Allina physicians strike over pay and AI; UC San Diego finds AI portal drafts slow complex replies; physician pay outpaces productivity; CPT panel weighs AI codes
THE SHORT READ
About 150 physicians at two Allina Health hospitals north of Minneapolis walked out Sept. 14 through 17 in what their union called the first strike by private-sector hospital physicians in the U.S. Among the union's stated demands was a say in what the hospitals' AI systems are allowed to suggest. A SullivanCotter survey of about 235,000 employed physicians, released Sept. 17, showed pay rising faster than productivity, with radiology pay up 12% to 18% since 2024. And a year of data from UC San Diego Health, published in NEJM AI, found that edits requiring clinical judgment, such as interpreting a radiology result, add 60% to 70% to the time a physician spends on an AI-drafted portal reply.
The Current Procedural Terminology (CPT) Editorial Panel of the American Medical Association (AMA) met Sept. 17 to 19 in Minneapolis on a 94-item agenda that included seven proposals for codes describing AI services. The Consumer Technology Association (CTA), the consumer electronics trade group, asked Congress in a Fierce Healthcare op-ed for a single national health-AI framework in place of the more than 240 state bills it counted this year. California Gov. Gavin Newsom issued an executive order on frontier models on Sept. 18 that does not address the four health bills on his desk; those face a Sept. 30 deadline. STAT published an interview with Dr. James Deardorff on why an accurate model can still be used badly; Deardorff, a geriatrician at the University of California, San Francisco, wrote the commentary on Epic's mortality score covered in Thursday's post.
THE DETAILS
- Workforce Sept 17, 2026Doctors Council SEIU, which represents hospitalists, OB-GYNs, psychiatrists and critical care physicians at Allina Health's Mercy campus in Coon Rapids and Unity campus in Fridley, filed its strike notice Sept. 5 and walked out Sept. 14 through 17, after three years without a first contract, in what it called the first strike by private-sector hospital physicians in the U.S. Its stated reasons were sick leave it said state law requires, an offer it described as pay cuts and more input into patient care decisions; the Minnesota Reformer counted about 73 physicians on the picket line, 37 who crossed and 40 not scheduled that week. Dr. John Wust, an OB-GYN on the bargaining team, told MPR News: "We have already seen AI starting to suggest diagnostic codes for patients, and we're concerned that it may start to suggest treatments." CBS Minnesota described the demand more broadly, as physicians seeking a voice in AI that makes billing and treatment suggestions without their input. Allina said the strikers were about 10% of the physicians at the two campuses and that both hospitals would run normally.
- Evidence Sept 17, 2026The study, published in NEJM AI, examined a year of Epic-integrated draft replies to patient portal messages across specialties at UC San Diego Health. Researchers sorted every physician edit into a 15-category taxonomy built with language models and expert review and used response time as the measure of workload. Scheduling changes were the most common edit, appearing in 38.5% of edited messages, followed by lifestyle and non-drug advice in 18.2% and added empathy in 16.1%; stopping or tapering a medication was the rarest edit at 2.4%. Clinical edits added the most time per message: interpreting a radiology result added 70.1% to the time spent on a message, clarifying or ruling out a diagnosis added 63.9% and interpreting laboratory results added 60.8%. The authors concluded that the burden is not uniform, with judgment edits costing the most per message and high-volume administrative edits costing the most across the system, according to reports in Becker's Physician Leadership and Healthcare Innovation; the paper is behind NEJM AI's paywall, and the figures come from those two reports.
- Workforce Sept 17, 2026SullivanCotter's 2026 Physician Compensation and Productivity Survey, released Sept. 17, covers 575 organizations, about 235,000 physicians and 240 specialties. It reports that median total cash compensation rose in every major specialty group and rose faster than productivity. Work relative value units grew 0.8% in primary care and 2.8% in adult medical specialties, against a 7.2% pay gain in adult medical specialties and a 25.2% gain over five years, the firm said. Anesthesiology pay is up 14% to 16% since 2024 and pay for certified registered nurse anesthetists nearly 12%; radiology, the specialty with the most algorithms cleared by the Food and Drug Administration, is up 12% to 18% across diagnostic, interventional and mammography. Nearly 95% of new hires received a sign-on bonus, clinical outcome measures now appear in 55% of incentive plans and nearly 65% of the organizations plan to expand their physician workforce next year, according to the release, whose only reference to AI is a managing director's statement that shortages and AI tools together mean compensation plans need to be "agile." The survey is based on employer-reported pay data and is the third dataset this year, after Doximity and the Medical Group Management Association, to show physician pay continuing to rise.
- Money Sept 19, 2026The panel met Sept. 17 to 19, the last meeting of the cycle that produces the 2028 code set, with 94 tabs on the agenda: 49 Category I proposals, 40 Category III proposals and a handful of guideline changes. Seven of the 94 describe AI services, four of them seeking permanent codes: quantitative MRI software for liver tissue composition (Tab 36), augmentative arrhythmia analysis from computational cardiac simulations (Tab 46), noninvasive measurement of left ventricular end-diastolic pressure to detect heart failure (Tab 48) and electrical impedance spectroscopy of skin lesions (Tab 49). The other three seek Category III codes: augmentative algorithmic analysis of digitized prostate slides (Tab 68), augmentative breast cancer risk prediction from screening mammograms (Tab 82) and noninvasive identification of ST-elevation myocardial infarction (STEMI) and its equivalents (Tab 84). Tab 94 would rewrite the pathology and laboratory code application to ask about software and its validation, which Madhani Healthcare Consulting and the Discoveries in Health Policy blog both flagged as the AI change that will outlast any single code. The AMA's Appendix S taxonomy, updated the day the meeting opened, supplies the vocabulary: assistive software provides data, augmentative software provides a score or parameter and autonomous software makes the interpretation itself. Three of the seven proposals are labeled augmentative in the agenda, and none is labeled autonomous. The votes are not yet public; the panel publishes a summary of its actions weeks after it meets. Whether Medicare pays for any code the panel accepts would be decided in the 2028 fee schedule.
- Regulation Sept 18, 2026Kinsey Fabrizio, the association's president and CEO, wrote in Fierce Healthcare on Sept. 18 that more than 240 health AI bills have been introduced across 43 states this year, that at least 20 state privacy laws now govern health data and that the patchwork will delay tools for the 77 million Americans in primary care shortage areas. The op-ed offers the association's own consensus standard for post-market AI monitoring, written with more than 75 organizations, as the model for a national rule, and it cites the association's own survey figures: 41% of consumers using AI for health information and 33% of users of web-based health assistants reporting satisfaction. The bill count and the survey figures are the trade group's own. The same day, Newsom's executive order directed California agencies to speed up SB 813 and AB 1405, the state's frontier-model oversight laws, and to convene experts within two months on an AI kill switch; the order does not mention health care or the four health bills on the governor's desk, which face a Sept. 30 deadline. The op-ed joins preemption arguments made this year by the Justice Department in Colorado and in the draft Great American AI Act.