The signs arrive quietly and out of order. Charts closing later than they used to. Two staff members who have started routing certain questions to a different physician without being asked to. A colleague who is somehow always the last to arrive on Mondays. Nobody says anything out loud, because the partners are friends and because the first person to say it becomes responsible for what happens next. That silence is the reason impairment provisions exist.
Physician impairment—from substance use, from psychiatric illness, from the cognitive changes that arrive with age or with disease—is not a hypothetical risk in a partnership with a 30-year horizon. It is a scheduled one. And the reflex that makes a partnership worth joining, which is loyalty, is precisely the reflex that produces delay when a partner is unwell. Loyalty without structure looks like six months of covering for someone. Meanwhile children are being seen, and every day of drift widens the practice’s exposure and narrows the colleague’s options for a confidential, non-disciplinary path back.
A working provision decides seven things before it is ever needed. It defines impairment broadly enough to cover substance use, mental health conditions, and cognitive decline, and neutrally enough that it never reads as an accusation of misconduct. It names who may initiate—a stated number of partners, or the managing partner with the concurrence of one other—so no individual carries the act alone. It states what initiation triggers: a confidential referral to the state physician health program, an independent fitness-for-duty evaluation, or both, by an evaluator chosen from a pre-agreed process rather than selected in the heat of the moment by whichever side moves first. It specifies interim measures that pause rather than punish, including a schedule modification or paid administrative leave and the compensation treatment that attaches to it. It imposes confidentiality on every partner who touches the process. It defines return to work: the evaluator’s clearance, a monitoring agreement, any practice-level conditions and their duration, and what a relapse means. And it states how this process relates to the disability trigger of article 33 and the for-cause machinery of article 42—because a partner routed into treatment is not being expelled, and an agreement that cannot tell those two paths apart will produce the wrong one.
Most states operate a physician health program offering confidential evaluation, referral, and monitoring, frequently with an alternative-to-discipline pathway available when a physician self-refers and complies with the program’s terms. Structure, confidentiality protections, funding, and the program’s relationship with the state medical board vary substantially from state to state. The practice should learn its own program’s specifics from the program itself and from counsel—the intake process, what triggers a report to the board and what does not, and how a monitoring agreement interacts with credentialing and payer enrollment—and then reference the program by function rather than by a name that may change.
The legal terrain around this provision is intricate, and it is named here rather than solved. The Americans with Disabilities Act and its state analogues govern when a medical examination may be required, what makes an inquiry job-related and consistent with business necessity, and how medical information must be segregated and kept. Whether a physician-owner is an employee for that purpose is itself a contested question that turns on control rather than on title, the analysis the Supreme Court set out in Clackamas Gastroenterology Associates v. Wells. State peer-review statutes may protect the process and its records, but generally only where it is conducted as peer review, with the structure that implies. Many states impose an affirmative duty to report a colleague reasonably believed to be practicing unsafely, and those duties do not lapse because the partners would prefer to handle it internally. Reporting obligations to the National Practitioner Data Bank follow their own rules. Every one of those questions is answered by healthcare and employment counsel in the practice’s own state, before the policy is drafted—never during the week it is first needed.
One structural point decides whether the whole provision survives contact with reality. Put it in the partnership agreement, or in a policy the agreement incorporates and protects with the same amendment threshold as any other governance provision. A process that lives only in a policy manual can be edited by whoever controls the manual, which means the provision most likely to be aimed at a partner is editable by the partners aiming it—or by the partner it is aimed at, if he happens to be the managing partner. That is not a hypothetical failure mode. It is the ordinary one.
Finish with the money, because a process nobody can afford to use does not get used. Confirm that disability coverage and any salary-continuation arrangement actually respond to a treatment leave, that call coverage during an absence is assigned rather than volunteered, and that the compensation formula does not punish a partner so severely for a 90-day absence that treatment becomes financially impossible. A provision that makes recovery unaffordable has quietly chosen the outcome it was written to prevent.
Red flags. No defined impairment process, so the partners improvise or avoid. No fitness-for-duty provision, leaving the practice with nothing between concern and expulsion. No confidentiality obligation, which guarantees the process is never started. No return-to-work criteria, so a recovering partner’s reinstatement becomes a negotiation instead of a standard. And a process defined only in a manual that can be changed by simple majority on a Tuesday.
The kindest provision in a partnership agreement is the one written years before anyone needed it, by people who did not yet know which of them it would protect.
The provision names the initiator in advance, so no individual carries the act alone. A stated number of partners, or the managing partner with the concurrence of one other, is the common construction. Initiation triggers a confidential referral to the state physician health program, an independent fitness-for-duty evaluation, or both, with the evaluator chosen from a pre-agreed process rather than in the moment.
No. A working provision states how the impairment process relates to the agreement’s disability trigger and its for-cause machinery, because a partner routed into treatment is not being expelled. Interim measures pause rather than punish, including a schedule modification or paid administrative leave with the compensation treatment stated, and return to work runs on evaluator clearance and any monitoring agreement.
In the partnership agreement, or in a policy the agreement incorporates and protects with the same amendment threshold as any other governance provision, since a process living only in a manual is editable by whoever controls the manual. The Americans with Disabilities Act, state peer-review statutes, and state reporting duties are answered by healthcare and employment counsel in the practice’s own state.
The Partnership Agreement Analyzer scores an existing agreement against the framework this series is built on — or schedule a discovery call to work through it with PMI. The full framework, with the arithmetic, lives in the textbook Pediatric Practice Management: The Fundamentals.