What worries doctors in 2026: AI dominates the writing, but pain and 'moral injury' win readers

This report — the first in a new quarterly series based on the KevinMD archive, a platform that has published first-person writing by clinicians since 2004 — analyzes the 805 articles published between July and September 2026, 574 of which were written by health professionals (physicians, residents, students, nurses, nurse practitioners and physician assistants) under 321 different bylines. The method matters in itself: instead of starting from a survey with preset questions, it observes what clinicians write about when no one assigns them a topic, and cross-references that with the actual reading time each subject generates among readers.

Three topics dominated the quarter's conversation by a wide margin: artificial intelligence (77 essays by 59 different authors), burnout (36 essays by 31 authors) and pain, opioids and addiction (30 essays by 18 authors). None of the other thirty topics the report tracks exceeded 12 essays. The report's authors stress an important methodological caveat: a topic can look like it is surging when in fact a single prolific pen is behind it — they cite malpractice, where 9 of 12 essays came from one physician — so they only consider a topic to be 'rising' when its growth is sustained by many voices and holds up against the previous four quarters.

Artificial intelligence was by far the most written-about subject: it went from 9.3% of clinician essays in the previous four quarters to 13.4% this quarter. The publication's podcast reflects the same shift, with 16 of 73 episodes focused on AI. But the most interesting figure is not how much is written, but about what: in 2023, three out of four clinician essays on AI (78 of 103) mentioned ChatGPT or chatbots repeatedly; this quarter only 15 of 77 did, fewer than one in five. The conversation has moved from the novelty of the chatbot to the tools already embedded in daily clinical work and their effects on care: a psychiatrist and internist questioned whether a simple 'consent obtained' checkbox in an AI ambient-scribe workflow amounts to real consent; a psychiatric nurse practitioner described patients arriving at appointments frightened by AI-generated prognoses, and proposed calling it 'AI anticipatory anxiety'; other physicians wrote about automation bias and about why prompt engineering fails a doctor on call at two in the morning. Readers, however, have not kept pace: AI essays drew, on average, only two-thirds the reading time of a typical article, and just 15 of the 118 measurable AI pieces (by any author) made the quarter's most-read quartile, when simple proportion would have predicted around 30.

Burnout, although still the second most written-about topic, shrank from 8.5% to 6.3% of clinician essays compared with the previous four quarters. The report warns that a single quarter does not make a trend, but points to a more consistent underlying shift: the term 'moral injury' has held steady at roughly 1–2% of clinician essays for about a year, and ten different authors addressed it this quarter. Psychiatrist Devina Maya Wadhwa explains why more and more clinicians are looking for another word: 'that word [burnout] is doing too much work.' Physiatrist Sarah A. Vlach puts it bluntly: 'moral injury is not exhaustion from overwork.' Reader interest backs this semantic shift: burnout essays drew 1.49 times the typical reading time, and moral injury essays 2.79 times, among the highest figures of the entire quarter.

Pain, opioids and addiction saw a notable surge: their share nearly doubled, from 2.8% to 5.2% of clinician essays. A single physician wrote a third of those pieces, but the report found that even excluding him, the other 17 authors on their own more than doubled the topic's share, ruling out a single-voice effect. The essays discuss both access to treatment and the drugs themselves: Simon Feng, a family physician and addiction medicine specialist, wrote that the system surrounding chronic pain 'benefits enormously from, and even demands, the suppression of opioid prescribing.' Francisco M. Torres, an interventional physiatrist, described patients who arrive at a pain referral expecting suspicion and who 'often say they feel judged, misunderstood and abandoned.' Other authors wrote about kratom and 7-OH, harm reduction, and postoperative pain control lost to health policy. Readers responded strongly: the typical essay on this topic generated 60% more reading than an average article, and Feng's piece on why doctors don't prescribe opioids was among the ten most read of the quarter.

GLP-1 drugs grew from a small base: 11 essays by 9 authors, from 1.3% to 1.9% of clinician essays. The focus was less on the drug than on what happens around it: endocrinologist Hana Kahleova wrote about weight regain after stopping treatment and argued that 'we are funding half of this equation with the less durable effect.' Nurse Patience C. Ennis addressed the stigma that follows patients even after they lose weight, captured in one patient's question: 'If you knew what it took to change my body, would you think less of me?' A family physician also explained what a tirzepatide study that had circulated in the media was actually measuring.

Misinformation and social media also grew from a small base, nearly doubling their share (from 0.9% to 1.7%, with 10 essays by 9 authors), largely because clinicians encounter it first in the exam room. Physician and writer Arthur Lazarus explained why corrections arrive too late: 'by the time a correction appears, the claim may already feel familiar, socially reinforced and emotionally true.' Physician Elizabeth Agyeman Prempeh made the opposite case: 'I believe true medical experts need to become influencers.' It is, however, one of the topics the report explicitly flags as unrewarded by the audience: misinformation essays drew 30% less reading than an average article.

Three subjects lost ground against the previous year: insurance, coverage denials and billing fell from 2.3% to 1.4%; vaccines from 1.4% to 1.0%; and, as noted, burnout from 8.5% to 6.3%. Private equity in health care, despite the media attention it usually receives, barely appears: it was the central topic of just 2 essays this quarter (though it was mentioned in 15), and across two decades of archive it has never been the main focus of more than one in 250 articles in any year. When a physician does take it on directly, the tone is more historical than accusatory: Brian Hudes, a gastroenterologist who owned his practice for twenty years before spending more than ten as a hospital employee, wrote that 'private equity did not make independent practice impossible. It came later and bought what was left, at a discount.'

One of the quarter's most striking findings is the list of the ten most-read articles, measured by engaged reading time: every single one was written by a physician, something the report itself says would happen by chance only about once in 250 tries (given that physicians wrote nearly six of every ten articles in the quarter). The titles include pieces such as 'Women physicians die younger. Hospitals still aren't acting' (Ali Novitsky), 'I don't like who I've become in primary care' (Wendy Shue), 'COVID trauma in emergency physicians can surface suddenly' (Howie Mell), 'Why doctors leave medicine after a malpractice lawsuit' (Jordan Grumet) and 'The administrative burden in primary care and the end of my career' (Susan Bennett), among others, alongside the Simon Feng essay on opioids already cited. Shue sums up the common thread in ten words: 'I don't like who I've become, working in primary care.' Tellingly, none of the ten is about artificial intelligence, even though AI accounted for one in six essays in the quarter — consistent with the broader pattern: the typical clinician article generated, on average, a third more reading than the typical article by any other kind of author.

The report adds a historical perspective that illuminates the present. For a decade, the electronic health record was the dominant technology in writing by health professionals: it led every year between 2009 and 2019, peaking in 2014 at roughly one in 27 articles, before settling from 2020 at around one in 100. Artificial intelligence stayed below 1% of articles until 2019, and in 2026 it is already the focus of roughly one in eight. Burnout followed a similar arc: in 2008 no article treated it as a central topic, and by 2025 one in 14 did. From those electronic health record years comes a line the report links directly to today's AI debates: obstetrician-gynecologist Matthias Muenzer summed up why those systems were designed the way they were in six words: 'the buyer is not the user.' Today's AI essays, the report notes, ask in various ways whether the new tools will repeat that same mistake.

Looking ahead to coming quarters, the report raises several open questions: whether the surge in pain and opioids holds and broadens beyond a few prolific voices; whether writing about AI keeps growing while readership lags; whether more clinicians adopt 'moral injury' as the term for their distress; and whether GLP-1s and misinformation keep growing from their still-small bases. Taken together, the picture this first quarterly report draws is of a profession writing prolifically about the technology transforming its daily work but which, when it comes to reading, gives more attention — and more time — to writing about the human cost of practicing medicine: exhaustion, untreated pain, and the gap between what practice demands and what the person practicing it can sustain.