By Brendan Ryu
It is essential to consider the ethical implications of payment models for AI radiology services, especially when startups, hospital systems, radiology firms are saving resources and trying to financially support early innovation. In this blog post, we will explore the ethical issues surrounding this topic, with a particular focus on justice and rights to AI access drawing upon insights from an interview with Dr. Eric Keller, Senior IR Resident at Stanford, and an expert in medical ethics.
The Notion of Justice in Medical Ethics
Justice, as one of the pillars of medical ethics, encompasses the fair distribution of health care resources and the equitable treatment of patients. It ensures that all individuals, regardless of their socioeconomic status, have access to the appropriate medical services. However, the introduction of AI within radiology raises questions about whether medical integrity is being upheld when patients are required to pay out-of-pocket for these services.
Keller states, “Given the current state of evidence, or lack thereof, is AI radiology something we consider a basic health care right? Should everyone have access to it? Currently, it feels more like a luxury to me, as it is not yet the standard of care. We still don’t fully understand its benefits and limitations. Some studies even suggest that AI may introduce biases and confirmation biases, leading to errors.”
This insight highlights the uncertainty surrounding AI radiology’s benefit and the potential downsides. While the technology holds great potential, it has yet to be integrated in societal guidelines and community standard of care. This raises concerns about the ethical implications of having patients pay for a technology that is not yet considered a component of the community standard of care.
Addressing Access and Justice
The issue at hand is whether AI radiology is a basic right to which everyone should have access or if it remains a “luxury elective service.” Keller pointed out two concerns that arise from the current situation.
The Ick Factor: Keller expressed discomfort with the direct-to-consumer sales approach. This approach may exploit patients’ fears of cancer and give rise to a false urge to make the extra payment. While this kind of strategy is already prevalent in health care, it highlights the ethical concerns surrounding a payment model that ultimately relies on the patient’s own decision-making.
Segregation and Inequity: Keller also raised concerns about the potential for segregating access to improved outcomes based on a patients’ ability to pay. If AI truly improves diagnostic accuracy and patient outcomes, those who cannot afford the technology might experience delays in detecting conditions like breast cancer with the technology further exacerbating existing health care disparities.
In conclusion, the ethical implications of the direct-to-patient approach for AI radiology warrant careful consideration. We must reflect on whether cutting-edge medical technologies should be considered basic rights or luxury services. The uncertainties surrounding their benefits and potential biases raise concerns about payment models and equitable access. Striking a balance between innovation support and fairness is essential, emphasizing justice and patient rights in our evolving health care landscape.
Challenges Associated with Getting New Technology in the Hands of Patients
One of the largest American radiology firms, RadNet, recently launched an Enhanced Breast Cancer Detection (EBCD) service that offers an FDA-cleared artificial intelligence (AI) technology with patients’ annual breast screening regimen. Clinical studies and real-world evidence (internal) support the technology’s ability to aid radiologists in providing an accurate mammogram report and improving overall cancer detection. But it doesn’t come free. RadNet faces the same conundrum most innovators face in the current U.S. health care market: how to get the technology reimbursed. RadNet has taken a bold move in inviting patients to pay an out-of-pocket fee for the service, at least for now. Some may question the appropriateness of this approach, particularly in terms of health equity. While others argue that without a direct-to-consumer approach, no patients would benefit from the innovation. Either way, this is a huge challenge in our health care system, and there is a strong need to provide a bridge from innovation to widespread patient access.
Out-of-Pocket Payment Model is a Necessary and Crucial Step
RadNet’s self-pay model for EBCD is similar to the approach taken with respect to the tomosynthesis model, which first hit the market with a $50 out-of-pocket fee before proving clinical benefit and being covered by Medicare. RadNet acknowledges the importance of balancing affordability with patients’ willingness to pay in the early stages. Therefore, RadNet conducted numerous focus group discussions early in their pilot to better understand patient response to the program and the impact of their out-of-pocket model, and received a generally positive response. Some participants expressed willingness to pay significantly more than the set charges to obtain the benefits of the AI.
In order to demonstrate the clinical benefits of the EBCD program, RadNet is actively conducting real-world evidence studies to prove the benefit and to assess the market response to the program. They are optimistic that patients will advocate for insurance coverage and CMS will begin looking into reimbursement for AI algorithm use, highlighting the potential benefits and long-term cost savings associated with early cancer detection. RadNet sees the self-pay model as a bridge to a longer-term solution. Most AI startups face a significant financial uphill climb and depend on many rounds of financing by patient investors with the hope of someday recovering the years of losses while bringing the lifesaving technologies to the masses. RadNet is invested in the long-run, and wants to ensure access to its technology for everyone for years to come. Its AI division is still operating in the red, but with the EBCD program there is a foundation to sustain continued progress until reimbursement is approved.
Striking a Balance
RadNet’s challenges with the EBCD program serve to highlight the ongoing problem innovators face bringing new health care products and services into widespread clinical practices so that all patients may benefit. Some potential solutions include:
- Evidence-Based Guidelines: Professional societies of radiology and medicine should collaborate to review the evidence on AI radiology, its clinical benefits and limitations. Moreover, AI radiology vendors promoting their product should conduct rigorous studies, so that societies can review the clinical data and help optimize the integration of AI radiology into those guidelines and promote reimbursement from CMS.
- Increased Government Funding: Government funding enables research to evaluate the effectiveness and impact of AI radiology, leading to the acceleration of evidence-based guidelines. Simultaneously, CMS exploring reimbursement models should ensure that AI radiology services can reduce the health care economic burden and provide value in the long run. These initiatives are not only driven by the goal of promoting access but also by the potential for cost savings and improved health outcomes. From a health economics standpoint, reimbursement for AI radiology services can be justified based on the value it brings to the health care system. For example, if AI radiology demonstrates the ability to reduce unnecessary scans, tests, and procedures, value is created by lowering the overall health care cost for the patient. Therefore, by investing in AI radiology and focusing on reimbursement pathways, the government recognizes the potential for greater clinical value, ensuring a sustainable health care system that benefits both patients and the health care system as a whole.
- Emphasis on Informed Consent: Informed consent is vital when patients are required to pay out-of-pocket for AI radiology services. Clear and unbiased communication is essential to ensure patients understand the benefits, limitations, and potential risks. Healthcare providers should engage in transparent discussions, explaining the added value of AI radiology and its implications. Emphasizing informed consent promotes patient autonomy and ethical decision-making.
As the field of AI in medicine continues to evolve, it is essential that payment models follow suit. Therefore, collaboration among stakeholders, payers, and regulatory bodies is crucial. By fostering ongoing discussions, exploring pricing models, and conducting studies to prove clinical benefit, the aim is to establish a sustainable and ethically sound landscape for AI radiology. RadNet is an example of how one leading AI business is trying to succeed in the face of these innovation challenges. The EBCD program serves as a catalyst for further discussion. It is a bold move with challenges, and it is yet to be determined if it will be successful driving new advancements in AI technology and promoting equitable access to enhanced healthcare services, but it provides a new potential path for future AI innovators to consider with the big financial mountains they must climb.
– Brendan Ryu, is a 2024 M.D. candidate at the Zucker School of Medicine at Hofstra/Northwell. He is on LinkedIn.
