Opinion: What STAT readers think about M.D. vs. D.O., AI in medicine, surrogacy, and more

First Opinion is STAT’s platform for interesting, illuminating, and provocative articles about the life sciences writ large, written by biotech insiders, health care workers, researchers, and others. To encourage robust, good-faith discussion about issues raised in First Opinion essays, STAT publishes selected Letters to the Editor received in response to them. You can submit a Letter to the Editor here, or find the submission form at the end of any First Opinion essay. The story “Tear down the wall between M.D. and D.O. education,” by Abigail MacKenzie and Vijay Rajput The response While I appreciate Drs. MacKenzie and Rajput for helping to shine a light on the bias and unfair challenges graduates of osteopathic medical schools face in residency placements, the answer is not for D.O. grads to surrender and give up the principles and practices that they bring to the health care system. While it may be hard for some M.D. physicians to understand, the separate pathways are more than just logistics. Organizations, examinations, accreditation systems, and professional structures matter, but they are not what makes osteopathic medical education worth preserving. We need to preserve something much more fundamental: the cognitive architecture of osteopathic medical education. By cognitive architecture, I mean the intentional organization of knowledge, experiences, scientific disciplines, and professional values to shape how physicians perceive, reason through, and solve clinical and health system problems. Every allopathic medical school (M.D.) teaches biomedical and clinical science. Increasingly, they teach health systems science as well. The distinction between allopathic and osteopathic education is not whether these subjects exist elsewhere. The distinction is how they are intentionally integrated. Osteopathic principles and practices are more than just an add-on elective. Osteopathic medical education intentionally integrates biomedical, clinical, health systems and biomechanical sciences, and concepts aligned with complexity science into a coherent framework for physician formation. Osteopathic philosophy serves as the integrative lens through which these domains influence clinical reasoning. For the D.O. community, the result is physicians who strive to excel at both promoting and maintaining health and treating disease; who value relationships as well as diagnoses; who recognize that structure and function are linked; who appreciate context, adaptation, prevention, and community; and who use medical, surgical, manual, behavioral, and social interventions as complementary tools in patient care. Given the incredible growth of osteopathic medical education, this philosophy and these principles are resonating with prospective medical students and patients in record numbers. Today, about 30% of all medical students in the U.S. are educated at a college of osteopathic medicine, and that percentage is growing every year. Those students deserve equal opportunities, and that is why the American Association of Colleges of Osteopathic Medicine is working with our peers in the medical community and why we’re supporting the Fair Access in Residency (FAIR) Act. The answer to our nation’s physician shortage crisis is to fully open the doors to residency for all qualified candidates, not to close the door on those who choose the osteopathic pathway. — Robert Cain, American Association of Colleges of Osteopathic Medicine The response Excellent article and well researched. Most of my colleagues never knew or cared that I was a D.O. They cared more about the care I offered and whether I could accept constructive criticism when it was due. I completely agree with the article, but I think the credentialing boards generate so much revenue that they are reluctant to combine (and possibly line their pockets a little less) and work together. Perhaps we even do away with the maintenance of certification altogether. Isn’t passing boards and maintaining continuing medical education credits, along with the threat of lawsuits always looming, enough to ensure we keep up with the science? — Sharon McKelvey The response There are more differences in the training of M.D. students and D.O. students than just the modality of osteopathic manipulative treatment. The entire approach to the patient is different. Holism (a term that has been grossly overused and misunderstood) is a tenet that stresses a more global understanding of a patient and all that influences their health and well-being. This approach begins in year one and continues throughout all (usually) four years. D.O. training focuses heavily on history and physical diagnosis and tends to create a higher percentage of primary care providers because of this broad approach. All facets of scientific evidence-based medicine and surgery are taught and complementary, and alternative medicine is not avoided. In fact, osteopathic manipulative medicine focusing on circulation and innervation is a central modality during training. Most of us who have spent careers as osteopathic physicians in a largely allopathic American health care system have learned to value and embrace these unique aspects. Vive la différence! — Alfred Sassler, University of Cincinnati Medical Center/UC Health The story “AI won’t enhance physician autonomy. It will further diminish it,” by Frances Mei Hardin The response Thank you for this article. I couldn’t agree more that “lack-of-autonomy creep” has been happening for a long time. It would be OK if the new structures and “tools” were mostly in the service of better patient care, but let’s be honest, most new “advancements” are not. As the author points out, our autonomy has been removed from us piece by piece. The power is in the hands of the owners of the means of production, and that is not the doctors. We are the workers, not the factory owner, albeit well-paid ones, which is one reason we don’t complain. How can money-driven incentives improve patient outcomes? Let’s be honest about that. Venture capital couldn’t care about patient outcomes as much as we do. AI will be deployed for cost savings primarily. The AI we happily use for our notes will be the loser to the AI the insurance companies use to tell us we have “overcoded.” Physicians are suffering from a history of autonomy that is currently an illusion. This long-defunct echo prevents us from organizing and exerting the power we do have to influence our destiny. I am by no means a Marxist, but in this case the metaphor is apt. We have gone from being respected autonomous experts to unempowered, well-paid “factory workers.” The attribution of liability that the author references is but one example. The answer is to not passively accept what is given but to organize to make our voices heard. And sorry, our professional organizations as they currently exist are not the answer, as they are in bed with corporate (insurance, pharma) interests, something almost too obvious to have to state. — Jody Whitehouse, M.D. The story “Whale songs: the sounds of love and loss as heard by an OB-GYN,” by Caroline Cherston The response These observations rang true to me because of my experience working in the pit right after Sept. 11. My therapy dog and I accompanied grieving families to ground zero in the days following the attacks. Mothers accompanied by their adult children would arrive at the still burning disaster, still in denial that their child had been killed. Once there, staring at the devastation, there was no denying the reality of death. These women would scream a very distinctive wail, something I had never heard before and will never forget. As a park ranger living with brown bears in Alaska, I have witnessed what I interpret as grief when a sow loses a cub. Now I work at a national park where we teach our visitors about whales. Grief is human, grief is universal. Would that we studied it more and embraced its process. — Naomi Boak, National Park Service The story “The Lindsay Clancy case shows the limits of postpartum psychiatry,” by Jamie Maguire, Lauren M. Osborne, and Jennifer Payne The response The authors are right on the mark arguing that scientific knowledge on postpartum psychopathology is lacking, and that this gap is likely to have played a role in the tragic Clancy case. I agree with their argument whereby the major flaw of extant knowledge is lack of precision, particularly concerning risk factors for developing severe, even lethal, conditions. Most importantly, by focusing on science, the authors steer the discussion away from (unsubstantiated) claims about psychiatry’s negligence and toward where we, collectively (science, mental health services, policymakers, and the public) should head to prevent such tragedies. I particularly identify with the article’s ending, which suggests that ultimately, precision enables compassion. One aspect, however, is conspicuously absent from the authors’ vision: the role of clinical psychological science in achieving such precision. The past four decades have seen a proliferation of methodologies enabling increasingly precise assessment of risk and protective factors in psychopathology, including psychometrically sophisticated self-report measures, investigator-based semi-structured interviews, computerized tasks targeting cognitive-affective processes, and the validation of these methodologies against neural processes and overt behavior. Perhaps the most intriguing risk factor for psychopathology gleaned from clinical psychological science, and one that is particularly relevant to the assessment and treatment of postpartum psychopathology, is malignant self-criticism. Defined as the tendency to set impossibly high self-standards and to adopt a punitive self-stance once these standards are (inevitably) unmet, self-criticism emerged as a robust predictor of virtually all forms of psychopathology, including suicide and violent behavior. Self-criticism has also been shown to derail evidence-based psychotherapeutic and psychopharmacological treatment of depression, even though more recent studies have shown that self-criticism responds well to distinct forms of treatment. Importantly, such research demonstrated the formidable risk associated with self-criticism for postpartum depression and anxiety, as well as for disruptions in mother-child communication, which might exacerbate postpartum psychopathology. Of course, we cannot know the extent to which malignant self-criticism played a role — if any — in the Clancy case. Other psychological risk factors may be in play. The bottom line, however, is that psychiatry and psychology should not compete over markers. Rather, they must join forces, in science and practice, in identifying a wide spectrum of factors that increase vulnerability to the chilling outcome manifested by the Clancy case and other tragic ones. The aim, therefore, is biopsychosocial precision in mental health science and practice. — Golan Shahar, Ben-Gurion University of the Negev The story “8 steps to making surrogacy more ethical for everyone involved,” by Arthur Caplan The response The problem is with the very concept of “surrogacy,” which means acting on behalf of someone else. Our ideas of surrogacy are based on a patriarchal notion that the essence of the baby is in the seed. While the original idea was that men had the seeds, medicine eventually had to acknowledge that women have seeds too, and babies are thus “half his and half hers.” Our society, and this discussion, claim that the real, the meaningful relationship is with the seed, and the pregnant woman is just a place, a spot you plant the precious seed. (In ordinary English we refer to the earth itself as dirt, soil in which we plant seeds.) But pregnancy is not a location — it is a relationship. Babies are born recognizing the language the pregnant woman spoke. The comfort “rocking” offers a baby is its duplication of the walking motion. Hiring someone to be the location disregards the relationship. Can I hire someone else to be the wife for a husband, live with him, interact physically and socially — and then fire her, take him back, claim she was only a surrogate? We need to acknowledge and respect the relationship that is pregnancy. — Barbara Katz Rothman, City University of New York The story “Medical aid in dying and the burden of choosing to live,” by Monika Piotrowska The response I am 82 years old and still healthy, but — let’s face it — for how much longer? The analogy the author makes between a student learning that they are not doing well academically and facing the inevitability of dying does not work for me at all. The author speaks of autonomy, and I don’t understand that, either. Because my husband has died of dementia with no autonomy at all and it wasn’t pretty. Because I want to be able to decide I am ready to die and I know that I soon will be ready to die (but still hoping to outlive Trump). I have had a long and really wonderful life and I am content. What I fear is burdening others with my care. I don’t want to impose that on my children. I want to be able to make the choice about the timing of my death so as to retain my autonomy and my ability to make choices, to have quality, not quantity of life. MAiD promises this to me, and is such a tremendous relief. — Nora Staffanell The story “AI has created a shadow medical system,” by Arya Rao and Marc Succi The response In their recent essay on AI and the “shadow medical system,” Arya Rao and Marc Succi close with the 1979 IBM assertion that “A computer can never be held accountable, therefore a computer must never make a management decision.” Applied to health care, that claim has been outdated at least since 2018. That was the year LumineticsCore received the first FDA de novo clearance for an autonomous medical diagnosis — diabetic retinopathy — and began carrying malpractice insurance for its no-clinician-in-the-loop system. And just this year, Doctronic launched a prescription-renewal pilot that has some physician oversight, but not over every clinical decision made by the AI. Accordingly, Doctronic also carries a malpractice policy holding the system to physician-level accountability. The problem of AI accountability for health decisions is real and important. It is no longer accurate to say it is unsolved. — Joel Selanikio, M.D., futurehealth.live The response While the article makes several insightful points about AI assuming clinician responsibilities, the bigger picture should be on who is accountable for those decisions. Rao and Succi correctly identify accountability as the dividing line between a useful medical tool and a functioning medical system. Accountability is more complicated than ensuring a physician remains nominally involved. “Human oversight” becomes meaningless when responsibility and authority are separated. We’ve seen this problem with clinical decision support for years. Technology identifies an abnormality or recommends action, while the difficult question — who owns the next step? — remains unanswered. AI makes that long-standing design failure much more consequential. If the default is to lay accountability at the clinician’s feet, we are in danger of creating an unsustainable system. A clinician cannot meaningfully be accountable for AI decisions if they did not select the model, cannot interrogate its reasoning, cannot see its underlying data, or do not have a realistic opportunity to intervene. The American Medical Association has called for transparency, accountability, physician oversight, regular auditing, and clearer standards around AI-enabled clinical decision support. It has also argued that liability should be appropriately apportioned rather than automatically falling on physicians. “Human in the loop” is not an accountability model. Requiring a clinician to approve an AI-generated recommendation may create the appearance of oversight without giving that clinician genuine control. “Clinicians must review an AI output” is closer to a disclaimer than a governance policy unless an organization clearly defines where human judgment is required and leaves space for clinicians to exercise that judgment. That thinking applies to the shadow system. A click, signature, or brief review should not magically transfer responsibility from an AI developer or platform to a physician. Accountability has to extend beyond the diagnosis to the care loop. Technology often succeeds at identifying a problem while failing to establish who owns what happens next. We must design clinical systems around explicit responsibility and completion of the care process, rather than merely generating alerts or recommendations. This idea maps to Rao and Succi’s concern about AI correctly identifying abnormal lab values and the trail of questions that follow: Who confirms it? Who determines its significance in context? Who ensures follow-up occurs? Who recognizes when the algorithm missed the dangerous alternative diagnosis? That’s where “answering a medical question” becomes practicing medicine. — Craig Joseph