicians.
Hospitals are reimbursed under two separate payment systems: the Inpatient Prospective Payment System (IPPS), which uses Diagnosis Related Groups (DRG) as its fundamental reimbursement system; and the Hospital Outpatient Prospective Payment System (HOPPS), which uses Ambulatory Payment Classification (APC) as its fundamental reimbursement system. Each of these systems accounts for the payments for medical equipment, devices, and supplies in different ways. And, while some private payers base their payment systems on Medicare, each private insurer has its own way of assigning reimbursement for medical equipment, devices, and supplies to each service.
If you think this is complicated, you are right (and we did not even discuss Category III CPT codes, which are temporary codes designed to track new procedures and emerging technology that may or may not be reimbursed). The system is made even more complicated because clearly there will not be a one-size-fits-all payment scheme for reimbursing the use of AI in healthcare. Some algorithms will affect payments to physicians, perhaps making our work either more efficient or more time-consuming as we bring more patient information into our care of complex patients. Some algorithms will improve the overall quality and efficiency of our practices and health systems but cannot be attributed or assigned to a specific service or procedure. While some algorithms may be directly reimbursable by third party payers, many won’t be. Finally, all algorithms that are adopted by physicians and our health systems must be proven to provide demonstrable value to patients in a safe and bias-free environment.
The CMS Quality Payment Program (QPP)3 is the next step in the development and adoption of alternate payment models in U.S. healthcare. The QPP includes the Merit-based Incentive Payment System (MIPS) and Advanced Alternative Payment Models (APMs). MIPS uses four categories – quality, improvement activities, cost, and advancing care information – to adjust Medicare FFS payments to physicians – up or down by as much as 9 percent in 2022 – based on their performance in each category. Measures for quality, improvement activities, and advancing care information are reported to CMS by physicians. If certain AI algorithms have documented value and have improved quality to our patients, use of these could potentially at some point be included as MIPS measures. While APMs are much less prevalent in the U.S., algorithms that increase overall efficiency for health systems will be welcome as the medical community strives to do more for our patients at a lower overall cost. When using APMs, assigning and attributing a per-unit cost for an AI algorithm to an individual CPT code will be much less important than ensuring that the algorithm functions in a way that augments the care provided to patients without taking away the common-sense decisions of physicians and patients.
Finally, the economics of AI in healthcare will have to include a discussion about potential disparities if AI is available to some patients and not available to others. While I fully expect to see market leaders emerge touting that their services include the latest AI innovations, we do not want to see our healthcare system devolve into a two-tier system where some can afford AI and others cannot. I believe our reimbursement system has a duty to protect our patients by making sure all physicians have access to these potentially revolutionary tools.
ACR has always been strategically involved in the federal regulatory and payment policy issues around the radiological sciences. Reimbursement for AI will be no exception. In order to provide education about regulatory payment policy issues around AI, the ACR Data Science Institute (ACR DSI) will be hosting the ACR DSI and SIIM Spring 2018 Data Science Summit: The Economics of AI in Health Care on May 30 in Washington, D.C., in conjunction with the Society for Imaging Informatics in Medicine (SIIM). Over the course of the day, the Spring 2018 Data Science Summit will provide developers, administrators, and physicians an in-depth look at all of the economic issues and challenges facing the healthcare community in moving AI algorithms into clinical practice. I encourage you to register today and look forward to seeing you there.
By Bibb Allen, Jr., Chief Medical Officer, ACR Data Science Institute