Access and transparency, AI challenges, women’s health — what directions are best for primary care?

© CrazyCloud - stock.adobe.com

Radical policy and financial change is needed to refocus the American health care system to emphasize primary care.

Meanwhile, artificial intelligence (AI) holds promise for treating patients, but there are still a number of challenges to overcome to ensure it proves beneficial to patients. And for women patients, their total health, not just reproductive care, must become a priority for primary care in years to come.

Three analysts shared their viewpoints as part of the discussion in this month’s meeting of the Standing Committee on Primary Care of the National Academies of Engineering, Science and Medicine. All were coauthors of recent articles published in Health Affairs with their prescriptions for pain points within the health care system, and the three panelists summarized those points of view.

Speakers were Donald Berwick, MD, MPP, president emeritus and senior fellow of the Institute for Healthcare Improvement; Brian Anderson, MD, CEO of the Coalition for Health AI; and Claire Brindis, DrPH, distinguished professor at the Philip R. Lee Institute for Health Policy Studies at University of California San Francisco.

‘The buck stops nowhere’

Donald Berwick, MD, MPP
_{© Institute for Healthcare Improvement}

“Health care in the United States is among the most technologically advanced in the world, but it is largely failing to meet the needs of the nation,” said the article, “From Laggard To Leader: Why Health Care In The United States Is Failing, And How To Fix It.”

“The American health care system is failing,” and it is really important for leaders from all sectors in the system to understand, said Berwick, a coauthor of the article.

The writers do not believe incremental change is going to help a country at the bottom of developed nations for health care performance, but at the top for cost, Berwick said. Instead, the United States needs a set of universally embraced aims for healthcare, starting with universal access, the elimination of medical debt, adding a decade of healthy life to the populace, and closing racial and economic gaps, he said.

The authors outlined eight principles that should be minimum commitments, Berwick said. “None of this is easy,” he added.

• “The first is that everyone should have access to care. It’s time for this country to have universal coverage. End of story. Period,” Berwick said. “No apologies for that, and anybody being left out of coverage is a failure for the country.” Embedded in that is elimination of medical debt at least for people with income at least 400% of the federal poverty level. There should be a Medicare plus a public option to buy into Medicare, including for people with employer-sponsored coverage. Medicaid payment should be brought to the level of Medicare.
• Integrated primary care is a flagship principle, with team-based care and teams accountable and rewarded for progress toward the aims.
• Fee-for-service payment should be eliminated in favor of population-based payment, with standardized payment rebalanced toward primary care and away from specialization.
• Money saved should be reallocated toward social determinants of health. That will mean some money leaves the health care system to be in the hands of communities or other entities.
• Financing, which is maddeningly complex and opaque, must be simplified to minimize administrative complexity and enhance efficiency.
• Health care needs transparency with a new level of organizational professionalism. Physicians take an oath to put patients first, and leaders who govern and finance health care organizations should embrace those ethics, with consequences for violating them.
• Health care organizations must become learning organizations, with data-based learning.
• The buck has to stop somewhere. “This is very difficult,” but there must be more clarification of a supervisory agent, such as a congressionally created health system accountability board that reports to the nation about progress on the aims.

“Right now, the buck stops nowhere,” Berwick said.

‘We haven’t figured this out yet’

Brian Anderson, MD
_{© Coalition for Health AI}

AI models have a life cycle, starting with development, then deployment, then monitoring it and managing it over time, said Anderson, coauthor of “Artificial Intelligence In Health And Health Care: Priorities For Action.”

That last phase, with monitoring and management, is going to pose an enormous challenged for health systems, he said. It will involve understanding how to monitor AI for degradation and drift, but frontline clinicians in primary care may not have the infrastructure, subject matter expertise, staff or resources to do that.

The authors noted the importance of investing at the federal level to create a policy that supports lower-resource health system to monitor performance so AI performance degradation does not put patient lives at risk, Anderson said.

In the deployment phase, health care needs best practices and frameworks for “implementation science” to configure AI in the complex information technology of a health system or primary care clinic, and to mitigate the risks. Physicians may ignore an annoying pop-up in the electronic health record, so an AI system must overcome those cognitive biases of doctors and nurses, Anderson said.

“And that is something that, candidly, we haven’t figured this out yet,” he said. Physicians and nurses need education to think critically about AI and understand things like unjustified bias.

“I would assert that one of the most important questions that any provider in this space should ask is, you know, is this AI tool appropriate for the patient that I have in front of me?” Anderson said. There is a concept of transparency or explainability to help physicians and nurses understand how an AI model was made and what data sets were used to train it. If the data sets were not representative of a patient in a primary care setting, the AI performance may be different from what a vendor stated when the AI mode was procured and deployed, Anderson said.

The authors made a policy recommendation to incentivize and create regulations so physicians and other clinicians have that information. Physicians also are legally responsible for the tools they use on their patients, so they must have information to decide whether to use AI as a tool on patients. They don’t have that now, Anderson said.

Because AI is an evolving part of health care, it needs a fundamentally different approach to oversight, favoring dynamism and learning over static rules and requirements. There are amazing stories of health care providers using AI to gain time with patients, summarize complex medical records or make diagnostic decisions. Yet, physicians and other clinicians don’t have all the answers on how to measure bias or manage performance across multiple different use cases for generative AI, Anderson said.

“Yet, we have these really moving testimonies and anecdotes and stories, and growing body of evidence, that these generative AI tools are really meaningfully helping doctors,” Anderson said. “And this juxtaposition of not knowing exactly what we're getting into and how to measure accuracy and performance or bias in the generative AI space, and not wanting to stifle innovation with unnecessary regulatory burden, is a real challenge. And the point of the passage is, we need a level of flexibility within the private sector and within the public sector, within the regulatory space, to figure this out together, in essence, because we haven't done it yet. And yet, we're deploying these already in, you know, somewhat consequential use cases.”

‘Prioritize investments in women’s health’

Claire Brindis, DrPH
_{© Philip R. Lee Institute for Health Policy Studies at University of California San Francisco}

Investing in women’s health would add quality days to women’s lives and spark economic growth up to $1 trillion by 2040, said Brindis, who discussed implications for primary care from the article, “New Directions for Women’s Health: Expanding Understanding, Improving Research, and Addressing Workforce Limitations.”

“Clearly, we need to define women's health, emphasizing the need for holistic, whole person approach to their health considerations of biology, genetics, sex, gender, and many other intersecting social, economic, behavioral, environmental, and structural upstream factors that affect their health,” Brindis said. “Historically women's health has been reduced to a singular focus, primarily reproductive health, that fails to recognize the full array of health issues that impact their health and the vital role that primary care plays in their lives both before and after some of their reproductive decision making.”

There is a challenge in that women’s disease symptoms and progression affect the diagnosis, management and treatment of conditions ranging from heart disease to neurological disorders to mental health to cancer to communicable and noncommunicable diseases. There are significant knowledge gaps, Brindis said.

In primary care, policy implications and potential solutions include:

• Increased investment in women’s health, with a stronger women’s health provider network and workforce payment rates.
• Eliminating disparities in reimbursement for female-centric procedures. That is clearly in the realm of the U.S. Center for Medicare & Medicaid Services and insurance companies.
• Improved provider capacity to more consistently apply evidence-driven practices across systems of care, geographies, and heterogenous segments of the population.

Brindis said she and her coauthors want to address issues ranging from the effects of the COVID-19 pandemic on the women’s health workforce, to the lack of sufficient funding and reimbursement for female-specific procedures, to gender-related contributors to burnout.

“We want to continue to strengthen and not eliminate social medicine initiatives that consider the holistic health of women and the influence and impact of social determinants of health,” she said. “In adopting a comprehensive intersectional and life course framework, we want to prioritize investments in women's health to enhance health outcomes and also achieves societal benefits. We want to close gaps preventing access to high-quality care across the lifespan.”