Contraindication¶

Flag the sparse patient conditions under which a normally-indicated treatment becomes inadvisable, by naming the specific context in which its risk-benefit balance flips sign.

Core Idea¶

A contraindication is the condition-triggered prohibition pattern in clinical medicine: a treatment that is generally indicated — meaning its expected benefit exceeds its expected risk for the typical patient population — becomes inadvisable or impermissible when a specific patient condition is present, because in that condition the risk-benefit ratio inverts. The condition does not change the treatment's mechanism; it changes the patient context against which the treatment's harm and benefit are calculated.

The pattern has three load-bearing parts. The default indication is the treatment-context pairing under which the action is appropriate — the baseline population and clinical setting for which the treatment was established. The conditioning context is the specific patient characteristic, diagnosis, physiological state, or co-existing treatment whose presence alters the expected harm-benefit balance enough to trigger the rule. The prohibition rule is the restriction itself, which clinical medicine codes in two gradations: absolute contraindications (the treatment must not be given under the conditioning context, with no room for compensating adjustments) and relative contraindications (the risk-benefit ratio worsens enough to require serious alternative consideration or protective co-intervention, but clinical judgment may still permit the treatment with appropriate safeguards). Classic instances in pharmacology include beta-blockers in asthma (bronchospasm risk), ACE inhibitors in pregnancy (fetal renal toxicity), and live vaccines in severely immunocompromised patients (uncontrolled infection risk). These are encoded in formularies, drug labelling, prescription-screening algorithms, and clinical decision-support systems, which operationalise the contraindication catalog into point-of-care flags against patient records.

Structural Signature¶

Sig role-phrases:

the default indication — the treatment-context pairing under which the action is generally appropriate (benefit exceeds risk for the typical population)
the conditioning context — the specific patient condition, diagnosis, physiological state, or co-treatment whose presence triggers the rule
the risk-benefit sign-reversal — the defining move: under the conditioning context the harm-benefit balance inverts, while the treatment's mechanism is unchanged
the prohibition rule — the explicit restriction fired by the trigger, the codified "do not give under condition X"
the absolute/relative gradient — the engineered gradation: absolute (no safeguard rescues the treatment) versus relative (a worsened balance reopened as a weighing under monitoring or co-intervention)
the fallback — the alternative class whose balance stays favourable under the condition, so the prohibition is generative of the next action
the codification apparatus — formularies, drug labelling, and prescription-screening systems that operationalize the rule into point-of-care flags against patient records
the scope boundary — its characteristic limit: a contraindication is the condition-triggered sign-flip, not a generic side effect, a narrow therapeutic window, or mere absence of effect

What It Is Not¶

Not a side effect. A side effect is an adverse outcome the treatment may produce; a contraindication is the specific patient condition under which an otherwise-favourable risk-benefit balance flips sign. Every drug has side effects without being contraindicated; the contraindication is the context that makes giving it inadvisable, not the harm it might cause in general.
Not a blanket "never use this drug." A contraindication is condition-triggered: the treatment remains indicated for the population at large and is withheld only when the conditioning context is present. It marks the sparse exceptions where the sign reverses, not a treatment that is bad everywhere.
Not all-or-nothing. Absolute and relative contraindications are different decisions: an absolute one is a flat prohibition no safeguard can lift, while a relative one reopens a weighing in which monitoring or co-intervention may still license use. Collapsing both into an undifferentiated "don't" discards exactly the gradient that tells the clinician what kind of decision is owed.
Not a narrow therapeutic window. A therapeutic window is a dose-range question — too little does nothing, too much harms — within a treatment that still applies. A contraindication is a context question: under the triggering condition there is no favourable use at any dose, so it is not about calibrating the amount but about whether to give the treatment at all.
Not a change to the drug's mechanism. The conditioning context does not alter how the treatment works; it alters the patient context against which its harm and benefit are scored. The same pharmacology that helps the typical patient harms the contraindicated one — what flipped is the balance, not the mechanism.
Not mere absence of benefit. A contraindication is an active sign-reversal (expected harm now exceeds expected benefit for this patient), not a case where the treatment simply does nothing useful. "No effect" is a different situation from "the balance has turned against use."

Scope of Application¶

Contraindication lives across the subfields of clinical medicine that pair an indicated treatment against patient states — the settings where a specific condition can invert an otherwise-favorable risk-benefit balance; its reach is bounded to that medical substrate, all of it codified uniformly in formularies, labelling, and prescription-screening systems. The conditional-constraint shape that recurs in engineering safe-operating limits or policy carve-outs travels by the parent composition constraint + risk + exclusion (with the absolute/relative gradient), not by "contraindication" as named; those stay out of this map.

Pharmacology — drug–disease, drug–drug, and drug–food contraindications (beta-blockers in asthma, ACE inhibitors in pregnancy, MAOI + tyramine, warfarin + certain antibiotics), the home turf.
Surgical and procedural medicine — absolute and relative contraindications for surgery: active infection, severe coagulopathy, patient refusal.
Imaging — condition-triggered prohibitions such as gadolinium contrast in renal failure, flagged against the patient record before the study.
Vaccination — anaphylaxis history and, for live vaccines, severe immunocompromise, where the sign of the balance inverts under the conditioning state.

Clarity¶

The concept's clarifying work is to surface, and force into the open, a piece of risk-benefit reasoning that would otherwise stay tacit in clinician training. A treatment carries one indication and an open-ended set of patient states in which that indication fails; without the contraindication frame those reversals live as scattered clinical lore, recalled or not at the bedside. Naming the pattern compels three questions to be answered separately rather than fused into a single intuition of "is this a good idea?": what is this treatment indicated for, what specific condition inverts its risk-benefit balance, and what is the fallback when it does. Structured formulary fields, drug labelling, and prescription-screening systems are only possible because the reasoning has been factored this way — the codification is what lets the risk-reversal be checked against a patient record by something other than the prescriber's memory.

Its sharpest distinction is the absolute-versus-relative gradient, which keeps two very different clinical situations from collapsing into one undifferentiated "don't." An absolute contraindication closes the question — no safeguard rescues the treatment under that condition — whereas a relative one reopens it as a weighing, where co-intervention or monitoring may still license use. Holding these apart tells the clinician what kind of decision is owed: a flat prohibition to honour, or a risk calculation to perform with eyes open. It also draws a clean line the field is prone to blur — a contraindication is not a mere side effect or a general risk, but the specific patient context in which an otherwise-favourable balance flips sign, which is exactly the question a point-of-care alert is built to ask.

Manages Complexity¶

The space a prescriber faces is, in principle, the full cross-product of every treatment against every patient state — each drug, procedure, and vaccine paired against each diagnosis, physiological condition, pregnancy status, age band, and concurrent medication, an unbounded lookup that no clinician could hold and no single risk-benefit intuition could compute case by case. Contraindication compresses that matrix by exploiting a structural fact: a treatment's risk-benefit balance is favourable across the great bulk of patient states and inverts only in a sparse, enumerable set of conditioning contexts. The analyst therefore does not evaluate the whole matrix but tracks, per treatment, just its default indication plus the short list of conditions that flip the sign — beta-blockers favourable except in asthma, ACE inhibitors favourable except in pregnancy, live vaccines favourable except under severe immunocompromise — and reads the qualitative outcome (give, withhold, or weigh) off whether any listed condition is present in the patient. The cross-product collapses to a sparse exception table because the sign of the balance is the only quantity that needs gating, and it is constant across nearly the entire context space. A second compression is the absolute–relative gradient, which reduces the decision type to a binary branch the moment a condition fires: an absolute contraindication terminates the evaluation with a flat prohibition no safeguard can lift, while a relative one reopens it as a bounded weighing in which monitoring or co-intervention may still license use — so the clinician reads not just whether the balance inverted but what kind of decision is now owed, without re-deriving the safeguard calculus from first principles. Because the reasoning has been factored into these three separable fields — indication, conditioning context, prohibition rule with its gradation — the sparse exception table is machine-checkable: formularies, drug labels, and prescription-screening systems run the patient record against the listed conditions and surface a point-of-care flag, so the high-dimensional treatment-by-condition problem is discharged by a lookup that fires only on the enumerated sign-reversals rather than by re-weighing every prescription against every state. The analyst tracks a default and its exceptions, reads off act-or-withhold from the exception list and absolute-or-relative from the gradient, and leaves the rest of the matrix — the vast favourable region — uninspected because the structure guarantees nothing there changes the sign.

Abstract Reasoning¶

A contraindication licenses a sign-of-the-balance inference that is sharper than ordinary risk talk: the clinician reasons not about whether a treatment carries risk (every treatment does) but about whether a specific patient condition inverts the sign of its risk-benefit balance. The move runs FROM "this patient has asthma and the proposed drug is a beta-blocker" TO "the bronchospasm risk flips an otherwise-favourable balance, so withhold," and crucially the clinician infers that the treatment's mechanism is unchanged — what changed is the patient context against which harm and benefit are scored. This is a diagnostic move on the decision, not on the patient: from the presence of a listed conditioning context the clinician concludes the default indication no longer holds for this patient, while it continues to hold for the population at large.

The governing predictive heuristic is default-with-exceptions: the clinician predicts a favourable balance across the vast bulk of patient states and treats sign-reversal as confined to a sparse, enumerable exception set, so the reasoning is to check the patient against the short list of sign-flipping conditions rather than to re-weigh the treatment from scratch. The act-selection is read off that check — no listed condition present, give as indicated; a listed condition present, do not give as default — which converts an open-ended "is this a good idea?" into a bounded lookup against named triggers.

The boundary-drawing move is the absolute-versus-relative gradient, which tells the clinician what kind of decision is owed the moment a condition fires. From an absolute contraindication the clinician infers a terminated question — a flat prohibition no safeguard can lift — and stops weighing; from a relative one the clinician infers a reopened weighing, predicting that monitoring or a protective co-intervention may pull the balance back to favourable and license use under safeguards. So the reasoner does not merely register "don't" but classifies the prohibition's force, and reasons differently downstream: honour-the-flag versus perform-the-calculation-with-eyes-open.

The interventionist move is fallback selection: triggering a contraindication is simultaneously a prediction that an alternative is needed, so the clinician reasons from "first-line class is contraindicated here" to "substitute a class whose balance stays favourable under this condition" — the prohibition is generative of the next action, not merely a stop. And the concept draws a clean scope line the clinician uses to reject mis-modelled cases: a contraindication is the specific patient context in which a favourable balance flips, not a general side effect, not a dose-range question, and not the absence of any useful effect — so the reasoner declines to treat an ordinary adverse-effect profile or a narrow therapeutic window as a contraindication, reserving the sign-reversal logic for the condition-triggered case a point-of-care alert is actually built to fire on.

Knowledge Transfer¶

Within clinical medicine the concept transfers as full mechanism — the three-part structure (default indication, conditioning context, prohibition rule), the absolute–relative gradient, the sign-of-the-balance inference, the default-with-exceptions heuristic, and the fallback-selection move all carry intact, together with the codification apparatus that operationalizes them. They move without translation across pharmacology (drug–disease, drug–drug, and drug–food contraindications: beta-blockers in asthma, ACE inhibitors in pregnancy, MAOI + tyramine, warfarin + certain antibiotics), surgical and procedural medicine (active infection, severe coagulopathy, patient refusal), imaging (gadolinium in renal failure), and vaccination (anaphylaxis history; live vaccines under severe immunocompromise) — all encoded uniformly in formularies, drug labelling, and prescription-screening systems, so a contraindication established for one treatment class is checked against a patient record by the same machinery as any other. The kin entries — drug–drug interaction, drug–disease interaction, black-box warning — are the same medical-domain codification of condition-triggered risk-reversal, and the mechanism is genuinely shared across them.

Outside medicine the honest reading is mostly case (A) shape-borrowing, with the portable mechanism belonging to parent primes (B). Engineering do-not-operate placards and safe-operating limits, software/security do-not-deploy lists and feature-flag exclusions, and policy "generally-permitted-but-prohibited-under-condition-X" carve-outs all share the structural shape of a conditional constraint with a sign-flip — but each domain has its own native vocabulary and operational apparatus (safe-operating limits, guardrails/exclusions, regulatory carve-outs), and "contraindication" is not the name any of them uses; there is no substrate-independent intervention vocabulary that ports without the medical-pharmacology framing. What genuinely travels is the structural residue the concept composes: constraint supplies the prohibition, general conditional logic ("if condition then prohibit") supplies the trigger, risk supplies the harm-benefit reasoning that motivates the rule, and exclusion/gatekeeping supplies the population-filtering. The most portable refinement is the absolute-versus-relative gradient, which sits naturally as a sub-pattern within constraint. So the honest cross-domain lesson should carry constraint + risk + exclusion (with the conditional-trigger and absolute/relative distinctions), and not "contraindication" as named — its distinctive cargo (the FDA/BNF labelling apparatus, structured formulary fields, the clinical-reasoning protocol, the decision-support implementations) is the medical codification and stays home. See Structural Core vs. Domain Accent.

Examples¶

Canonical¶

ACE inhibitors (e.g., lisinopril, enalapril) are a first-line, generally-favorable therapy for hypertension and heart failure: for the typical adult, expected benefit clearly exceeds risk. In pregnancy — particularly the second and third trimesters — that balance inverts. The same mechanism (angiotensin-converting-enzyme blockade) that helps the typical patient impairs the fetal renin-angiotensin system, causing reduced fetal renal perfusion, oligohydramnios, skull ossification defects, and neonatal renal failure. Pregnancy therefore fires an absolute contraindication: the drug is withdrawn and a pregnancy-compatible antihypertensive (labetalol, methyldopa, nifedipine) substituted. The mechanism of the drug is unchanged; what changed is the patient context against which its harm and benefit are scored.

Mapped back: Hypertension therapy is the default indication; pregnancy is the conditioning context whose presence produces the risk-benefit sign-reversal — fetal toxicity now outweighs maternal benefit. Because no safeguard rescues the drug here, the absolute/relative gradient reads absolute, a flat prohibition rule; and the switch to labetalol or methyldopa is the fallback the prohibition generates.

Applied / In Practice¶

Modern hospitals operationalize the contraindication catalog through computerized provider order entry (CPOE) with clinical decision support. When a clinician orders a non-selective beta-blocker (propranolol) for a patient whose electronic record carries an active asthma diagnosis, the system runs the order against the coded drug–disease exception table and surfaces a point-of-care alert: beta-blockade risks bronchospasm by antagonizing β2 receptors in airway smooth muscle. The prescriber then either honors the flag or, treating it as relative, switches to a cardioselective agent or a different antihypertensive class with monitoring.

Mapped back: The order-entry system is the codification apparatus checking the patient record against the enumerated conditioning context (asthma) rather than relying on the prescriber's memory. The alert encodes the risk-benefit sign-reversal (bronchospasm), and because cardioselective use may still be permitted with care, the case sits on the relative side of the absolute/relative gradient — a reopened weighing whose fallback is class substitution.

Structural Tensions¶

T1: Absolute prohibition versus reopened weighing (the gradient that both closes and reopens). The absolute–relative gradient is the concept's sharpest tool: it tells the clinician what kind of decision is owed the instant a condition fires — a flat prohibition to honour, or a risk calculation to perform with eyes open. But the two branches have opposite computability. An absolute contraindication terminates the question and is fully machine-checkable: the system fires the flag and the answer is settled. A relative contraindication reopens the question as a judgment no catalog can close, requiring the clinician to weigh monitoring and co-intervention against the worsened balance. So the same codification that discharges absolutes cannot decide relatives — it can only flag them and hand the weighing back. The gradient's value in classifying the decision is inseparable from the fact that half of it defeats the automation the concept otherwise enables. Diagnostic: Is this contraindication one the system can settle by firing a prohibition, or one it can only surface for a weighing the clinician must still perform?

T2: Sparse enumerable exceptions versus completeness (the compression is only as good as the list). The whole efficiency of contraindication rests on a structural bet: a treatment's balance is favourable across nearly the entire patient-state space and inverts only in a sparse, enumerable exception set, so the vast favourable region can be left uninspected. That bet buys the machine-checkable lookup — but it is only as sound as the enumeration. A sign-flipping condition, drug–drug combination, or rare physiological state not on the list falls into the "uninspected because guaranteed safe" region, where the very structure that makes the concept efficient suppresses the check that would have caught it. The compression's power and its blind spot are the same move: trusting that nothing outside the table changes the sign. Diagnostic: Is this patient's context actually covered by the enumerated exception set, or is it an unlisted combination the default-with-exceptions structure is silently treating as favourable?

T3: Codified flag versus alert fatigue (checkability breeds a new failure). Factoring the reasoning into indication, conditioning context, and prohibition rule is exactly what lets a prescription-screening system check the patient record instead of the prescriber's memory — the concept's operational triumph. But that same codification, applied across every drug–disease and drug–drug pair, floods point-of-care with flags, many of them relative, low-relevance, or already known to the clinician. The result is the well-documented erosion where alerts are overridden reflexively and the one consequential contraindication is dismissed with the ninety trivial ones. The mechanism that guarantees no sign-reversal goes unflagged also guarantees that flags lose their signal value through sheer volume. Comprehensiveness and attention are in direct tension. Diagnostic: Is this flag surfacing a genuine sign-reversal that changes the decision, or adding to a stream of alerts whose volume is training the prescriber to dismiss them?

T4: Prohibition-as-generative versus the stranded patient (the fallback is presumed, not guaranteed). A contraindication is not a mere stop: it is generative, predicting that an alternative is needed and pointing the clinician toward a class whose balance stays favourable under the conditioning context — pregnancy fires the ACE-inhibitor prohibition and summons labetalol or methyldopa. The concept's usefulness leans on this fallback being available. But nothing in the structure guarantees one exists. When the contraindicated treatment is the only effective option and no substitute keeps a favourable balance under the condition, the sign-reversal logic delivers a clean prohibition and no path, stranding the patient between an inadvisable treatment and no treatment. The generative reading is a strength exactly where a fallback exists and silent about the cases where it does not. Diagnostic: Does a class with a favourable balance under this condition actually exist, or does honouring the contraindication leave the patient with no adequately effective option?

T5: Autonomy versus reduction (a named medical rule or a composition of constraint, risk, and exclusion). "Contraindication" is a fully developed clinical mechanism, and within medicine it transfers as full mechanism — the three-part structure, the absolute/relative gradient, the sign-of-the-balance inference, all riding the same formulary, labelling, and screening apparatus across pharmacology, surgery, imaging, and vaccination. But off the medical substrate what travels is not the named concept: engineering do-not-operate placards, software do-not-deploy exclusions, and policy carve-outs share only the shape of a conditional constraint with a sign-flip, and each uses its own native vocabulary. The portable residue is the parent composition — constraint supplies the prohibition, conditional logic the trigger, risk the harm-benefit weighing, exclusion/gatekeeping the population filter, with the absolute/relative gradient sitting naturally inside constraint. The concept's distinctive cargo (FDA/BNF labelling, structured formulary fields, decision-support implementations) is medical codification and stays home. Diagnostic: Resolve toward constraint + risk + exclusion when carrying the conditional-prohibition shape to engineering, software, or policy; toward "contraindication" when diagnosing a condition-triggered risk-benefit sign-reversal inside clinical medicine.

Structural–Framed Character¶

Contraindication sits at the framed-leaning position — well onto the framed side because it is a normatively charged, practice-and-institution-constituted clinical rule, though held just off the framed pole because the sign-reversal it flags is anchored in a real, observer-free physiological harm rather than being pure verdict. On evaluative weight it reads strongly framed: to call a treatment "contraindicated" is to render a normative decision — do not give — not to describe a neutral mechanism; the concept exists precisely to fire a prohibition, and its whole content is a prescription about what a clinician ought to do. On human-practice-bound it is framed: the concept is constituted by the practice of medicine — a default indication established for a population, a clinician weighing a risk-benefit balance, a fallback to substitute — and it dissolves without that practice, since off the clinical stage there is no "indication" for the condition to invert. On institutional origin it is emphatically framed: the entry's distinctive cargo is codification apparatus — formularies, FDA/BNF drug labelling, structured formulary fields, prescription-screening and CPOE decision-support — an artifact of a specific medical-regulatory tradition, and even the absolute/relative gradient is an engineered gradation drawn inside that tradition. On vocab-travels it is framed (the operative terms are pharmacological), and on import-vs-recognize the entry is explicit that engineering do-not-operate placards, software do-not-deploy lists, and policy carve-outs share only the shape and use their own native vocabulary, so the cross-domain move is analogy carried by parent primes, not recognition of "contraindication."

What tempers the placement toward framed-leaning rather than the pole is that, unlike a pure fallacy-verdict, contraindication sits on top of a genuinely mechanistic fact: under the conditioning context the harm-benefit balance really inverts, and it does so through observer-free physiology — ACE inhibitors impair the fetal renin-angiotensin system, beta-blockade provokes bronchospasm, whether or not anyone codes the rule. But that underlying harm is a side-effect mechanism, not the concept; "contraindication" is the codified, condition-triggered prohibition built on top of it, and that is the framed part. The one portable structural skeleton is the conditional constraint with a sign-flip — the entry's parent composition of constraint (the prohibition), conditional logic (the trigger), risk (the harm-benefit weighing), and exclusion/gatekeeping (the population filter), with the absolute/relative gradient sitting naturally inside constraint. That composition is what contraindication instantiates and what actually travels to engineering, software, and policy; the medical codification — the labelling apparatus, the formulary fields, the clinical-reasoning protocol — stays home. Its character: a normatively charged, institution-codified clinical prohibition whose portable core is a conditional risk-constraint borrowed from its parent composition, framed in every feature that makes it specifically medical.

Structural Core vs. Domain Accent¶

This section decides why contraindication is a domain-specific abstraction and not a prime, and it carries the case for its domain-specificity — so it is worth being exact about what could lift and what stays home. Its skeleton is genuinely a composition: contraindication instantiates several primes at once rather than one.

What is skeletal (could lift toward a cross-domain prime). Strip the medicine and a thin relational structure survives: a generally-permitted action is prohibited when a specific triggering condition is present, because under that condition its cost-benefit balance flips sign. The portable pieces are abstract, and they resolve into a composition of catalogued primes: constraint supplies the prohibition itself; a conditional "if condition then prohibit" trigger supplies the gating; risk supplies the harm-benefit weighing that motivates the rule; and exclusion / gatekeeping supplies the filtering of a sparse sub-population out of the default. The most portable refinement — the absolute-versus-relative gradation (a flat prohibition versus a reopened weighing under safeguards) — sits naturally as a sub-pattern inside constraint. That composition is genuinely substrate-portable and recurs as a conditional-constraint-with-sign-flip in engineering safe-operating limits, software do-not-deploy exclusions, and policy carve-outs. But it is the core contraindication shares, not what makes it contraindication.

What is domain-bound. Almost everything that makes the entry contraindication in particular is clinical-medicine furniture, and none of it survives extraction. The default is a treatment's indication established for a patient population; the trigger is a patient condition, diagnosis, physiological state, or co-treatment; the sign-reversal runs through observer-free physiology (fetal renin-angiotensin impairment, β2-mediated bronchospasm); and the whole rule is operationalized by a specific codification apparatus — formularies, FDA/BNF drug labelling, structured formulary fields, prescription-screening and CPOE decision-support that flag the patient record. The decisive test: engineering do-not-operate placards, software do-not-deploy lists, and policy carve-outs share only the shape of a conditional constraint with a sign-flip, and each uses its own native vocabulary (safe-operating limits, guardrails, regulatory carve-outs) — "contraindication" is the name none of them uses. Remove the medical practice — the indication, the risk-benefit weighing against a patient, the fallback class, the labelling apparatus — and there is no indication for the condition to invert, only the bare conditional constraint. The concept is constituted by the very clinical codification the prime bar would ask it to shed.

Why this does not clear the prime bar. A prime is a relational structure whose vocabulary travels and whose cross-domain transfer is recognition of the same mechanism, not analogy. Contraindication's transfer is bimodal. Within clinical medicine it travels as full mechanism — the three-part structure (default indication, conditioning context, prohibition rule), the absolute/relative gradient, the sign-of-the-balance inference, and the fallback-selection move carry intact across pharmacology, surgery, imaging, and vaccination, all checked by the same formulary-and-screening apparatus, with drug–drug interaction and black-box warning as kin codifications. Beyond medicine what travels is not the named concept but the shape: each domain has its own vocabulary and operational apparatus, so calling an engineering interlock or a policy carve-out a "contraindication" borrows the conditional-constraint form while renaming every component — analogy, not the same mechanism. When the bare structural lesson — prohibit the generally-permitted act under the sign-flipping condition — is needed cross-domain, it is already carried, in more general form, by constraint + risk + exclusion (with the conditional trigger and the absolute/relative gradient inside constraint). The cross-domain reach belongs to that composition; "contraindication," as named, carries the labelling apparatus, formulary fields, and clinical-reasoning protocol that keep it a medical rule rather than a free-floating prime.

Relationships to Other Abstractions¶

Current abstraction Contraindication Domain-specific

Parents (2) — more general patterns this builds on

Contraindication presupposes Risk Prime

Contraindication presupposes Risk because its trigger is defined by a context- specific probability-and-severity shift that reverses expected net benefit.
Contraindication is a decomposition of Constraint Prime

Removing clinical vocabulary leaves a conditional constraint that prohibits a generally permitted action when a named context flips its balance against use.

Hierarchy paths (4) — routes to 4 parentless roots

Contraindication → Risk → Uncertainty

Show alternative paths (3)

Not to Be Confused With¶

Side effect / adverse effect. An undesirable outcome a treatment may produce. A contraindication is not the harm itself but the specific patient condition under which an otherwise-favourable risk-benefit balance flips sign. Every drug has side effects without being contraindicated; the side effect is the mechanism, the contraindication is the context that makes giving the drug inadvisable. Tell: Are you naming a harm the treatment can cause (side effect), or the patient condition that inverts the balance so it should not be given at all (contraindication)?
Drug–drug / drug–disease interaction. The kin codifications of condition-triggered risk in medicine — an interaction is precisely a conditioning context (another drug, a disease) that can fire a contraindication. They are the same medical apparatus, often the trigger rather than a distinct concept; a contraindication is the resulting prohibition rule, an interaction is one class of thing that raises it. Tell: Is the focus the co-present agent or state that alters the balance (interaction, the trigger), or the codified do-not-give rule it produces (contraindication)?
Black-box warning. The most severe labelling alert — flagging a serious risk that demands caution but does not, by itself, prohibit use. A contraindication is the actual prohibition (absolute) or reopened weighing (relative). A boxed warning heightens vigilance; a contraindication withholds. Tell: Does the label demand heightened caution while still permitting the drug (black-box warning), or does it bar use under the triggering condition (contraindication)?
Narrow therapeutic window. A dose-range problem — too little does nothing, too much harms — within a treatment that still applies; the fix is calibrating the amount. A contraindication is a context question: under the triggering condition there is no favourable use at any dose. Tell: Is the issue getting the dose right within a usable range (therapeutic window), or that no dose is advisable given the patient's condition (contraindication)?
Mere absence of benefit / ineffectiveness. A treatment that simply does nothing useful here. A contraindication is an active sign-reversal — expected harm now exceeds expected benefit — not a neutral zero. Tell: Does the treatment merely fail to help (ineffectiveness), or has the balance turned actively against use (contraindication)?
The conditional-constraint composition (constraint + risk + exclusion). The substrate-neutral parent shape — a generally-permitted action prohibited when a triggering condition flips its cost-benefit sign, with the absolute/relative gradient sitting inside constraint. Engineering safe-operating limits, software do-not-deploy exclusions, and policy carve-outs are co-instances under their own native vocabulary. Contraindication is the clinical-medicine instance. Tell: Is the setting a treatment weighed against a patient with formulary/labelling codification (contraindication), or a conditional prohibition in engineering, software, or policy (the constraint + risk + exclusion composition, which carries the cross-domain lesson)?

Neighborhood in Abstraction Space¶

Contraindication sits in a sparse region of the domain-specific corpus (97^th percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (309 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-07-12