Integrating AI into family medicine education and practice

In a 2021 editorial, Drs. Winston Liaw, Ioannis Kakadiaris, and Zhou Yang asserted that embracing artificial intelligence (AI) is “the key to reclaiming relationships in primary care.” For example, AI tools can efficiently identify patients at high risk for poor outcomes, perform triage, provide clinical decision support, and assist with visit documentation. On the other hand, AI “could just as easily make things worse by leading to endless alerts, nonsensical notes, misdiagnoses, and data breaches.” To avoid having AI reenact the cautionary tale of electronic health records and cause more problems than it solves, Dr. Liaw and colleagues encouraged family physicians to partner with researchers, participate on health information technology committees, and lend their primary care expertise to computer scientists developing AI tools.

In the future, medical students, residents, and practicing clinicians will need to meet basic competencies for the effective deployment of AI-based tools in primary care. In a recent special report in the Annals of Family Medicine, Drs. Liaw, Kakadiaris, and colleagues proposed six competency domains for family medicine:

1) foundational knowledge (what is this tool?), (2) critical appraisal (should I use this tool?), (3) medical decision making (when should I use this tool?), (4) technical use (how do I use this tool?), (5) patient communication (how should I communicate with patients regarding the use of this tool?), and (6) awareness of unintended consequences (what are the “side effects” of this tool?)

The report provided examples of AI competencies within each domain based on learner roles (student, resident and faculty) and noted that primary care team members other than physicians would also benefit from additional training in AI.

AI-enabled chatbots, which can be trained to write intelligible text and complete essays in response to specific queries, are already changing the way universities assess students and have the potential to distort the scientific literature. This month, the World Association of Medical Editors released a preliminary statement advising medical journals that chatbots cannot be authors, and that authors who use chatbots for writing assistance remain fully responsible for their work and should be transparent about how chatbots were used. (The journal Cureus has taken a different approach, inviting the submission of case reports written with the assistance of the chatbot ChatGPT and asking that the AI tool be named as an author.)

In August 2022, the U.S. Department of Health and Human Services (DHHS) announced its intention to confront health care discrimination resulting from the application of biased clinical algorithms and tools through a proposed rule that could hold clinicians liable for clinical decisions made by relying on flawed AI-based tools. A JAMA Viewpoint recommended that DHHS shield clinicians from liability if they are following the accepted standard of care (e.g., utilizing the American College of Cardiology / American Heart Association Pooled Cohort Equations, which generate higher cardiovascular risk estimates for patients who identify as Black) and work closely with the U.S. Food and Drug Administration to determine how to best assess algorithmic software for bias.

Clinical algorithms are not the only way that AI could worsen health inequities in primary care. In a systematic scoping review in Family Medicine and Community Health, Dr. Alexander d’Elia and colleagues identified 86 publications that discussed potential negative effects of AI on access (“digital divide”), patient trust, dehumanization / biomedicalization, and agency for self-care. It described approaches to improving equity in AI implementation that included prioritizing community involvement and participation and considering system-wide effects outside of the primary care setting.

**

This post first appeared on the AFP Community Blog.

New clinical recommendations on osteoporosis treatment

The American College of Physicians (ACP) has updated its 2017 clinical practice guideline on treatment of primary osteoporosis or low bone mass to prevent fractures in adults. The previous version, which was endorsed by the American Academy of Family Physicians, recommended treating women with osteoporosis for 5 years with alendronate, risedronate, zoledronic acid, or denosumab to reduce the risk of hip fractures and vertebral fractures. It also suggested that men with clinically recognized osteoporosis be offered bisphosphonates to reduce the risk of vertebral fractures. Treating older women with low bone mass (osteopenia) at high risk for fracture was deemed to be optional based on patient preferences and medication benefits, harms, and costs. The publication of additional studies on existing therapies and the availability of new therapies such as abaloparatide prompted this guideline update.

An independent evidence review team performed a systematic review and network meta-analysis of osteoporosis treatments that analyzed 34 randomized, controlled trials and 36 observational studies. The review confirmed the effectiveness of bisphosphonates and denosumab in reducing hip, vertebral, and other clinical fractures. In older postmenopausal females at very high fracture risk, abaloparatide, teriparatide, and sequential romosozumab, then alendronate, appeared to be more effective at reducing clinical fractures over 24 months than bisphosphonates. Harms of therapies included an increased risk of adverse events with abaloparatide and teriparatide and a small absolute increased risk of atypical femoral fractures and osteonecrosis of the jaw persons taking bisphosphonates for 36 months or more.

In the updated guideline, the ACP now recommends that clinicians preferentially use bisphosphonates first-line in women and men with osteoporosis, with the exception of women at very high risk of fracture. In this group, either romosozumab or teriparatide can be used, followed by a bisphosphonate. Denosumab is endorsed as a second-line therapy for adults with contraindications to or who experience adverse effects from bisphosphonates. Similar to the 2017 guideline, the ACP suggests an individualized approach to prescribing bisphosphonates in women over age 65 with low bone mass. Rather than revisiting the recommendations in a specific time frame, the ACP plans to perform quarterly literature surveillance and maintain this topic as a “living” guideline that will be updated periodically when new evidence becomes available.

The U.S. Preventive Services Task Force currently recommends screening for osteoporosis in all women age 65 years and older and postmenopausal women younger than 65 years at increased risk using a clinical risk assessment tool. It found insufficient evidence to assess the balance of benefits and harms of screening for osteoporosis in men. The American College of Obstetricians and Gynecologists recently made similar screening recommendations. A 2020 report from the Women’s Health Initiative study found that repeating bone mineral density (BMD) testing after 3 years did not provide more clinical information than a baseline measurement. Additional information on osteoporosis diagnosis and treatment is available in AFP By Topic, including a Lown Right Care article on making decisions about fracture prevention in older adults.

**

This post first appeared on the AFP Community Blog.

A world without Twitter

Sometime during the week after Christmas, my Twitter account was hacked. Someone with an IP address in Quebec stole my password, logged in as me, and changed my e-mail recovery address so that I would not be able to reset my password. I have contacted Twitter Support three times since December 29 and have not received a response. After Elon Musk's reported "firing frenzy" after his acquisition of Twitter, it isn't clear to me when or if I ever will.

In the meantime, if you don't subscribe to Common Sense Family Doctor by e-mail or an RSS feed reader, you can also find notifications of new posts at the following websites:

Facebook: https://www.facebook.com/commonsensefamilydoc/

Substack: https://commonsensemd.substack.com/

LinkedIn: https://www.linkedin.com/in/kennylinafp/recent-activity/shares/

Improving early cancer diagnosis: it's (mostly) not about screening

In a recent medical news item that you may have missed, an analysis from NORC at the University of Chicago determined that only 14% of cancers in the U.S. are diagnosed by a recommended screening test for breast, cervical, colorectal, or lung cancer.  An additional 11% represent prostate cancers detected through PSA screening, which isn't technically recommended. Adding these two percentages together and subtracting from 100% means that 75% of cancers are either detected incidentally or after patients develop symptoms that cause them to seek medical care. Notably, the study was funded by GRAIL, which sells the Galleri blood test for screening for many types of cancer at once (most without currently recommended tests), and the company no doubt plans to use the results to increase demand for its unproven $949 test. 

However, there is another way to respond to this analysis. If three-quarters of cancers are detected after symptoms develop, the medical community should focus on improving outcomes by reducing the time from symptoms to cancer diagnosis. In a 2022 JAMA Viewpoint and a more detailed paper in Cancer Prevention Research, Dr. Elizabeth Sarma and colleagues made the case for this approach, arguing that symptom detection should be viewed as a "partner to screening" in primary care. People with possible cancer symptoms don't always seek timely care; a mixed-methods systematic review of 80 studies suggested that older adults often initially attribute symptoms to normal aging, but when they do recognize them as potentially serious, they are quicker to see a doctor than younger persons. Another study found that patients with more than two chronic conditions had a longer diagnostic interval (time from primary care presentation to cancer diagnosis) and a higher likelihood of seeking emergency care, possibly because clinicians incorrectly attributed the symptom to the pre-existing condition rather than cancer.

The diagnostic challenge that family physicians face is that most patients with common symptoms that could be due to cancer don't have cancer. If I ordered a CT scan or referred to a gastroenterologist every adult who presented to my office with abdominal pain, many patients would endure a lot of unnecessary procedures to identify the few with colorectal cancer. A 2019 review found few electronic clinical decision support tools for cancer diagnosis in primary care. However, a Veterans Affairs health system study concluded that visiting a primary care clinician at least annually is associated with substantially lower risks of metastatic disease at time of diagnosis and cancer-related death. So what factors influence our decisions to perform tests or refer patients with symptoms that could represent cancer? A systematic review found that the only factors that consistently prompted more diagnostic workups and referrals were alarm symptoms (e.g., fever or unexplained weight loss) and a "gut feeling" that a serious cause was responsible. This isn't good enough. Alarm symptoms are generally obvious, and gut feelings may overestimate or underestimate risk depending on the physician's training and experience.

Although they perform less cancer screening than we do in the U.S., the United Kingdom is well ahead of us in refining systems for early diagnosis. Dr. Sarma observed that U.K. researchers used data from their national heath system to "generate symptom lists and corresponding positive predictive values ... [that] were used to develop interactive calculators for primary care practice to predict an individual's risk of cancer." She endorsed a three-pronged research agenda: describing pre-diagnostic care pathways for symptomatic cancers; identifying signs and symptoms that can be used to identify patients at higher risk for specific cancers; and improving diagnostic pathways for symptomatic patients by increasing patient awareness and improving point-of-care tests in primary care. Can the U.S. successfully emulate the U.K. model of improving early cancer diagnosis?

A "hot take" on screening colonoscopy

Screening for colorectal cancer is an important preventive health practice that saves lives. But is colonoscopy really the "gold standard" for colorectal cancer screening? In Episode 172 of the American Family Physician podcast, I provided my "hot take" on a recent randomized trial that was designed to inform the answer to this question. You can listen to it in the embedded player starting at 22:50 or read the transcript below. Health care professionals may also be interested in a more in-depth discussion that I participated in for Medscape.

Hi, I’m Kenny Lin, deputy editor of AFP and an expert in cancer screening guidelines.

In 2002, the US Preventive Services Task Force first recommended colonoscopy as a primary screening test for colorectal cancer (CRC) in adults. This was an uncharacteristic decision, since the first randomized trial of colonoscopy would not be published for another 20 years.

Since then, flexible sigmoidoscopy has virtually disappeared as a screening option, and colonoscopy has become the primary screening method in the U.S. Gastroenterologists call it the “gold standard” and portray stool-based tests as an inferior alternative to be offered only to patients who refuse.

Of course, colonoscopy is less convenient and has serious risks that stool tests don’t: perforations, bleeding, and infections. The Task Force and others have assumed that colonoscopy saves more lives than stool tests, which reduce CRC mortality by around 15%, or flex sig, which lowers it by 25 to 30%. So the first trial results were surprising, even shocking. After 10 years, the group invited to undergo screening colonoscopies developed fewer cancers, but there was no change in CRC mortality.

Some have argued that a longer follow-up period, higher adherence in the intervention group, and better trained endoscopists might have produced better results. But at a minimum, this landmark trial suggests that it is not accurate to inform patients that colonoscopy is the best test for CRC screening or ethical to recommend it preferentially.

Instead, we should explain that stool-based tests and colonoscopy have different benefits, harms, and screening intervals; that either test is better than none; and then let them decide. On a health system level, it may be worth taking another look at flex sig, an office procedure that older family physicians like me were trained to perform before the promotion of screening colonoscopy got out ahead of the evidence.