Record of Care · Episode 5 of 7 · Medical fiction

The Second Signature

A colleague asks Rowan to sign. First, they must agree what the signature will mean.

  • Medical fiction · Episode 5 of 7
  • With notes: The Second Signature

Published · revised as medical fiction

8 min story · 14 min notes · sources in the notes checked through (Pacific)

Conceptual illustration: a note with two signature lines, one signed and one blank beneath an outlined button.
Conceptual illustration. No real patient, clinician, screen or record is depicted.

The note was good. Lena could tell that by the second paragraph.

Mr. Harlan Pike, sixty-eight, a retired harbor pilot with blood pressure that had ignored one pill for three months. Home readings recorded. Kidney function from the spring. A clear assessment and a plan that moved from one thing to the next without hurrying. Eli had signed it on Monday. The drafting tool had helped write it, and it read better than most of what Lena produced on her own.

At the bottom of the screen sat the button.

In the story

Reviewed and agree.

“The pharmacy called about what?” she asked.

“He wants to know if he’s supposed to be cutting them in half.” Eli had not sat down. “I haven’t called him back yet. I wanted this closed first so there’s one version of the chart when I do.”

“Cutting what in half?”

“The lisinopril. I went up to forty.”

Lena read the plan again.

In the story

Continue lisinopril 20 mg daily. Recheck labs in two weeks.

She opened the orders. Lisinopril, 40 milligrams, one tablet daily, sent to the pharmacy Monday at 2:15.

“Your note says continue twenty.”

“It doesn’t.” He came around the desk and looked. She watched him find it. His face stayed composed, but it cost him something. “I said forty. I told him forty. We had a whole conversation about it, because he asked if that meant he’d been on the wrong dose, and I said no, you’ve been on twenty, twenty was the right place to start—”

He heard it as he said it.

“He said, so I stay on twenty,” Eli said slowly. “And I said, you’ve been on twenty. We’re going to forty.”

“And the note kept the first half.”

“I read that note.”

“I’m sure you did.”

“I did read it, Lena.” He stepped back. “Is this what the queue’s for now? I’ve been doing this nine years. I flag a chart because the agreement says dose increases in his age group route to you, and you go through it looking for—”

“Eli. What does his after-visit summary say?”

He stopped.

She had it open already. The page printed for Mr. Pike on Monday, the one on his refrigerator or in his glove box, said twenty.

“So he’s got a bottle that says forty and a paper that says twenty,” Lena said. “And he’s a careful man, and he’s cutting them in half. That’s what the pharmacy’s calling about. Can we call him before we do anything about that button?”

Whatever Eli had been about to say, he did not say it. He took out his phone.


Mr. Pike answered on the second ring. Eli put him on speaker and said who was in the room.

He had been cutting them in half since Tuesday. He had a pill splitter; his late wife had used it. The paper said twenty and the bottle said forty. In forty years of bringing ships into the harbor he had learned that when two instruments disagreed, you went with the more conservative one until someone told you otherwise. His pressure that morning had been 162 over 94. He had written it down.

“Which one’s right?” he asked.

“The bottle,” Eli said. “That’s my mistake, Mr. Pike. The paper’s wrong. I meant forty, I ordered forty, and the summary didn’t get it right and I didn’t catch it. One whole tablet, once a day, starting tomorrow morning.”

“Not tonight?”

Eli looked at Lena. It was a small look, and it was the first time that afternoon he had asked her anything.

“You’ve had twenty today,” she said toward the phone. “Tomorrow morning is right. This is Dr. Rowan, I work with Eli.”

“And the blood test?”

“Two weeks from tomorrow,” Eli said. “I’m going to move it so it counts from the real dose. Somebody will call you with the day. And I’m sending you a new paper through the portal tonight, and a printed one in the mail, because I know you like to have the paper.”

“I do like to have the paper,” said Mr. Pike.

When the call ended Eli sat down at last, in the chair patients used.

“Thank you for not doing that part first,” he said. “The chart part.”

“I nearly did.”

“I know. I was ready for it.” He rubbed his eyes. “I thought you were checking my work.”

“I was checking the chart against the orders. That’s all I can check. I wasn’t in the room.” She turned the monitor so they could both see the button. “That’s my problem with this. It says reviewed and agree. Agree with what? I agree with forty. I think forty’s right. I’d have done the same. I don’t agree with the note, because the note said twenty until a minute ago. And it doesn’t say what I did. I read a chart and found two numbers that didn’t match, and then you fixed it.”

“The agreement says you review dose increases in that group.”

“What does it say review means?”

Eli opened his mouth, and then he laughed, briefly, for the first time since he had knocked.

“I signed it four years ago. I’d have to look.”

They looked. It took a while to find. The practice agreement lived in a credentialing folder neither of them had opened since the day they signed it. It was six pages. It said that the supervising physician would be available for consultation by telephone or electronic means. It listed the situations that routed a chart, and it said the physician would review the record and discuss with the PA as needed. It did not say examine. It did not say agree.

“So the button’s asking for more than the paper does,” Eli said.

“Or less. It depends who reads it in a year.” Lena opened the free-text field beneath the button. In four years she had typed into it perhaps twice. “I’m not going to refuse to sign your chart. That leaves it sitting in the queue, and that’s worse. I’m going to say what I did.”

She typed it and read it aloud as she went.

In the story

Record review at the treating PA’s request under our practice agreement. Narrative and after-visit summary stated lisinopril 20 mg; order was for 40 mg. Discussed with E. Navarro, PA, who confirmed 40 mg as the intended dose. Patient contacted by the PA today and instructed; labs rescheduled. Note addended by the PA. I did not examine the patient.

“That last line makes it sound like you’re covering yourself,” Eli said.

“Does it?”

He thought about it honestly.

“No. It makes it sound like what happened.” He pulled his laptop onto his knee and began his own addendum. “If Pike had a problem in six months and somebody pulled this, I’d rather it said that than agree.”

“It doesn’t make it not mine,” Lena said. “I want to be clear about that. If the agreement says I review these, I review them. Writing a careful sentence doesn’t get me out of anything.”

“I didn’t think it did.”

“I’m saying it for me.”


Ava Chen came to the door while they were both typing, a tea in one hand, plainly on her way somewhere else.

“Are you two doing queue? Because I’ve got one sitting in yours, Lena, and I want to be sure you’re not about to supervise me.”

“I’m not?”

“You are not.” Ava leaned on the frame. She was a nurse practitioner, eight years at Marsh Road and two at Pier Street. Lena had never been asked to oversee her practice and had never needed to read the documents that defined it. “Mine’s the quarterly quality sample. You’re my assigned reviewer this quarter, like Sam was last quarter. Five charts, the rubric, done. It’s got nothing to do with whether I can adjust a medication. It lands in the same list as Eli’s and it has the same button.”

Lena found the item. It was indistinguishable from the one she had just closed.

“If I’d clicked agree on that—”

“Then a year from now somebody reads it as you signing off on my patient. You didn’t see her, I didn’t ask you about her, and you were never supposed to be responsible for her.” Ava shrugged. “I brought it up at the regional association meeting in the spring. Half the room had the same queue. I said the button was going to get somebody hurt or somebody blamed.”

“You gave a talk?”

“Ten minutes at an open session. I didn’t name us.” She sipped the tea. “I did wonder, when my sample got pulled early this quarter. For about a day. Then I asked Noor, and the sample schedule was set in the first week of the quarter. My talk was in the fifth. So unless Julian has a time machine, it’s a coincidence.” She said it lightly, but she had checked. “I’d still like the button fixed.”

“So would I,” Eli said, without looking up.

“Write it down, then. Both of you. Julian’s collecting things for Friday.” Ava straightened. “I’d say three kinds of item, three labels, and a box that makes you say what you did. But nobody’s asked me.”

“I’m asking you,” Lena said. “Send it to me. I’ll bring it tomorrow with your name on it.”

Ava looked at her for a moment, as if adjusting an estimate.

“All right,” she said.

When she had gone, Eli finished his addendum and closed the laptop.

“Every one of these takes this long?”

“This one took twenty minutes, and it had a patient cutting pills in half. Most of them are fine and take two.” Lena looked at the clock. It was 4:52. The discharge summary was still open on her other screen, with the blood thinner and its two doses. “Celia wanted to put a visit in this slot.”

“Somebody should tell her what’s in it.”

“I did. Today, actually.” She pulled the summary back to the front.

A message arrived as she did, at the top of the inbox, where the routing rule now left only the ones meant for her.

It was from Mara Ellison. It was four lines long.

In the story

Thank you for the appointment. I have read your letter. I have also read my record. It says in three places that I refused. I would like to know how to put down what I actually said.

The questions behind this episode

Clinical and governance notesThe Second SignatureWhen AI writes the note, the next clinician must know what the next signature means.

Everything below is commentary on real clinical, technical and legal questions. It refers to the story only as an illustration. Characters named here are fictional.

Evidence at a glance

Established by the cited sources
California’s physician-assistant framework makes supervision substantive under a practice agreement without a universal countersignature rule; California recognizes nurse-practitioner practice under standardized procedures and in the 103 and 104 categories; HRSA requires periodic quality assessment while leaving methodology to the health center.
Not established, and not claimed
The terms of any real practice agreement, any individual clinician’s certification status, or any finding that a real clinician practiced beyond competence.
The analysis in these notes
Record review, consultation, PA supervision, NP collaboration and periodic quality review are different activities. Each signature should truthfully describe the activity actually performed.

A button in the episode reads Reviewed and agree. The difficulty is deciding what a click on it says about the person who clicks. That the record was read? That the management was discussed? That a required supervisory review was completed? That the reviewer examined the patient? That the clinical work described in the note was independently verified? The button offers less space than the question deserves.

Three kinds of item reach the same queue in the story: a chart routed under a physician assistant’s practice agreement, a nurse practitioner’s quarterly quality sample, and requests for consultation. The interface makes them look alike. Their purposes are not interchangeable, and the notes below take them one at a time.

Begin with the claim

A signature needs a subject. It must authenticate something, confirm something, or record an action by someone. Without that connection, the institution may count a completed click while the people involved understand different things.

The treating clinician might understand the second signature as acknowledgment that a colleague is available for consultation. The physician might understand it as a review of the written plan. An administrator might understand it as evidence that a supervisory requirement has been fulfilled.

Those interpretations can all feel reasonable until a material discrepancy appears. Then the institution needs to know which review actually occurred, which information was available, and what response followed.

The first governance task is therefore to identify the activity before refining the wording. An elegant attestation cannot repair a process that never defined its purpose.

Rowan asks the clinic to distinguish record review, patient-specific consultation, examination, supervisory concurrence, and periodic quality assessment. This does not require a separate essay for every note. It requires the workflow to identify the correct category and the reviewer to make an accurate statement about the work performed.

The same restraint applies to the word verified. A clinician may verify an order against an order record. That does not mean the clinician independently verified every sentence describing the patient encounter.

Fluency is a poor test of judgment

Rowan’s first impression is that the note is unusually polished. That observation provides very little information about the treating clinician’s judgment.

A clinician can carefully review an accurate AI draft, reject inappropriate suggestions elsewhere in the workflow, confirm the plan, and leave the remaining wording unchanged. Another can make many cosmetic edits while overlooking an important contradiction. The amount of editing does not resolve the question.

This matters for professional respect as well as accuracy. A colleague should not be accused of thoughtless adoption because a note sounds polished. Nor should awkward phrasing be treated as proof that a human thought more deeply about the patient.

The useful review begins with clinical substance. Are the findings supported by the available record? Does the assessment fit the stated evidence? Is a possibility presented as a confirmed condition? Does the plan match the orders and instructions? Is an important uncertainty acknowledged?

In the fictional encounter, Rowan finds a concrete discrepancy: the medication instructions in the narrative conflict with the available order record. That finding justifies clarification. It does not establish whether the discrepancy originated in speech, software, editing, or transfer between systems.

The treating clinician explains the intended plan. The conflicting information can then be reconciled through the appropriate process, and any necessary patient communication can be assigned. The review has added something useful without pretending to reconstruct an encounter from writing style.

This is the standard AI should encourage: less confidence in literary clues, more attention to evidence and clinical reasoning.

The PA agreement is a working document

California’s physician-assistant framework makes supervision substantive. Business and Professions Code sections 3501 and 3502 connect PA services with physician supervision, an applicable practice agreement, competence, and preparation. Supervision includes agreed oversight and availability through telephone or other electronic communication; it does not generally require physical presence.[1]

Section 3502(c) does not establish a universal requirement that a physician countersign every PA record. It preserves requirements in the practice agreement and specified disciplinary conditions. Absence of a universal countersignature rule does not mean absence of supervisory responsibility.[1]

The Physician Assistant Board identifies the agreement’s required subjects: authorized services, adequate supervision, continuing evaluation of competence, drugs or devices, and additional agreed provisions. Section 3502.3 expressly addresses communication, availability, consultation, and referral within the supervision provisions.[1, 2]

For Rowan, the immediate task is to obtain and understand the agreement governing the actual relationship. A generic assurance that the author is responsible for the note answers only part of the question. What consultation is expected? When must it occur? How are concerns resolved? Which reviews are required, and by whom?

California’s PA regulations also require consultation or appropriate referral when a task or diagnostic problem exceeds the PA’s competence, and preserve ongoing supervisory duties.[1]

An AI-generated plan cannot expand that competence. Equally, use of AI does not establish that a PA lacks it. The relevant inquiry concerns the individual, the service, the required support, and the work actually performed.

The agreement should be usable during a difficult clinical afternoon. If it exists only as a signed document that nobody consults, its language and the clinic’s actual practice can drift apart.

Nurse practitioners require a separate analysis

The next item in Rowan’s queue comes from a nurse practitioner. It should not be processed by simply substituting NP for PA in the preceding explanation.

California retains traditional NP practice under standardized procedures. It also recognizes the categories commonly called 103 and 104 NPs. A qualified 103 NP may practice without standardized procedures in specified group settings with physicians; the 104 framework permits practice outside those settings within the population focus of national certification when its requirements are satisfied.[3]

The Board of Registered Nursing also states that an employer may require a 103 NP to continue using standardized procedures. The actual certification, setting, authorized activities, and applicable organizational requirements therefore matter.[3]

These distinctions are easily lost in a queue labeled provider review. A label does not determine the legal relationship. Nor does a job title establish which certification pathway an individual holds.

The clinic should verify the relevant status and documents before assigning responsibilities. It should identify whether Rowan is acting as a required supervisor, a consultant, a reviewer under organizational policy, or a colleague participating in quality improvement.

That clarity benefits the NP as well as the physician. It prevents the organization from implying an authority relationship that does not apply, while preserving requirements that do apply. Independent authority and access to consultation can coexist; neither is accurately represented by a vague request to approve everything.

AI should not flatten professional distinctions that law, credentialing, and clinical practice deliberately preserve.

Periodic quality review is another activity

A later message asks Rowan to complete a periodic review sample. This is a third task, even if the EHR presents another familiar button.

HRSA’s Health Center Program framework requires systematic quality assessments at least quarterly and gives health centers discretion over methodology and report focus. The chapter does not set a single chart-sample size for every center. Its assessment requirements are not identical to an encounter-level supervisory concurrence.[4]

A periodic reviewer may evaluate trends, documentation quality, appropriateness of care, and the need for improvement. The review does not retroactively establish that consultation occurred during a particular visit. It also does not replace a timely response when a current clinical concern is identified.

Rowan’s clinic should therefore avoid using the completion of one process as evidence that another was performed. A quality report can document what was reviewed, the criteria used, the findings supported, and the action recommended. A supervisory entry should describe the relevant supervisory activity.

This distinction is not administrative fussiness. It prevents an institution from asking one signature to carry several different accounts of care.

Make the attestation describe an action

An attestation becomes more useful when its verbs are specific.

Read, discussed, examined, reconciled, agreed, corrected, and referred describe different actions. A template that substitutes one for another creates ambiguity even if the signer acts in good faith.

In a fictional example, “I reviewed the available encounter documentation and discussed the conflicting medication instructions with the treating clinician” describes more than a bare concur. It also describes less than a personal examination or verification of the original conversation.

That sentence is an illustration, not a universal attestation to copy into charts. The wording must fit the actual work and the applicable agreement or policy. If a required activity has not occurred, the solution is to complete or appropriately escalate it, not select narrower language that conceals the gap.

The useful design principle is that a reader should be able to understand what the second clinician contributed. If the review identified a material discrepancy, the record should show how it was addressed. If a consultation altered management, that contribution should be documented accurately.

The treating clinician’s authentication remains important. It establishes that the clinician adopted the final documentation through the relevant signing process. But it does not tell a later reader everything about how the draft was generated or which parts another person reviewed.

A second signature should add a distinct, truthful statement. Repeating a broad assurance may create apparent certainty without adding information.

For the organization, a small set of clearly defined review pathways may be more practical than one expansive attestation used everywhere. The design should make the correct action easy to choose and make exceptions possible when the ordinary pathway does not fit.

Five pathways, five statements

The distinctions above can be reduced to a working table. Its contents are general; the applicable agreement, standardized procedures, organizational policy and law govern any actual relationship.

PathwayWhat prompts itWhat the reviewer doesWhat the entry can truthfully sayWhat it does not certify
Record review under organizational policyA policy-defined sample or routing ruleReads the documentation against stated criteriaThat the record was reviewed, and any discrepancy foundThat the reviewer observed the encounter or examined the patient
Patient-specific consultationA question from the treating clinicianAdvises on a particular patient’s managementWhat was asked, what was advised and whenResponsibility for parts of care that were not discussed
PA supervision under a practice agreementThe terms of the applicable agreementThe supervisory activity the agreement specifiesThe specific supervisory activity performedSupervisory work the agreement requires but that has not yet occurred
Consultation or collaboration with an NPThe NP’s practice framework and the setting’s policiesConsultation, standardized-procedure functions or review, as applicableThe activity actually performed under the applicable frameworkA supervisory relationship the framework does not establish
Periodic quality reviewA quality program’s schedule and methodologyEvaluates sampled records for patterns and improvementThe criteria applied and the findings supportedThat any particular consultation occurred at the time of care

The fourth row is deliberately general. California’s nurse-practitioner frameworks differ, and the correct description depends on the individual’s certification, the setting and any requirements the employer imposes. The purpose of the table is to keep one button from carrying all five meanings.

A clinic that adopts something like this table gains an immediate practical benefit. When a material discrepancy appears months later, the record will already say which kind of review occurred, and no one will have to reconstruct the meaning of a click from memory.

Review requires something to review

There are limits to what a later clinician can establish from a final note.

Rowan can compare the narrative with available medications, orders, results, and messages. Rowan can ask the treating clinician to explain a decision. If authorized source material or version history is available, it may help investigate a specific discrepancy.

But the final document alone may not establish whether a statement was spoken, inferred by software, introduced during editing, or altered during transfer. A signature does not supply missing evidence.

That limitation should narrow the conclusion, not automatically end the review. If the note contains contradictory instructions, the contradiction can be identified even when its origin is uncertain. If an action is documented as completed but remains pending, the current care need still requires attention.

Conversely, uncertainty about origin should not be converted into certainty about blame. The reviewer can distinguish the established discrepancy, the patient’s present needs, the correction required, and the unanswered question about how it arose.

For material concerns, the organization should have a proportionate way to preserve and examine relevant information consistent with privacy, access rules, and applicable retention obligations. That is a proposed governance safeguard, not a claim that every encounter must be retained as audio indefinitely.

The reviewer should not export sensitive material into a personal collection to compensate for a poorly designed process. Access and investigation should occur through authorized channels that protect the patient and preserve the reliability of the evidence.

Competence is more than a well-written plan

AI makes it especially important to distinguish the quality of the prose from the competence of the person providing care.

A polished assessment can organize a sound clinical decision. It can also obscure a missing step if nobody asks how the conclusion was reached. The appropriate response is an assessment of competence that uses more than the finished paragraph.

HRSA’s staffing framework treats credentialing and privileging as distinct activities and recognizes peer review or comparable methods in evaluating continuing competence. A license, an approved privilege, and a particular review result answer related but different questions.[5]

For the story’s clinic, useful education could include discussion of why a diagnosis was supported, what would change management, when consultation is needed, and how the clinician checks generated content against the encounter. The purpose would be to improve reasoning and tool use, not reward the ability to imitate a preferred writing style.

Training should also address correction. A clinician needs to know how to remove an unsupported suggestion, preserve accurate content, reconcile orders, and report a material system problem. Knowing the medicine and knowing the local software are different competencies; a safe workflow needs both.

An experienced clinician may need instruction on a new integration. A newer clinician may need clinical supervision even when the software behaves exactly as intended. The response should match the identified need.

The most useful question is not whether the author touched every sentence. It is whether the author understood and verified the clinically consequential content, acted within the applicable role, and sought help when needed.

The time belongs in the plan

A meaningful review requires attention, and attention requires room in the working day.

If the clinic assigns Rowan responsibility for reviewing clinical decisions, resolving discrepancies, and being available for consultation, it should specify how that work fits alongside patient appointments and other duties. Counting completed signatures does not reveal whether the expected review was feasible.

The answer need not be maximal review of every detail. The process can distinguish routine cases from unfamiliar care, high-consequence discrepancies, repeated documentation problems, or matters that exceed the treating clinician’s competence. Applicable agreements and requirements still control; convenience cannot silently rewrite them.

Proportionate review is a design choice that should be explicit. It identifies what deserves immediate attention, what can be sampled, what requires discussion, and what should reach clinical leadership.

It also requires a route for disagreement. A clinician who cannot support a proposed attestation should have a way to explain the concern and obtain a resolution. An indefinitely parked chart is not a satisfactory substitute for that process, particularly when the patient’s care remains unfinished.

The organization should learn from the pattern. If reviewers repeatedly encounter the same mismatch between generated narrative and workflow, the answer may involve training, template changes, integration repair, or a revised assignment of work. Requiring additional clicks alone may leave the problem intact.

When the reviewer cannot sign

There will be occasions when Rowan cannot truthfully complete the requested entry: the discrepancy remains unresolved, the treating clinician is unavailable, or the entry would certify something Rowan has not done. The process should anticipate that moment rather than leave the chart in limbo.

A sound sequence is straightforward. First, address any present patient-safety need, whatever happens to the paperwork. Second, record the review activity that did occur, in accurate terms. Third, identify the unresolved discrepancy and the person responsible for resolving it. Fourth, route a disagreement the clinicians cannot resolve to a designated clinical leader within a stated time. Fifth, close the loop: the final entry should show how the matter was resolved and by whom.

None of these steps permits a reviewer to refuse a legitimate obligation. They require the organization to recognize that a truthful partial statement, followed by a resolution, is worth more than a complete-looking signature describing work no one performed.

The volume of such exceptions is itself information. If reviewers regularly cannot sign, the problem may lie in the template, the routing, the time allotted or the generated documentation rather than in the reviewers.

The strongest objections deserve an answer

An administrator may reasonably ask whether this approach makes every note a formal investigation. It should not. Most encounters need an ordinary, well-defined workflow. Additional inquiry should follow a meaningful clinical question or a requirement of the actual supervisory arrangement.

A treating clinician may ask why an accurate note needs scrutiny simply because AI helped write it. It does not deserve suspicion on that ground alone. Review should apply clear criteria and examine substance. AI use may change what evidence is available or how an error can arise, but it does not justify assuming incompetence.

A physician may ask whether a carefully limited attestation eliminates responsibility for the underlying care. It does not provide blanket immunity. The person’s real role, actions, applicable law, and agreement remain relevant. Precise language helps describe the work; it cannot make required work optional.

A technology team may say that the final author already accepted the note. That is an important part of the process. It does not answer every question about supervision, consultation, source access, or a recurring integration problem.

And a patient may ask why any of this should matter. The answer should be concrete: because the review should help identify and resolve a concern about care, and because everyone involved should understand who is responsible for the next step.

What the second signature should add

Rowan returns to the fictional encounter after the discrepancy is clarified. The note now accurately states the clarified plan, and the medication instructions have been reconciled. Any necessary patient communication has an identified owner. The supervisory process applicable to this encounter has been completed and documented.

The final entry is less grand than an assurance that everything has been independently verified. It is more useful because it identifies the actual contribution.

That is the opportunity AI creates for clinical governance. A system capable of generating finished-looking prose should prompt institutions to become more exact about unfinished questions, professional roles, and the meaning of review.

The first signature should reflect the treating clinician’s adoption of an accurate record. The second should explain the second clinician’s work. Periodic quality review should evaluate the process on its own terms.

When those activities are distinct and connected, the signatures can support care rather than merely complete the screen.

Sources and scope

Sources checked through 19 November 2026 (Pacific). Each source establishes only what its own scope supports; vendor documentation describes product capabilities, not any organization’s configuration. Source numbering restarts in each essay of the series.

  1. California Physician Assistant Board. Laws and Regulations Relating to the Practice of Physician Assistants. Business and Professions Code §§3501(f),3502; 16 CCR §§1399.540,1399.545.www.pab.ca.gov
  2. California Physician Assistant Board. Frequently Asked Questions for Physician Assistant Applicants, question 8; Business and Professions Code §3502.3.www.pab.ca.gov
  3. California Board of Registered Nursing. Assembly Bill 890 guidance and frequently asked questions; practice categories, standardized procedures, and employer requirements.www.rn.ca.gov
  4. Health Resources and Services Administration. Health Center Program Compliance Manual, Chapter 10: Quality Improvement/Assurance.bphc.hrsa.gov
  5. Health Resources and Services Administration. Chapter 5: Clinical Staffing.bphc.hrsa.gov

Related reporting and resources on this site

About this page. This is Episode 5 of Record of Care, a fictional medical drama in seven episodes, followed by its clinical and governance notes. The story is fiction: its characters, clinic, documents and events are constructed, and nothing in it is evidence about a real person, organization, product or deployment. The notes cite real sources with their limits and label their proposals as proposals. First published 19 November 2026 as part of the essay series Clinical AI, Physician Autonomy, and the Record of Care; revised as a serialized work of medical fiction on 19 November 2026. The earlier edition is retained in the site’s editorial archive. Drafted with AI assistance and published on the author’s instruction. Sources were carried over from the first edition, checked through 19 November 2026 (Pacific). General information, not individualized medical or legal advice. Corrections: editorial standards · methodology.

Medical fiction with clinical and governance notes. First published 19 November 2026; revised as a serialized work of medical fiction on 19 November 2026, on the author’s instruction. Sources in the notes checked through 19 November 2026 (Pacific). Corrections: editorial standards.

Search KPSGill.com

Search the full publication — patient guides, policy, investigations, enforcement records and research.