Hospital spending on contract staff is almost back to its pre-pandemic level. The contracts themselves never came back.<br><br>
That’s the quiet finding in our look at how contract staff normalization masls employment risks staff normalization masks physician employment risks</a>. Maine’s hospitals can point to a spending line that looks like 2019, and fair enough, it does. But crisis-era locum deals didn’t vanish. They settled into long-term outsourcing agreements at lower rates. The per-hour cost came down. What the physician gave up (retirement contributions, tail coverage, guaranteed hours) mostly stayed gone. A fixed labor cost became a variable one, and somebody has to hold the variability. It is rarely the hospital.<br><br>
Most of what moved in workforce news this week moved in that same direction. Pay held up fine. Risk traveled, from institutions and federal policy down to the individual physician, resident, or first-year medical student, which is usually the person holding the pen.<br><br>
Start with the contract. A new federal bill would void most physician noncompetes, several states already have, and the retention work those clauses used to do is migrating somewhere. Mostly it’s going to clawbacks: signing bonus and relocation repayments that can reach six figures, sometimes tied to productivity targets set in a busier year. Add termination without cause on 90 days’ notice or less. Add tail coverage, which can run two to three times an annual premium and is increasingly offered at the physician’s expense. None of this is hidden, exactly. It’s on page seven. (Page seven has always been where the good stuff lives.)<br><br>
Then liability. Health systems bought ambient AI to give physicians their evenings back, and in fairness some evenings came back. But clinicians in a UC San Diego study edited roughly 70% of AI-drafted messages before sending them. As our piece on AI review time as uncompensated labor review time as uncompensated labor</a> points out, the physician who signs the note owns every error in it, including the inflated complexity language she didn’t write. The review generates no RVUs. The malpractice exposure is total. The efficiency lands with the institution, and the audit lands on the signature line.<br><br>
Then debt, the version of this that starts before anyone has a contract to read. The 2025 reconciliation law capped federal borrowing for medical students at $50,000 a year and $200,000 overall and closed PLUS loans to new borrowers. At the Kansas College of Osteopathic Medicine, tuition runs about $15,000 a year above the new limit. Students are covering the gap with private loans, credit checks and co-signers included. This is for a profession whose graduates have among the lowest federal default rates of anyone. You’d expect a borrower that reliable to get better terms. You’d be thinking like a lender, which is apparently no longer the assignment.<br><br>
And then the risk nobody even invoices. States keep funding medical schools, but the Medicare-backed residency slots that actually turn students into practicing physicians are still pegged to 1996. Our look at the GME bottleneck undermining medical school expansion</a> found Georgia sent 65% of its matched graduates out of state this year, and about a quarter of those wanted to stay. Georgia paid for the training. Another state gets the doctor. Someone in rural Georgia gets the drive.<br><br>
The pattern is consistent enough to stop calling it a coincidence. Institutions have been buying flexibility: variable labor costs, AI throughput, easy exits, the option not to fund something. Flexibility for one party is exposure for the other. And unlike most workforce stories, nothing here is accidental or unmeasured. It’s written down clause by clause, and someone signed it.<br><br>
Which leaves physicians in an odd spot. They have more market power than they’ve had in a decade, and the terms that matter most don’t show up in the salary line. A signing bonus is easy to negotiate. A prorated forgiveness schedule, an AI-review carve-out from the productivity denominator, a tail clause that says who pays: those require a reader who makes it to page seven.<br><br>
Somewhere in Kansas, a first-year medical student is filling out a private loan application with a line for a co-signer. Her first employment contract is about seven years away.<br><br>
P.S. PhysEmp is an AI-powered job board for physicians, which this week mostly means we’d like you to find the job before you find the clause. Physicians can browse roles and see how specialties and markets are actually moving, and recruiters can reach the physicians who fit instead of everyone with a license. Fewer surprises on page seven is a perfectly reasonable goal.<br><br>
All sources are analyzed and curated from PhysEmp’s industry alert network. AI assists with synthesis and pattern recognition; editorial judgment stays with the PhysEmp Editorial Team. How we make this newsletter →