Published · revised as medical fiction
8 min story · 15 min notes · sources in the notes checked through (Pacific)
Mara Ellison read her chart on Monday night, standing at the kitchen counter with her reading glasses on and her dinner going cold beside the phone.
She had not meant to read all of it. She had sent a message at ten past six that morning from the loading dock at Tern Street Bakery, between the second and third bake. It had been two weeks, and nobody from any rheumatology office had called, and she wanted to know whose number she was supposed to be looking for. No answer had come by evening. So she opened the portal to see whether she had missed one, and the visit summary was there, and she tapped it.
In the story
You declined further evaluation.
She read it twice. Then she found the longer note underneath, the doctor’s note, and read that too.
Rheumatoid arthritis. Not possible. Not one of several things. Her mother’s sister had rheumatoid arthritis and had not been able to button a coat by sixty. Mara had sat in that exam room two weeks ago and heard Dr. Rowan say, carefully, that this was one of the things it could be and there were others, and that blood tests and pictures of her hands and a specialist would sort out which. She had gone home and told her husband it might be nothing. Now she was looking at a screen that said it was not nothing, in the flat way a screen says things.
Patient refused further evaluation.
She took her glasses off.
She had not refused. She had said she could not do the X-ray that week because two of her bakers were out and she was covering the three o’clock start herself. She had said give me two weeks. She had gone straight to the lab and let them take four tubes of blood, and that was not what a person who was refusing did. She had even come back in afterward and asked for a letter about the flour tubs.
Her husband asked from the other room what was wrong. She said nothing was wrong. She would tell him when she understood it.
What she felt, mostly, was embarrassed, and that made her angry. Somebody she had never met would open this and think she was a woman who was told she had a serious disease and walked out.
There was a form on the counter under the fruit bowl. They were refinancing the house, and the lender wanted a life insurance policy. Question eleven asked whether she had ever been diagnosed with any of the following. She had been going to do it that weekend.
She moved the fruit bowl back on top of it.
The letter from Dr. Rowan came on Wednesday. Mara read it on her break, on an upturned crate in the yard behind the ovens.
It was long. It was a good letter, in its way. It said the note she had read was wrong and that it was the doctor’s responsibility. It explained that the clinic used a tool that drafted notes from the conversation, that the doctor had corrected the draft, that a second draft had replaced the first when Mara came back about the work letter, and that the doctor had signed it without reading the plan again. It explained that a correction had been added. It explained the difference between an addendum and an amendment. It apologized twice, and Mara believed both.
It did not say which office to call.
She read it again to be sure. Then she laughed, once. The dishwasher looked over. She waved him off.
Tomas called at one.
She was home by then, in the dim bedroom. She nearly did not answer an unknown number. He said he ran the referral desk at Calder Bay, that he was calling every patient whose referral had gone to Seaview since the spring, and that hers was one.
“Seaview’s the place I was sent to last year,” she said. “I never went.”
“Yes, ma’am. They stopped taking our patients in the spring. Your new referral went there anyway. That’s our mistake. It’s not anything you did. They sent it back, and it didn’t get to anybody for a few days. I’ve asked the county clinic at Calder General to see you, and I’m waiting for them to call me.”
“When will they?”
“I don’t know. I called yesterday and again this morning. I’m not going to tell you someone will be in touch.” He seemed to hear how that sounded. “I’ll call you tomorrow by five, whether I’ve heard or not. That’s my direct line on your phone now.”
It was the first thing anyone at the clinic had said to her in two weeks that she could act on. She wrote the number on the back of her hand, out of habit, though the phone had it.
“Can I ask you something else? While I’m waiting. If my hands get worse. Who’s in charge of them?”
There was a short pause. He was writing it down, she thought.
“Dr. Rowan is. I’ll put that question to her today and somebody will call you with a real answer. Not me. I’m not clinical. But I’ll make sure it’s asked.”
A nurse called at four. A hot, swollen joint, a fever, or a hand she suddenly could not grip with meant call the same day and say so. Otherwise, the anti-inflammatory she was already taking, with food. The blood tests had not shown the markers they were looking for, which made some things less likely and ruled nothing out. The nurse said that twice, carefully. Mara thought she had been told to.
On Thursday Tomas called at 4:20, forty minutes early.
A Monday, three weeks out, nine in the morning, with hand X-rays at the hospital at eight so the specialist would have them. He had asked for a Monday because her chart said the bakery was closed Mondays. She had not known her chart said that. It was the first thing in it that made her feel she had been listened to.
“There’s something else,” she said. “The note says I refused.”
“Yes, ma’am. I saw that.”
“Can you take it out?”
“No. I can’t change a clinical note, and I wouldn’t want to be able to.” He said it kindly. “I can tell you the two ways it gets fixed, because I’ve walked people through them. One, you ask the clinic to correct it. It goes to Dr. Rowan and the records office, and they have to answer you in writing, yes or no. Two is yours, and it doesn’t matter whether they agree. You can write your own statement about anything in your record you think is wrong. There’s a length limit, about a page. They have to attach it. Any time they send that part of your chart to anyone, yours goes with it.”
“Even if the doctor says I’m wrong?”
“Even then. It’d say what she says and what you say.”
“She doesn’t say I’m wrong. She wrote me a letter.”
“I heard it was a long one,” Tomas said.
Mara sent her message to Dr. Rowan before she could think better of it. Four lines. She did not say she was angry. She thought it showed.
The doctor called her at six.
“I got your message. I’d like to—” A breath. “No. I wrote you three pages already. What do you want the record to say?”
Mara had expected to be explained to. She sat down on the edge of the bed.
“That I didn’t refuse. That I asked for two weeks because of work, and I did the blood tests the same day. And I don’t want it to say I have rheumatoid arthritis if nobody knows that.”
“Nobody knows that. I don’t know it. I said possible, and that’s still what I think.” Papers, or a keyboard. “Here’s what’s already done, and you can check each one from your phone tonight. My correction’s attached to the note, under today’s date. The old wording is still there above it. I can’t make it vanish, and I don’t think I should, because people have already seen it. The diagnosis on the visit doesn’t say rheumatoid arthritis any more. It says joint pain in the hands, inflammatory arthritis being evaluated. I changed that this afternoon. I didn’t know until Tuesday that it was a separate thing I had to change. And you have a new visit summary. The old one said you declined. The new one says what you just told me.”
“It said it in three places.”
“Where was the third?”
“The letter that went to Seaview. It’s in my documents.”
Silence. Then, “I didn’t know you could see that.” It did not sound like an excuse. “Tomas is sending the county clinic the referral tomorrow. The corrected note goes with it. If you write your statement, that goes with it too. Before they ever meet you.”
Mara looked at her hands. They were not bad tonight.
“Was it you or the machine?”
“The draft was the machine’s. The signature was mine.” The doctor did not hurry past it. “There’s a piece I can’t answer yet. The first draft said you declined imaging at this time. The second said refused. I don’t know why it changed. Do you remember what you said when you came back in? About the X-ray?”
Mara tried. “I might’ve said I can’t today. I might’ve said I’m not doing it today, I’ve got to be up at two. I didn’t say never.”
“No. I was there. You didn’t.”
She wrote the statement that night at the counter, on the back of the lender’s envelope first and then on the phone. It took four tries to get the anger out and leave the facts.
In the story
I did not refuse evaluation. On the day of my visit I had blood drawn as recommended. I asked for two weeks to arrange hand X-rays and a specialist visit around my work schedule. I was told I might have an inflammatory arthritis and that this was not certain. I ask that this statement be kept with my record.
It came to fifty-nine words. She had been allowed a great many more and did not want them.
Then she moved the fruit bowl, and answered question eleven truthfully. The answer was no.
Lena Rowan opened the packet again at nine that night, at her own kitchen table, to see whether Julian had changed the sentence.
He had not yet. But there was a new section, added at 4:52 that afternoon. She had been reading a discharge summary then, and Eli Navarro had been in her doorway.
In the story
Current item 2 (entered by S. Okoye). Visit, last month: clinician dismissed AI-suggested diagnosis of atrial fibrillation. Patient subsequently diagnosed with atrial fibrillation in the emergency department. Question for conference: missed diagnosis.
She knew the name before she scrolled to it. Mr. Reyes. Inez had pointed at that line on Tuesday, and Julian had written it down.
She remembered the visit. A man who felt his heart race after coffee, whose brother had atrial fibrillation and talked about it at every family dinner. She remembered the tracing she had run in the room.
The item told the conference what the tool had suggested. It told them what an emergency department had found three weeks later.
She read it to the end, and then again. Nothing in it said what the tracing in her office had shown that day.
End of Episode 6
Continue to Episode 7: The Finding →
Or read the clinical and governance notes behind this episode ↓
The questions behind this episode
Clinical and governance notesThe Patient Who Inherits the ErrorWhat happens when an uncertain conversation becomes a certain record—and the patient must live with the difference
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
- HIPAA gives patients of covered entities a right to request amendment and to submit a statement of disagreement if it is denied; California law lets a patient who has inspected her records add a written addendum of up to 250 words per disputed item, which must accompany later disclosures of that portion; ICD-10-CM outpatient guidelines do not code uncertain diagnoses as confirmed.
- Not established, and not claimed
- That any real patient’s record contains the errors described. Mara, her encounters and the cause of the disputed wording are invented.
- The analysis in these notes
- When an uncertain conversation becomes a certain record, the patient inherits the difference. Correction needs a destination, a referral needs an owner, and the patient needs a usable way to place her own account in the record.
The episode belongs to the patient. She reads a visit summary that says she declined evaluation and a note that states a diagnosis her physician had called only possible. At that point the fictional clinic does not know how the wording arose. It could have come from the clinician’s own phrasing, a template, an imported field, a transcription error, a generated draft or a later edit. A feature being available does not establish that it caused a particular sentence. By the end of the story one part of the origin is still unknown.
The patient is already living with the result, whatever its origin. The notes below follow what a record does once it leaves the visit: how uncertainty is lost, how a request for time becomes a refusal, how a referral can exist only on paper, and which routes a patient in California has to correct the account.
The chart becomes the next conversation
A medical record travels ahead of its subject. It may shape the next clinician’s questions, the patient’s understanding of a diagnosis, a referral discussion, or an administrative assessment. Which parts travel, and how, depend on the actual system and workflow. A generated draft does not automatically become a problem-list entry or an order merely because it contains the relevant words.
Nevertheless, once a statement is accepted into a record and reused, its original uncertainty can become difficult to recover. A sentence that began as a possibility may be encountered later as background fact. A later writer may assume someone earlier confirmed it. Repetition then creates the appearance of corroboration without supplying a new observation.
Imagine that Mara sees another clinician several months later. The clinician reads the disputed label before entering the room. Mara now has to explain why the chart contains it. The visit begins with reconciliation of the record, consuming time that could have been used to address her current concern.
This is a plausible consequence, not a claim that it occurred to a particular patient. Its significance is easy to understand: the burden of correcting an institutional statement can be transferred to the person with the least access to the process that created it.
A safe documentation system should make that burden smaller. It should preserve meaningful distinctions between the patient’s report, the clinician’s observation, a considered diagnosis, a working diagnosis, and an established condition. Those distinctions are part of the care.
Uncertainty is information
The word “possible” can carry a great deal of clinical meaning. So can “resolved,” “historical,” “reported,” “not yet obtained,” and “discussed but deferred.” Removing those words may make a paragraph shorter or more decisive while changing what it says.
A patient can understand the difference between “your symptoms could have several causes” and “you have this disease.” The EHR should preserve that difference when it converts an encounter into prose, structured data, or a proposed code.
The current outpatient ICD-10-CM guidelines provide a useful, limited illustration. They direct coding of uncertain diagnoses according to the highest degree of certainty established for the encounter, rather than treating a probable or suspected outpatient condition as confirmed. This is a coding rule with a defined scope, not a complete definition of clinical reasoning. The applicable annual version and service setting matter.[1]
The larger editorial principle is that uncertainty should survive every transformation until evidence changes it. A physician should be able to discuss a serious possibility without software or a later reader converting that discussion into a confirmed disease. A patient should be able to ask about a condition without acquiring it in the chart.
Preserving uncertainty also requires specificity. A vague note that never identifies the concern or the follow-up plan may fail the next clinician. Precision is neither maximal certainty nor maximal hedging. It is an accurate account of what was known and what remained to be resolved.
A record of refusal can become a judgment of character
Mara’s portal message raises another distinction. A patient who declines a recommendation after an informed discussion has made a decision that may need to be documented. A patient who cannot obtain transportation faces an access problem. A patient who requests more time has expressed a different preference. A patient who cannot be reached presents an unresolved communication problem.
These situations can lead to the same empty box on a dashboard. They should not automatically lead to the same sentence in a medical record.
The physician should not manufacture agreement when the patient declined care. Nor should the institution turn every incomplete service into evidence that the patient was unwilling or the physician failed to offer it. The record should describe what happened, including the recommendation, the discussion actually held, the patient’s stated position, and any relevant next step.
An AI summary may have difficulty when the conversation is unfinished or the patient’s reason emerges gradually. Human authors can make the same error. The safeguard is an accurate final account and a practical opportunity to correct it, whoever produced the first wording.
This matters for trust. Mara may tolerate a typographical error. She may experience an inaccurate statement about refusal as a judgment about her reliability. If the clinic responds only that the note was signed, it answers a question about authentication while leaving her question about truth untouched.
The signature gives the clinic a responsible person to ask. It does not make asking unnecessary.
The referral that exists only on paper
In the story a contract change means that patients need a different pathway for a specialty service. Imagine that the organization has a transition plan and sends notices. Those are real steps, and they deserve recognition.
But a letter is evidence of a letter. It does not, by itself, establish that the patient understood the change, obtained an appointment, reached an appropriate receiving clinician, or knew whom to contact while waiting.
Mara has questions that an organizational announcement may not answer. Is the new clinician taking patients? Has the referral arrived? Who handles a time-sensitive concern before the first visit? Which practice follows up an outstanding result? What should she do if she receives conflicting instructions?
A referral can be clinically appropriate and operationally incomplete. The answer is to specify the unfinished work, not to declare that no plan exists.
For Health Center Program participants, HRSA’s requirements for formal referral arrangements address how referrals are managed and how patients are tracked and referred back for appropriate follow-up. The chapter concerns services within the approved program structure; it is not a universal instruction governing every referral in every practice.[2]
The practical proposal is broader: each transition should have a recognizable owner, a receiving pathway, a way to notice failed connections, and an escalation route proportionate to the clinical need. The patient should not have to infer these arrangements from the movement of an order through a computer.
The physician’s limits do not end the organization’s work
A clinician may appropriately recognize that a service exceeds their preparation or the support available in their practice. That recognition can protect patients. It should lead to an appropriate assessment of competence, consultation, referral, supervision, or service organization.
It should not become a general claim that an entire category of patients is outside primary care. Competence depends on the actual service and circumstances, and patients deserve respectful, nondiscriminatory access to appropriate care. Nor does the presence of a willing clinician eliminate the need for adequate facilities, staffing, follow-up, and emergency arrangements.
In the fictional clinic, the question therefore changes from “who will take these patients?” to “how will the organization maintain a safe transition for each affected patient?” The second question includes administrative work that physicians cannot complete by writing a better note.
The Medical Board of California’s guidance on closing or leaving a practice emphasizes advance notice, information about records, and assistance obtaining another physician. That guidance addresses a defined transition setting; it does not decide every specialty-service change. It nevertheless illustrates why notification and continuity are related but distinct responsibilities.[3]
An institution that announces a transition should be able to examine whether its process worked. A physician who raises a remaining gap should be able to identify the gap accurately. Neither side helps Mara by describing an incomplete handoff as a completed one, or a genuine plan as no plan at all.
The result can arrive after the visit has disappeared
Consider a second fictional patient, returning to the clinic after a hospital stay. The appointment occurs, a note is signed, and the encounter is closed. An outside result or medication clarification arrives afterward.
From the scheduling system’s perspective, the visit is over. From the patient’s perspective, the episode of care is still unfolding.
The important questions are who receives the information, who recognizes its significance, who acts, who contacts the patient when appropriate, and who provides coverage if the usual clinician is unavailable. A polished account of the original appointment cannot substitute for those later actions.
HRSA’s continuity chapter makes this distinction concrete for participating health centers: hospital and emergency-department procedures must address receipt and recording of relevant information and appropriate follow-up, with evidence that the procedures are actually followed. The local tracking method remains a matter of implementation.[4]
An AI documentation tool may help assemble a coherent summary. That is useful. It cannot be assumed to have created staff capacity, assigned the result to a responsible clinician, or completed a conversation with the patient. Those functions require their own evidence.
When an organization measures only how quickly a note becomes complete, the unfinished work can become less visible precisely because the visible record looks so orderly.
What the dashboard cannot explain alone
In Rowan’s clinic, a quality dashboard shows a missing service. The underlying patient relationship has changed, the panel assignment may be outdated, and part of the care occurred elsewhere. Any of those facts might matter. None automatically excuses a missed opportunity or removes the patient from the measure.
The correct question is what the measure requires, who was responsible during the relevant period, what information was available, and what the encounter record supports. The current dashboard may not reproduce exactly what the physician saw at the time.
This is where patient safety and fair physician assessment meet. If the organization assigns a missed service to the wrong clinician, it may also send the remedial task to the wrong team. Correct attribution helps the patient reach someone who can act.
Conversely, discovering an attribution problem should not end the inquiry when care remains outstanding. Administrative ambiguity is a reason to resolve responsibility. It is not a clinical plan.
A useful quality process can hold both ideas: the performance conclusion may need correction, and the patient may still need follow-up. The first protects fairness; the second prevents a dispute about fairness from consuming the purpose of the review.
Thoroughness that never happened
An AI-assisted note can make an ordinary visit read like a comprehensive counseling session. The danger is greatest when a template asks for a finished narrative even though the source material supports only a proposed action.
“Discuss options” is not the same statement as “options were discussed.” “Consider referral” is not “referral placed.” “Will review outside records” is not “outside records reviewed.” A change of tense can manufacture an event.
Mara may read that the risks and benefits were explained and wonder whether she missed something. The next clinician may assume that discussion already occurred. A quality reviewer may credit the note for an action that the underlying encounter never established.
The solution is faithful documentation and appropriate care. If counseling was required and omitted, the omission should be addressed. Adding a completed-action sentence does not repair it. If an action was only contemplated, the record should retain that status until the action occurs.
This principle applies equally to human dictation, smart phrases, copied text, and generative systems. The source of the prose does not change the difference between an event and a proposed event.
The patient benefits when quality review rewards that distinction. An accurate note may reveal work still to be done. That visibility is more useful than a record that appears complete because it has converted every intention into a past-tense accomplishment.
A correction needs a destination
Rowan reads Mara’s message and agrees that the record needs attention. The first task is to determine the appropriate correction through the clinic’s authorized process. The next is to ask where the disputed statement exists.
Is it only in the narrative? Is it also in a structured diagnosis field? Did it enter a problem list, an after-visit summary, or a referral document? Was it shared with another clinician? Each possibility needs verification; none should be assumed from the presence of the sentence alone.
If a correction is appropriate, the institution should consider the affected locations and the need for communication. Changing one paragraph may not resolve a separate structured entry. Conversely, an inaccurate historical note should not be silently rewritten to make its original existence impossible to understand. An appropriate amendment can preserve the sequence while making the correction clear.
HIPAA gives patients of covered entities rights to request amendments. HHS explains that the entity must respond and that a patient can submit a statement of disagreement when a request is denied. When a covered entity accepts a requested amendment, the regulation requires it to identify the affected records in its designated record set, append or link the amendment, and make reasonable efforts to provide it to the recipients specified in the rule. Those duties do not mean that every connected system updates automatically. The right is not a guarantee of erasing every disputed statement or replacing professional judgment with the patient’s preferred account.[5]
An accessible correction process should explain what will be reviewed, who will respond, and how the patient can understand the outcome. Mara should not need to identify the software vendor, distinguish a model from an integration, or prove which employee clicked a button before her concern receives attention.
What California law adds for Mara
Mara is a patient in California, and California law gives her a route that does not depend on the clinic agreeing with her.
California’s patient-access statute allows a patient to inspect and obtain copies of her records within time limits the statute sets. After inspecting her records, a patient may submit a written addendum about any item or statement she believes is incomplete or incorrect. The addendum is limited to 250 words for each such item and must state clearly that the patient wants it made part of the record. The provider must attach it to the record and include it whenever the disputed portion is disclosed to a third party.[6]
The addendum and the HIPAA amendment request are different instruments. An amendment request asks the covered entity to change its record, and it may be accepted or denied. The California addendum places the patient’s own account beside the provider’s, whether or not the provider agrees. A patient can use either route or both. Neither requires her to prove how the disputed sentence was created.[5, 6]
For the clinic, the practical lesson is that correction is not only a clinical-documentation task. It is a patient-rights process with its own requirements, and the people who handle it, including health-information staff, clinicians and privacy officers, need a shared understanding of which route a patient is using and what each route requires.
For Rowan, the addendum is no threat to clinical judgment. If the physician disagrees with Mara’s account, the record can contain both accounts, clearly attributed. That is more honest than a chart that preserves only one voice, and more useful to the next clinician, who will at least know that the statement was disputed.
A patient’s complaint should not become evidence against the wrong person
Suppose a quality reviewer later sees Mara’s complaint. It is an important source. The reviewer should investigate the discrepancy and any clinical consequence. But the complaint does not, by itself, identify the origin of the wording or establish every element of a performance allegation.
The signed record matters. So do the available source material, the clinician’s explanation, the patient’s account, subsequent corrections, and the workflow through which the information moved. A credible review can find a physician’s verification inadequate without pretending the software played no role. It can find a software or process defect without pretending the clinician had no responsibility.
This is particularly important when the physician has already raised concerns about the same workflow. That history may identify a known problem or a missed opportunity to investigate. It does not prove the current error resulted from that problem, and it does not automatically establish retaliation if a review follows.
The patient’s interests call for specific answers: what was inaccurate, what care or communication was affected, what has been corrected, and what will reduce recurrence. A contest over which party can absorb all the blame is unlikely to produce those answers.
The organization should preserve enough information to examine a meaningful concern, consistent with privacy, security, applicable retention requirements, and authorized preservation duties. It should avoid promising that every source will exist forever or implying that a missing recording prevents any reliable finding.
Questions a patient can ask
Mara should not need to understand software architecture to protect her own care. A short set of questions, asked through the clinic’s ordinary channels, can move a concern toward a useful answer. This is general information, not legal advice about any particular situation.
- Where else does this statement appear: in a diagnosis list, a problem list, an after-visit summary or a referral?
- Has the record containing it been sent to another clinician or organization?
- How do I request an amendment, and how do I submit a written addendum in my own words?
- Who will respond, and will the response be in writing?
- What is the current status of my referral, who is responsible for it, and what is the plan while I wait?
- What symptoms or changes should prompt me to call before the next appointment, and whom should I call?
- If I need an interpreter to discuss this, how do I arrange one?
A clinic that can answer these questions promptly has a working correction process. A clinic that cannot has learned something important about its own system, and has learned it from the person the system exists to serve.
Why physician autonomy matters to Mara
From Mara’s perspective, physician autonomy is useful when it protects the clinician’s ability to listen, evaluate evidence, preserve uncertainty, challenge an unsuitable recommendation, and arrange appropriate care. It is less useful as an abstract claim of status.
The physician needs room to disagree with a generated diagnosis. The organization needs a way to assess whether that disagreement was clinically sound. The patient needs both.
Similarly, the physician needs to raise a concern about an unsafe handoff without that question being casually converted into resistance to change. The organization needs clinicians who will participate in improvements and complete legitimate requirements. These responsibilities can coexist when the expectations and the evidence are clear.
A practical test for a new workflow is to follow one fictional patient through it. Can the patient understand the record? Can an uncertain diagnosis remain uncertain? Can a factual error be corrected without losing an accurate plan? Can an incomplete referral be found? Does someone own the next action after the encounter closes? Can the clinician explain a decision without competing against a machine-generated account of a conversation that never occurred?
Those questions can be tested before implementation and revisited after changes. They are proposed design and governance tests, not a claim that one regulation mandates this exact exercise.
The last reader
The episode begins with a patient reading her record. The series ends with another clinician reading it.
Mara does not need the clinic to promise that no human or machine will ever make a mistake. She needs an institution that takes an unfamiliar sentence seriously, examines it without defensiveness, corrects what should be corrected, and keeps the work of care moving while the explanation is developed.
She also needs the next reader to know what is established, what was considered, what was declined, what could not be obtained, and what still needs to happen.
Physicians, administrators, developers, and reviewers each control part of that account. Their responsibilities differ. The patient’s dependence on the account is shared across all of them.
When the AI-assisted record becomes evidence, it describes more than the clinician’s performance. It becomes part of the patient’s future. The patient is entitled to a place in the conversation about how that future is written.
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.
- Centers for Medicare & Medicaid Services and the National Center for Health Statistics, ICD-10-CM Official Guidelines for Coding and Reporting, FY 2027, Section IV.H, outpatient uncertain diagnoses. Effective October 1, 2026–September 30, 2027. This version is not applied retroactively to earlier encounters.www.cms.gov
- Health Resources and Services Administration, Health Center Program Compliance Manual, Chapter 4, Required and Additional Health Services. The discussion concerns formal arrangements for services in the approved scope of project.bphc.hrsa.gov
- Medical Board of California, Information Pertaining to the Practice of Medicine, Closing Your Medical Practice. Guidance on closure or departure, used here as a limited illustration of transition responsibilities.www.mbc.ca.gov
- Health Resources and Services Administration, Health Center Program Compliance Manual, Chapter 8, Continuity of Care and Hospital Admitting. The cited requirements concern hospital and emergency-department continuity for participating health centers.bphc.hrsa.gov
- U.S. Department of Health and Human Services, Your Medical Records, including the corrections discussion; 45 C.F.R. § 164.526, particularly subsection (c) on accepted amendments and communication to specified recipients.www.hhs.gov www.ecfr.gov
- California Health and Safety Code section 123111, in the chapter on Patient Access to Health Records that begins at section 123100: a patient who has inspected records under section 123110 may submit a written addendum of up to 250 words per item believed incomplete or incorrect; the provider must attach it and include it in any disclosure of that portion to a third party. Statutory text as reproduced by Justia; the statute’s conditions and exceptions control. Checked November 19, 2026.law.justia.com
Related reporting and resources on this site
About this page. This is Episode 6 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.