How are clinicians utilizing AI to answer medical questions?

At the turn of the century, Dr. John Ely and colleagues observed 103 family physicians for two half-day clinical sessions and analyzed 1,101 questions that arose during patient care​. Physicians spent an average of less than 2 minutes seeking answers to 399 (36%) clinical questions. Nearly all information sources were colleagues or printed materials; only 10 times did physicians consult a computer. In contrast, a 2022 study​ of 31 family and other generalist physicians found that 93% of clinical questions were answerable with online resources, mostly using summaries of pre-appraised and synthesized evidence. However, for 32% of questions it took more than 5 minutes to retrieve an answer, too slow to be useful at the point of care.

Clinicians can use artificial intelligence platforms such as Open Evidence (OE) to answer questions more efficiently. (Full disclosure: OE partners with the American Academy of Family Physicians to surface content from American Family Physician in responses to queries.) A 2025 retrospective study​ evaluated OE’s performance in providing evidence-based recommendations for 5 primary care patients with common chronic conditions (hypertension, hyperlipidemia, type 2 diabetes, depression, and obesity). Physician reviewers rated OE’s recommendations highly for clarity, relevance, and evidence support, but noted that these generally reinforced physician plans rather than modifying them. Earlier this year, OE added strength of evidence ratings to answers​, ranging from A (strong) to D (minimal).

In a descriptive study in JAMA​, Dr. Vishal Patel and colleagues reported the content of more than 270 million queries submitted to OE by more than 950,000 US clinicians (including 556,200 physicians) from January 2024 to June 2026. On average, nurse practitioners and physician assistants submitted more queries per month than physicians. Residents, fellows, and early career physicians used OE more frequently than physicians with more than 10 years in practice.

Across medical specialties, hematologist-oncologists, internists, and family physicians submitted the most queries relative to their proportions in the workforce. Large numbers of queries were directed outside of a physician’s specialty; for example, 70% of queries from surgeons involved nonsurgical topics. Although the majority of queries (51.9%) were tied to a specific patient care decision, other purposes included general clinical knowledge, retrieval of research studies or guidelines, administrative tasks (eg, insurance appeal letters, patient education), teaching materials, and professional development.

An accompanying editorial​ highlighted some of the study’s limitations and directions for future research:

Patel and colleagues did not evaluate the quality of generated responses to queries, clinicians’ assessments of their accuracy or usefulness, or how the information was applied to or affected patient care. The study also could not distinguish use during clinical encounters from “precharting” preparation or general educational efforts. Returning users suggest perceived value, but engagement is not evidence of patient benefit.

It has never been easier for clinicians to answer questions that come up at the point of care. As always, the relevance and validity of retrieved information​ determines its usefulness, potential to change practice, and likelihood of improving patient-oriented outcomes.

**

This post first appeared on the AFP Community Blog.

Reflections and reality checks on the state of primary care

It has become fashionable among high-impact research journals to publish content bemoaning the "crisis" in primary care. Last year, Dr. Lisa Rosenbaum wrote a series of Medicine and Society columns in the New England Journal of Medicine, including one titled "Why Have We Chosen Not to Fix Primary Care?" In July, Health Affairs held a virtual event with researchers and policymakers on "Confronting the Primary Care Crisis." Last week, Dr. Eli Canan wrote a news article in JAMA titled "Inside the Growing Crisis of Primary Care."

The causes of the crisis boil down to several key points. Well-resourced primary care is associated with better population health, lower health care spending, and lower mortality, but there are too few primary care clinicians in many parts of the U.S. This shortage is worsening because not enough medical students are choosing family medicine or other primary care fields. Financial incentives built into health systems and medical training environments are stacked in favor of subspecialists and against primary care. Medical students are largely selected from socioeconomically advantaged backgrounds, and those interested in primary care are often actively discouraged by academic advisors and institutional structures that overvalue subspecialization. At a recent conference, the new president of the American Academy of Family Physicians presented two decades of data showing that while the number of family medicine residency programs has more than doubled, the number of American medical graduates entering the field has barely budged. That has led to an increasing reliance on international medical graduates to fill residency slots, which has been endangered by new restrictions on immigration, particularly a $100,000 fee for H-1B work visas.

Two major articles by family physician colleagues provide deeper dives into the distressed state of primary care and potential solutions. Writing in the Journal of the American Board of Family Medicine, Dr. Richard Young asks "Why Are There Never Enough Family Physicians to Provide Care for the People of the United States?" The short answer is that despite the efforts of many well-meaning policymakers, American health care institutions prefer to prioritize the construction of shiny proton therapy centers than build the less visible physical and intangible architecture of primary care. The longer answer is a list of problems that need solving:

1. The US medical school environment is toxic to family medicine

2. AMA CPT codes undervalue the work of family medicine

3. CMS documentation rules are unnecessarily burdensome to family medicine

4. There is little funding for family physician care research

5. Too little of US healthcare spending is for primary care

In a Milbank Quarterly perspective on "The State of Primary Care in the United States," Dr. Robert Phillips and colleagues illustrate how federal efforts to promote high-quality primary care have stalled or gone backwards since the 2021 National Academy of Medicine report. A promising initiative within the Department of Health and Human Services (HHS) to strengthen primary care was abandoned by the Trump Administration in favor of studying (and exaggerating) childhood vaccine harms, claiming that AI is better informed about medicine than doctors, and paving the way for people to inject untested and unregulated peptides. It is not at all clear whom in the health care workforce HHS Secretary Kennedy believes will be left to tackle the chronic disease epidemic in lieu of family doctors.

Will improving breast cancer risk assessment with AI lead to better outcomes?

The U.S. Preventive Services Task Force recommends that clinicians offer risk-reducing medications to women who are at increased risk of breast cancer and at low risk for adverse medication effects. However, clinical tools used to estimate risk of breast cancer, such as the Breast Cancer Risk Assessment Tool​ and the Tyler-Cuzick Risk Model​, perform only modestly better than random chance, with lower performance in African American and Hispanic women​. The limitations of these tools have stimulated the development of artificial intelligence (AI) models that analyze patterns in screening mammography images​ to more accurately predict breast cancer risk in all populations.

A 2023 systematic review​ of 16 studies found that the median area under the receiver operating characteristic curve (AUC) for AI imaging tools was 0.72, compared to 0.61 for breast density or clinical risk factor-based tools. Of note, in 7 of 8 studies that compared AI risk prediction to assessments that combined AI with clinical risk factors, the latter did not produce more accurate risk estimates than AI alone.

In a later study​ of an AI model that utilized retrospective data from more than 300,000 patients aged 50 to 70 years who had screening mammograms in the United Kingdom, the overall AUC was 0.7, with no differences in performance across ethnic or age groups. Similarly, in a recent U.S. study​, a deep learning model involving mammograms from more than 67,000 patients produced an AUC of 0.71, compared to an AUC of 0.53 for breast density.

Last year, the U.S. Food and Drug Administration approved an AI test that analyzes a single screening mammogram performed within the past 12 months to classify patients as at average (less than 1.7%), intermediate (1.7% to 3%), or high (greater than 3%) risk of developing breast cancer over the next 5 years. Everlywell, a digital health company, is charging $249 for this test​, which does not require a doctor’s order and is not covered by insurance. Patients are directed to share their report with their primary care clinicians to discuss next steps, which for those at intermediate or high risk could include more intensive screening or risk-reducing medications.

The trouble is, we don’t know if more accurate cancer risk prediction improves patient-oriented outcomes. Previous trials of risk-reducing medications such as tamoxifen and raloxifene used eligibility thresholds based on clinical risk assessments and showed reductions in breast cancer diagnoses, but not breast cancer deaths. Lifestyle modifications such as increasing physical activity, healthy eating patterns, and limiting alcohol consumption also reduce breast cancer risk, but whether women with increased risk will be more motivated to adopt these behaviors is unclear. In the absence of such evidence, AI tools may only increase preventive care costs without producing health benefits.

**

This post first appeared on the AFP Community Blog.

Time, access, and continuity: unleashing the power of primary care

In a recent editorial coauthored with the editors of six other US family medicine journals, Dr. Sumi Sexton, American Family Physician’s editor-in-chief, highlighted the established benefits of having continuous primary care. These include higher screening rates, fewer emergency department visits and hospitalizations, and lower costs of care for patients with chronic diseases. A 2020 systematic review found that better primary care continuity is associated with lower all-cause mortality. A retrospective cohort study​ of more than 100,000 adults in 48 Dutch general practices suggested that the benefits of continuity increase with longer duration: people registered with the same practice for greater than 5 years had lower odds of urgent hospital admission and lower hospital costs.

How does continuous primary care achieve positive impacts on health? Dr. Sexton and colleagues wrote:

The beneficial impact of primary care on population health likely reflects the core functions of primary care, including continuous relationships, whole-person care, better access, better quality, a greater focus on prevention, more appropriate and cost-effective testing, and early management of medical issues. Yet there is little emphasis on these core principles and continuous care in the quality measures to evaluate health care.

Paying primary care physicians based on value rather than volume of services, also known as value-based payment (VBP), is intended to promote wellness and improved health outcomes. However, Dr. Jeffrey Millstein observed in a commentary in the Journal of General Internal Medicine​ that misaligned VBP incentives sabotage effective primary care by reducing appointment access, impairing healing relationships, and discouraging medical students and advanced practice providers from choosing primary care careers:

What does the VBP-influenced primary care workday look like? It begins with a schedule overcrowded with perfunctory wellness visits that require excessive documentation. Visits that satisfy metrics crowd out access for patients with more acute symptoms or other concerns that cannot wait for weeks or months. … Diagnosis code suggestion pop-up alerts are now a constant distraction. … Capturing illness severity matters, but the current state is eroding PCPs’ ability to focus on the concerns patients bring to the exam room.

Access for patients is important, but so are sustainable clinician workloads. Medicare billing data suggest that family physicians require an average total time of 26 minutes per office visit, including non–face-to-face work. Based on this figure, Dr. Clark Trapp estimated in an FPM article that a sustainable workload for a full-time clinician is 20 patient visits per day and a panel size of 1,354, considerably lower than the current average of 1,727. Evidence demonstrates that reduced visit time pressure is associated with more comprehensive care​, particularly for patients with three or more chronic conditions. In contrast, primary care physicians with overloaded schedules who chronically experience “time scarcity”​ report disillusionment and dissatisfaction and cope by ordering unneeded testing, placing unnecessary referrals, and triaging sick patients to urgent care and the emergency department rather than seeing them in the office.

Sufficient time, access, and continuity are essential to unleashing the power of primary care to improve and maintain good health. Missing any of these elements greatly diminishes that power.

**

This post first appeared on the AFP Community Blog.

The breadth of family medicine in one meeting

My wife and I returned last night from the annual Family Medicine Education Consortium meeting, which was held in Atlantic City this year. Despite the rainy weather, it was an enjoyable conference. I presented the latest evidence on multicancer blood tests, which are limited and imperfect tools that family doctors should think twice about ordering for their patients even if Galleri is approved by the U.S. Food and Drug Administration. Plenary speakers included family physician and YouTube sensation Mike Varshavski, DO, better known as Doctor Mike. Over two packed days, I was only able to sample a small slice of the 400+ presentations (including posters) by Northeast family medicine faculty, residents, and medical students. Nonetheless, these demonstrate the tremendous clinical and policy breadth of the specialty of family medicine. To show that breadth, and for the curious who want to know what sorts of talks I gravitate to at a conference like this, here are the topics of the oral presentations I attended:

Shaping the future of size-inclusive care: lessons from a medical student elective

Continuous glucose monitor utilization for insulin-dependent patients in family medicine clinic

A mixed-methods evaluation of Pennsylvania IMPLICIT Network sites

A 40-year retrospective cohort study of primary care alumni of NEOMED

Beyond statins

The Pap stops here: cervical cancer screening for women with HIV

From PCE to PREVENT: rethinking CVD risk in family medicine

This is our lane: building confidence in firearm storage safety conversations with patients

Primary care for all Americans: the current evidence and the plan to fix

Joint hypermobility in family medicine

An integrative approach to metabolic syndrome

Enhancing your research workflow with artificial intelligence

Shaping the future of care: where leadership, policy, and AI converge

When ideology dictates science policy: Trofim Lysenko and Robert F. Kennedy, Jr.

30 years ago, I took a college class on the history of Russian and Soviet science. One of the more infamous figures in an otherwise accomplished pantheon of scientists was a man named Trofim Lysenko, who rejected Mendelian genetics and espoused a form of agricultural pseudoscience. In a naturally self-correcting scientific system, he should have been an historical footnote, but by cleverly aligning his theories with Marxist ideology and ingratiating himself with Joseph Stalin, he managed to hold on to power for decades, at the expense of thousands of legitimate scientists and the Soviet people who depended on successful crop yields to feed their families.

Trofim Lysenko

Studying Lysenko, I would have been shocked if I had known that history would repeat itself in the United States, where a similar figure would promote thoroughly debunked theories about childhood vaccines and elevate his ideology above science at every turn. Thanks to elections and Presidential term limits, Robert F. Kennedy Jr.'s reign as the head of the U.S. Department of Health and Human Services (HHS) is likely to be much shorter than Lysenko's in the Soviet Union. However, the damage he has done to our scientific institutions and the scourge of preventable infectious diseases he has unleashed may take many more years to be repaired.

Robert F. Kennedy, Jr.

I am not the first person to compare the HHS Secretary's actions to those of Lysenko, but I don't believe there is a more definitive treatment of this analogy than my perspective article in the Fall 2026 issue of the Journal of Lancaster General Hospital.