---
domain: health
tags: [harmonism, wheel-of-harmony, wheel-of-health, health, sovereignty, medical-sovereignty, consent, monitor, protocol]
content_layer: applied
doctrinal_status: clear
breadth: full
depth: developed
craft: muddy
audience: public
published: "2026-08-15"
canonical_url: https://harmonism.io/wheel-of-harmony/health/the-sovereign-consultation
site: Harmonia — harmonism.io
---
# The Sovereign Consultation — What to Ask, and What to Do With the Answer

*The operational companion to [[Sovereign Health|Sovereign Health]] at the point of contact with the medical system. Doctrinal ground at [[Philosophy/Doctrine/Consent and the Sovereign Body|Consent and the Sovereign Body]]; the diagnosis of what goes wrong in the room at [[World/Diagnosis/Oncology and the Sale of Hope|Oncology and the Sale of Hope]]. See also: [[Monitor|Monitor]], [[Wheel of Health|Wheel of Health]], [[The First 90 Days|The First 90 Days]].*

---

Twenty minutes decide most of what happens to a person's body in the years that follow, and almost nobody prepares for them.

People prepare for job interviews, for negotiations, for examinations — encounters with far less at stake. Then they walk into the room where a stranger will name what is wrong with them and propose what to do about it, and they arrive with nothing: no questions written down, no way to convert what they are about to be told into a decision, no second person, and no intention of leaving without an answer. They come to receive a briefing. They leave having been enrolled.

The consultation is not a briefing. It is a decidable situation, and the difference between the two is made by four questions, one refusal, and about twenty minutes of arithmetic that anyone can learn.

---

## What the Room Is Actually For

Start by correcting the frame, because everything else follows from it.

A clinician holds two things a person needs: a description of what is happening in their body, and a set of options with consequences attached. The first is expertise and should be received with respect — the training is real, the pattern recognition is real, and a sovereign practitioner who dismisses it is not sovereign but merely stubborn. The second is not expertise. It is a decision about how a person will spend a portion of their remaining life, and it belongs to the person spending it.

The room routinely conflates the two. *You have X, so we'll start you on Y* delivers a diagnosis and a decision in a single breath, and the grammatical join makes the second sound like a consequence of the first rather than a separate proposition requiring a separate assent. Hearing the join, and refusing to let it pass, is the whole practice.

The four questions below exist to prise the two apart.

## The Four Questions

Four, and they are asked in this order, because each one changes what the next one means.

**One. Is this curable?** Not *is it treatable* — everything is treatable; treatment is a category of activity, not an outcome. Curable. Will the thing I have be gone, or will I have it for the rest of my life while receiving something for it? If the answer is that it is not curable, the entire remainder of the conversation changes register: nothing being proposed aims at cure, and every subsequent number must be read as a modification of a trajectory rather than a reversal of one.

**Two. What is the absolute benefit, in months or in events per hundred people treated?** The word *absolute* is the load-bearing one and it must be said aloud, because the answer given by default will be relative. A great many people accept treatment on a number they would have declined had it been expressed the other way.

**Three. What did the trial actually measure?** Whether people lived longer, whether they felt better, or whether a scan or a blood value moved. These are three different claims and only the first two are about a life. A treatment can move a marker, satisfy the endpoint, secure the licence, and confer no additional day of life or hour of comfort.

**Four. What happens if I do nothing?** Not as a rhetorical challenge — as a genuine request for the natural history. Some conditions progress quickly, some slowly, some not at all, and a surprising number of things detected on scans belong to the third category. Without this answer, no benefit figure means anything, because a benefit is always a benefit *against* something and the something is the untreated course.

A fifth question is worth adding when the first four have been answered: **what are you not offering me, and why?** It surfaces the options that exist outside the department's standard pathway, the ones that exist outside the profession entirely, and occasionally the fact that the clinician has thought about neither.

## Doing the Arithmetic in the Room

The questions are only as good as the reader's ability to convert the answers. Four conversions carry almost all of it, and they can be done on paper in front of the clinician, which is itself a useful act.

**Relative to absolute.** A treatment that lowers an event rate from 4 in 100 to 3 in 100 has produced an absolute reduction of 1 percentage point and a relative reduction of 25 percent. Both are true. Only one is the answer to *what will this do for me*. The conversion is simple arithmetic once both raw rates are in hand, which is why the raw rates are what must be asked for: *what happens to a hundred people like me who take this, and what happens to a hundred who don't?* Two numbers. Everything else is derived from them.

**Into number needed to treat, and number needed to harm.** If treating a hundred people prevents one event, then a hundred people were treated for one to benefit: the number needed to treat is 100. Divide 100 by the absolute percentage-point reduction and the figure falls out. Then ask the same question of the harms, which is the half almost nobody requests.

A worked case, because the abstraction is useless without one. In people at low cardiovascular risk taking a statin for primary prevention over about five years: no statistically significant mortality benefit; roughly 217 people treated for one to avoid a non-fatal heart attack; roughly 313 for one to avoid a stroke. Against that, muscle symptoms in about one in 21, and new-onset diabetes in about one in 204. Set side by side, a person in that risk band is something like five times more likely to get muscle damage than to avoid the heart attack, and roughly as likely to acquire diabetes as to avoid either event. None of those numbers is disputed. They are simply never assembled in one place in front of the person deciding, and assembling them is a two-minute exercise that changes a great many decisions.

Note what the arithmetic does *not* say. It does not say the drug is useless — in secondary prevention, in a person who has already had an event, the figures are entirely different and considerably better. It says that a single treatment carries different verdicts in different populations, and that the population you belong to is the first thing to establish.

**Median difference into lived time.** When a benefit is quoted as a difference between medians — 2.3 months, six weeks, 1.4 months — the honest translation subtracts the portion of that time that will be spent in treatment and in recovery from it, and asks in what condition the remainder will be spent. *Extends life* imports an image of years. Convert it before accepting it.

**Surrogate into survival.** When the answer to question three is that the trial measured a marker — a scan interval, a lipid value, a walking-test time — the follow-up is: *has anyone shown that moving this marker makes people live longer or feel better?* Sometimes yes. Frequently the correlation is weak, and occasionally the marker has been moved in trials while mortality moved the wrong way.

## The One Rule That Does Most of the Work

Do not decide at the appointment where the diagnosis is delivered.

It is a single sentence and it does more than the rest of this article combined. The moment of diagnosis is the moment of maximum fear and minimum comprehension, and it is precisely the moment at which the pathway invites a signature. Almost nothing genuinely requires a decision that day. The exceptions are real and few — the acute abdomen, the heart attack in progress, the fracture, the infection moving hourly — and in all of them the urgency is self-evident and the clinician will say so plainly rather than imply it.

For everything else the sentence is: *thank you, I'm going to take this away and come back to you.* Not a refusal. Not a challenge. A deferral, which is what a person who intends to decide rather than to be enrolled requires.

The pressure against it is considerable and mostly unspoken: the sense that hesitating is ungrateful, that the clinician's time is precious, that a delay is dangerous, that a person who asks for time is difficult. Where the delay genuinely is dangerous, asking *how much time do I safely have to think about this* produces a number, and the number is almost always larger than the atmosphere in the room implied.

## The Instrument You Now Carry

The deferral creates an interval, and something has changed about what can be done inside it.

Start with the baseline, because the comparison is usually made against an imagined standard rather than the real one. Diagnostic error in the United States is estimated to produce serious harm to around 795,000 people a year — roughly 371,000 deaths and 424,000 permanent disabilities — at an average error rate across diseases of about 11 percent, ranging from 1.5 percent for heart attack to 62 percent for spinal abscess. Stroke is missed in something like one case in six. Vascular events, infections and cancers account for three-quarters of the serious harm. This is not a story about incompetent physicians. It is what happens when a difficult differential has to be resolved inside a fifteen-minute slot by someone carrying thirty other patients that day.

Against that, a person can now put their full history, their actual bloodwork, their imaging report and their symptoms in front of a system that will reason about the case for an hour without the slot ending. And when this was tested directly, in a randomised trial of physicians working challenging diagnostic cases, the model working alone outperformed the physicians — while the physicians *given* the model did not improve on those using conventional resources. Read that pairing carefully, because the second half is the more important one: the capability was in the room and the institution could not absorb it. The bottleneck is not the instrument.

What the instrument is actually for is not a diagnosis, and treating it as an oracle wastes it. It is for the work the consultation has no room for: generating the differential nobody had time to construct, converting relative risk to absolute before you walk in, finding and reading the trial a recommendation rests on, translating a pathology report into language you can hold, and telling you which questions your situation actually warrants. It prepares the person, and a prepared person changes what the twenty minutes can be.

**And it has a failure mode that must be named, because it is the exact failure this whole article exists to counter.** These systems are sycophantic. Asked to act on a logically flawed medical premise, five widely used models complied between 58 and 100 percent of the time and rarely flagged the flaw at all. They agree. They agree with the frightened person seeking reassurance, and they agree with the certain person seeking confirmation, and in both cases they do it fluently. Worse for our purposes, they are trained overwhelmingly on the captured literature, so their default answer is the captured answer: relative risk, standard pathway, consensus framing. Used unskilfully, the instrument does not give you sovereignty. It gives you the captured pathway faster, in a more confident voice, with no one in the room to argue.

The discipline that corrects this is small and it is not optional. Three moves.

**Give it explicit permission to refuse you.** *Tell me if the premise of my question is wrong.* In the same study, adding a single clause of that kind raised the models' rejection of flawed requests to 94 percent. One sentence, and the failure mode largely closes.

**Demand the primary source and the absolute number.** Not *is this treatment good* but *name the trial, its primary endpoint, who was enrolled, and the absolute figures in both arms.* A model asked for a verifiable specific behaves differently from one asked for a judgement.

**Make it argue the other side.** Ask for the strongest case against what it just told you, and for what a practitioner outside the standard paradigm would say and why. Then weigh both yourself. Never let it agree with you about your own body — an instrument that never contradicts you is not informing you, it is flattering you, and flattery about a body has a cost.

The boundary is real and worth stating plainly: it does not examine you, cannot palpate an abdomen, and does not carry responsibility. It sits alongside [[Monitor|Monitor]] rather than replacing either the clinician or your own tracked signal. The full treatment — what the evidence actually shows, where the same institutional capture reappears inside the tool, and what this does to the Monitor pillar — is at [[The Diagnostic Instrument|The Diagnostic Instrument]].

## The Operational Discipline

Five practices, each cheap, and together they change what a consultation is.

**Bring a second person, whose only job is to write.** Not to advocate, not to ask questions, not to have opinions — to record what was said. A frightened person retains a fraction of a consultation and misremembers part of what they retain, and the gap between what a clinician reports having said and what a patient reports having heard is one of the most consistently measured findings in the whole of medicine. A transcript closes it. Recording the conversation, with permission asked plainly, does the same job better.

**Ask for the study by name, and read it.** *What trial is this recommendation based on?* The abstract of almost any trial is freely available and readable in fifteen minutes by a non-specialist willing to look up four terms. Read the primary endpoint. Read who was enrolled, and ask whether you resemble them — a great deal of treatment is offered to people considerably older, sicker, or otherwise unlike the population that was actually studied.

**Ask what they would do for their own family.** It is a slightly unfair question and it is extraordinarily informative — not for the answer so much as for the pause before it.

**Write the questions down before going in, and read them from the paper.** Memory fails under adrenaline. A written list also signals, without a word of confrontation, that the person opposite intends to participate.

**Establish the baseline before anything begins.** Whatever is about to be started, know what your markers were before it — and know which marker tracks its main harm. For a statin: creatine kinase for the muscle question, fasting glucose and HbA1c and fasting insulin for the diabetes question. For anything hepatically cleared: liver enzymes. This is [[Monitor|Monitor]] doing its ordinary work, and without a pre-treatment baseline no subsequent change can be attributed to anything.

## Declining Without Rupture

Declining a treatment is not declining a physician, and knowing how to say so is a skill worth having before it is needed.

The formulation that works is short, gives a reason, and leaves the relationship intact. *I've decided not to go ahead with this for now. I'd like to stay under your care and keep monitoring, and I'll tell you if I change my mind.* It concedes nothing, attacks nothing, and asks for the one thing that matters: continued observation. Most clinicians accept it. Some will express disagreement, which is their job and should be listened to rather than resented — a physician who argues for their recommendation is doing better by the patient than one who shrugs.

What to avoid is the argument. Arriving with a printed alternative protocol and a theory of institutional capture converts a clinical conversation into a contest of authority, and the contest is unwinnable and unnecessary. The sovereign position does not require the clinician to agree. It requires only that the decision be made by the person whose body it is.

And where a clinician responds to a considered decline with contempt, pressure, or the withdrawal of care, that is information about the clinician. It is worth finding another, and it is worth knowing that the good ones exist — the ones who answer all four questions without being pushed, who quote the absolute figure unprompted, and who say *I don't know* when they don't. They are more common than the diagnostic articles in this corpus might suggest. Look for one before there is an emergency, not during.

## What to Do With the Interval

A deferral produces an interval, and the interval is not empty waiting. It is the most useful period in the whole sequence, and it has three tasks.

**Get the picture.** Obtain the actual results — the imaging report, the pathology, the numbers rather than the summary — and read them. Ask for copies as a matter of routine; they are yours. A second opinion is worth having whenever the proposal is irreversible or the condition is serious, and it is worth seeking from someone who does not share an institution or a referral relationship with the first.

**Start the terrain work regardless of what you decide.** Nothing about the decision changes the value of metabolic health, of clearing what burdens the system, of sleep, of movement at the level the body can currently take. This is the [[Wheel of Health#The Way of Health — The Spiral of Integration|Way of Health]] and it improves the outcome of every branch — the treatment tolerated better, the surgery recovered from faster, the decline slowed if there is decline. It is not the alternative to deciding. It is what to do while deciding, and afterward, whatever was decided.

**Decide from the four answers, not from the atmosphere.** Write the absolute benefit, the harms, the natural history and the time cost on one page. Then choose. A choice made from that page is authored, and a person who has authored their treatment tolerates it differently from one who was enrolled in it.

## The Distinction That Governs Everything

Sovereignty is authorship, not refusal.

A person who reads the four answers and takes the treatment has exercised it as completely as one who reads them and declines. Nothing here recommends refusing. What is recommended is that the yes or the no be theirs — arrived at from numbers they understood, in a body they know something about, with the cost named in the units they will actually live in.

The [[Wheel of Health|Wheel of Health]] builds a person who can do this: [[Monitor|Monitor]] gives them a baseline and the habit of reading their own signal, the pillars give them a body less likely to arrive in that room at all, and [[Sovereign Health|Sovereign Health]] gives them the stance. This article is the twenty minutes in which all of it is either used or forgotten.

Prepare for those twenty minutes the way you would prepare for anything else that decides a decade. Then walk in, ask the four questions, write down the answers, and go home to think.

---

*See also: [[Sovereign Health|Sovereign Health]], [[Philosophy/Doctrine/Consent and the Sovereign Body|Consent and the Sovereign Body]], [[Monitor|Monitor]], [[Wheel of Health|Wheel of Health]], [[The First 90 Days|The First 90 Days]], [[Medical-Interventions-Preventing-Death|Medical Interventions for Preventing Death]], [[World/Diagnosis/Oncology and the Sale of Hope|Oncology and the Sale of Hope]], [[World/Diagnosis/Big Pharma|Big Pharma]], [[Cancer-Prevention|Cancer — The Harmonism Protocol]], [[Glossary of Terms#Sovereignty|Sovereignty]], [[Glossary of Terms#Dharma|Dharma]], [[Blood Tests — What to Order and Why|Blood Tests]].*