# SBA generator mistake bank

Lightweight record of recurring clinical and SBA generator/reviewer issues found during testing.

This document is **not** a rules engine. It does **not** change generation, validation, or review behaviour. Use it to track patterns for later prioritisation and rule design.

**Active in combined AI tutor reviewer** (`prompts/sba_combined_tutor_reviewer_prompt.php`):  
`wrong_emergency_pathway_imported`, `justification_invents_unsupported_diagnosis`, `copd_uncontrolled_oxygen_key`, `oxygen_prescription_missing_target_range`, `condition_stage_ambiguity`, `result_key_contradiction`, `specialist_monitoring_test_keyed_routinely`, `anticoagulation_pathway_ambiguity`, `acute_followup_context_mismatch`, consolidated `emergency_bundle_completeness` (missing essential / watchlist / safe-but-incomplete), merged `asthma_step_mismatch` patterns, plus existing quick-reference items (overspecific diagnosis, unstable ACS, batch repetition, justification–options mismatch, etc.). Repair-classification patterns remain **out of scope** for that prompt.

---

## Entries

### Essential component omitted from acute management

| Field | Detail |
|-------|--------|
| **Short code** | `acute_management_missing_essential_component` |
| **Affected skill** | Emergency Management |
| **Affected conditions so far** | Community-acquired pneumonia |
| **Potentially affected conditions** | Sepsis, asthma, anaphylaxis, pulmonary embolism, ACS, DKA, meningitis |
| **Current status** | Recorded for later review. Do not fix immediately unless the pattern recurs or is prioritised. |

**Example**

In a CAP Emergency Management item, the patient had oxygen saturation 88% breathing air, hypotension and severe pneumonia. The keyed answer included IV fluids and broad-spectrum IV antibiotics but omitted oxygen therapy, while a distractor included oxygen.

**Why it matters**

The key is clinically incomplete and creates one-best-answer unfairness.

---

### Acute management bundle incompleteness (watchlist)

| Field | Detail |
|-------|--------|
| **Short code** | `acute_management_bundle_incompleteness_watchlist` |
| **Severity** | Watchlist / moderate |
| **Affected skill** | Emergency Management |
| **Affected conditions so far** | Acute severe asthma; COPD batch (exacerbation items) |
| **Potentially affected conditions** | COPD, anaphylaxis, sepsis, DKA |
| **Current status** | Watchlist. Do not fix yet. Check whether this repeats across exacerbation skills. |

**Example (COPD batch)**

COPD exacerbation item keyed on bronchodilators but did not include systemic corticosteroids in the acute bundle, while other options implied incomplete exacerbation treatment.

**Why it matters**

The keyed answer may be a clinically incomplete acute bundle — missing a standard component tutors or candidates expect for the severity in the stem.

---

### Reviewer repair overrides tutor-intended distractor

| Field | Detail |
|-------|--------|
| **Short code** | `repair_overrides_tutor_intended_distractor` |
| **Severity** | Medium |
| **Affected skill** | Prescribing / monitoring (and similar misconception-testing items) |
| **Affected conditions so far** | Rivaroxaban monitoring SBA |
| **Potentially affected conditions** | Any item where tutor comments specify deliberate “wrong but tempting” distractors |
| **Current status** | Recorded for later. Do not urgently fix unless it recurs. |

**Example**

Tutor comments asked for INR or PT as distractors in a rivaroxaban monitoring SBA. Reviewer 1 (AI repair) replaced INR, PT and aPTT with more clinically relevant distractors, removing the deliberate safety-check distractor.

**Why it matters**

The repaired question may be clinically better, but it no longer tests the specific misconception the tutor requested.

---

### Repair changes educational content but labels it as style

| Field | Detail |
|-------|--------|
| **Short code** | `repair_content_change_labelled_style` |
| **Severity** | Medium |
| **Affected stage** | AI repair pipeline |
| **Affected skill** | Any (notably distractor design, Diagnosis, Investigation / interpretation) |
| **Affected conditions so far** | COPD prescribing batch; DVT / VTE batch (diagnosis, ultrasound interpretation) |
| **Current status** | **Recurring** after COPD and VTE batches. Repair classification tightened again (aggregated educational fingerprint + token-overlap fallback). Monitor in use. |

**Examples**

- **COPD batch:** pulmonary fibrosis → interstitial lung disease; pneumonia → anaphylaxis; budesonide → oral corticosteroids; theophylline → LAMA — all mislabelled **style**.
- **DVT / VTE batch (repair mislabelled style):** lymphoedema → superficial thrombophlebitis; Baker's cyst rupture → osteomyelitis; chronic venous insufficiency → peripheral arterial occlusion; proximal deep vein thrombosis → acute limb ischaemia; normal venous ultrasound → post-thrombotic syndrome; full blood count → prothrombin time; liver function tests → activated partial thromboplastin time.

**Why it matters**

The tutor may think the repair only changed wording, when it actually changed the **clinical distinction** being tested, even when the key letter is unchanged.

---

### Condition stage ambiguity

| Field | Detail |
|-------|--------|
| **Short code** | `condition_stage_ambiguity` |
| **Severity** | Medium |
| **Affected skill** | Prescribing (stable vs exacerbation vs rescue) |
| **Affected conditions so far** | COPD prescribing batch |
| **Potentially affected conditions** | Asthma, heart failure, diabetes, CKD |
| **Current status** | Recorded for later. Do not fix immediately unless prioritised. |

**Example (COPD batch)**

COPD item described increasing breathlessness and salbutamol no longer sufficient, but it was unclear whether the question tested stable COPD escalation, acute exacerbation treatment, or immediate symptom relief.

**Why it matters**

The best medication depends on which clinical stage the stem is targeting.

---

### Result key contradiction

| Field | Detail |
|-------|--------|
| **Short code** | `result_key_contradiction` |
| **Severity** | Medium–high |
| **Affected stage** | Generation, review, or repair |
| **Affected conditions so far** | COPD prescribing batch |
| **Current status** | Recorded for later. Watch in validator and repair keyed-answer checks. |

**Example (COPD batch)**

Keyed answer option stated **start tiotropium** (LAMA) but the justification emphasised **oral corticosteroids for exacerbation**, or the stem implied acute exacerbation while the key reflected maintenance therapy only.

**Why it matters**

Candidates cannot reconcile the stem, key, and justification — fairness and teaching value are reduced.

---

### Monitoring target too narrow

| Field | Detail |
|-------|--------|
| **Short code** | `monitoring_target_too_narrow` |
| **Severity** | Medium |
| **Affected skill** | Prescribing / monitoring |
| **Affected conditions so far** | COPD batch (when monitoring is tested); anticoagulant items |
| **Current status** | Recorded for later. |

**Example (COPD batch)**

Item tested monitoring of COPD therapy but options listed only pulse oximetry or peak flow, omitting other reasonable monitoring (symptoms, exacerbations, inhaler technique, FEV1 review) so distractors were not homogeneously plausible.

**Why it matters**

The option set may not fairly test the breadth of monitoring the lead-in implies.

---

### Distractor overlaps with keyed condition

| Field | Detail |
|-------|--------|
| **Short code** | `distractor_key_overlap` |
| **Severity** | Medium |
| **Affected skill** | Diagnosis / differential diagnosis |
| **Affected conditions so far** | COPD diagnosis batch |
| **Potentially affected conditions** | Any item where distractors are drawn from the same disease spectrum as the key (e.g. COPD, chronic bronchitis, emphysema) |
| **Current status** | Recorded for later. Do not fix now unless repeated. |

**Example (COPD batch)**

COPD diagnosis item replaced “asthma exacerbation” with “chronic bronchitis”, but chronic bronchitis can overlap conceptually with COPD.

**Why it matters**

A distractor may be clinically related to, part of, or a subtype of the keyed diagnosis, reducing one-best-answer fairness.

---

### Non-condition topic causes skill drift

| Field | Detail |
|-------|--------|
| **Short code** | `non_condition_topic_skill_drift` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected stage** | Generation (condition / skill selection) |
| **Affected topics so far** | Oxygen therapy (with Diagnosis skill) |
| **Potentially affected topics** | Other treatment-principle playbook topics (e.g. fluids, analgesia principles) paired with disease-specific skills |
| **Current status** | Recorded for later review. Later: restrict skills for treatment-principle topics. |

**Example (oxygen therapy batch)**

Oxygen therapy used as the condition with **Diagnosis** selected. The generated item keyed **pneumothorax**, so the actual topic became pneumothorax diagnosis rather than oxygen therapy.

**Why it matters**

Some playbook topics are treatment principles rather than diseases. They may not support all skills, especially Diagnosis — generation drifts to a concrete disease that fits the skill.

---

### Uncontrolled oxygen keyed in COPD risk

| Field | Detail |
|-------|--------|
| **Short code** | `copd_uncontrolled_oxygen_key` |
| **Severity** | High (if repeated) |
| **Batch** | Oxygen therapy |
| **Affected skills** | Emergency Management, Prescribing, Management |
| **Affected conditions so far** | COPD (emergency / oxygen therapy items) |
| **Potentially affected conditions** | Other hypercapnia-prone presentations where controlled oxygen is standard |
| **Current status** | Recorded for later review. High priority if repeated. |

**Example (oxygen therapy batch)**

COPD emergency item with SpO₂ 84%, confusion and cyanosis keyed oxygen via **non-rebreather mask** rather than controlled oxygen with target saturations, ABG, NIV, or escalation.

**Why it matters**

High-concentration oxygen can worsen hypercapnia in COPD. A question on oxygen therapy should test controlled oxygen targets unless immediate life-threatening hypoxia is explicitly framed.

---

### Safety reviewer misses oxygen-target nuance

| Field | Detail |
|-------|--------|
| **Short code** | `safety_reviewer_misses_oxygen_target_nuance` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected stage** | Clinical safety reviewer (Reviewer 2) |
| **Related codes** | `copd_uncontrolled_oxygen_key` |
| **Affected conditions so far** | COPD with hypoxia / confusion |
| **Current status** | Recorded for later review. **Safety reviewer prompt updated** with oxygen/COPD checks; monitor in use. |

**Example (oxygen therapy batch)**

Clinical safety reviewer accepted **non-rebreather oxygen** as the key in COPD with hypoxia and confusion without commenting on **controlled oxygen**, **target saturations**, or whether **ABG / NIV / escalation** should be considered.

**Why it matters**

Reviewer 2 may miss important condition-specific safety principles unless prompted or supported by rules.

---

### Repair introduces plausible competing answer

| Field | Detail |
|-------|--------|
| **Short code** | `repair_introduces_competing_answer` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected stage** | AI repair pipeline (Reviewer 1) |
| **Affected skills** | Diagnosis, Investigation / interpretation |
| **Affected conditions so far** | Oxygen saturation interpretation (pneumonia vs ARDS) |
| **Current status** | Recorded for later review. Medium priority. |

**Example (oxygen therapy batch)**

Oxygen saturation interpretation item keyed **acute hypoxaemia due to pneumonia**. Repair changed a distractor to **acute respiratory distress syndrome**, making the item borderline because bilateral infiltrates and severe hypoxaemia can support ARDS.

**Why it matters**

Reviewer 1 may strengthen distractors too much and create multiple defensible answers.

---

### Oxygen prescription lacks target saturation

| Field | Detail |
|-------|--------|
| **Short code** | `oxygen_prescription_missing_target_range` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected skill** | Prescribing |
| **Affected conditions so far** | COPD (oxygen prescribing items) |
| **Potentially affected conditions** | Other hypercapnia-risk or long-term oxygen therapy items |
| **Related codes** | `copd_uncontrolled_oxygen_key`, `safety_reviewer_misses_oxygen_target_nuance` |
| **Current status** | Recorded for later review. |

**Example (oxygen therapy batch)**

COPD oxygen prescribing item keyed **oxygen 2 L/min via nasal cannula** but did not include **target oxygen saturation range** (e.g. 88–92%).

**Why it matters**

Oxygen should be prescribed to a target range, particularly in COPD and risk of hypercapnic respiratory failure.

---

### Oxygen delivery changes labelled as style

| Field | Detail |
|-------|--------|
| **Short code** | `oxygen_delivery_change_labelled_style` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected stage** | AI repair pipeline |
| **Related codes** | `copd_uncontrolled_oxygen_key`, `oxygen_prescription_missing_target_range`, `repair_content_change_labelled_style` |
| **Affected skills** | Prescribing, Emergency Management, Management |
| **Current status** | Recorded for later review. **Repair classifier** covers device, flow, target range, ABG/NIV/escalation; monitor in use. |

**Example (oxygen therapy batch)**

Repair changed oxygen **flow rates** and **devices** in distractors (e.g. nasal cannula 2 L/min → non-rebreather 15 L/min) or removed **target SpO₂** / **ABG** wording but labelled changes as **style**-only.

**Why it matters**

Oxygen device and flow rate are clinical prescribing content, not wording style.

---

### Acute-follow-up context mismatch

| Field | Detail |
|-------|--------|
| **Short code** | `acute_followup_context_mismatch` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected skill** | Monitoring, Investigation, Management |
| **Affected conditions so far** | COPD / oxygen monitoring items |
| **Potentially affected conditions** | Any item mixing acute presentation language with outpatient follow-up lead-ins |
| **Related codes** | `condition_stage_ambiguity` |
| **Current status** | Recorded for later review. |

**Example (oxygen therapy batch)**

Monitoring item stem describes **acute breathlessness and mild confusion**, but the lead-in asks about **follow-up monitoring** (outpatient), so the time frame is unclear.

**Why it matters**

The time frame of the stem and lead-in may not match, making it unclear whether the question tests acute assessment or outpatient monitoring.

---

### Safe but incomplete emergency key

| Field | Detail |
|-------|--------|
| **Short code** | `safe_but_incomplete_emergency_key` |
| **Severity** | Medium |
| **Batch** | Oxygen therapy |
| **Affected skills** | Emergency Management, Management |
| **Affected conditions so far** | COPD oxygen emergency |
| **Potentially affected conditions** | Asthma, sepsis, DKA, PE, anaphylaxis |
| **Related codes** | `copd_uncontrolled_oxygen_key`, `acute_management_bundle_incompleteness_watchlist` |
| **Current status** | Recorded for later review. Do not fix all similar cases yet; watch for repetition across COPD, asthma, sepsis, DKA, PE and anaphylaxis. |

**Example (oxygen therapy batch)**

COPD oxygen emergency item keyed **controlled oxygen targeting 88–92%** and **reassessment**, but the stem included **confusion, cyanosis, SpO₂ 85%, accessory muscle use and exhaustion**. Reviewer 2 correctly flagged missing **ABG / NIV / escalation** considerations.

**Why it matters**

The keyed answer is not unsafe, but may be too narrow for the severity of the emergency stem, creating borderline one-best-answer fairness.

---

### Inappropriate specialist monitoring test keyed

| Field | Detail |
|-------|--------|
| **Short code** | `specialist_monitoring_test_keyed_routinely` |
| **Severity** | High (if repeated) |
| **Affected skills** | Monitoring, Prescribing |
| **Affected condition** | Venous thromboembolic disease (DVT / PE) |
| **Potentially affected conditions** | DOAC monitoring items; other drugs where anti-Xa, drug levels, or specialist assays are exceptional not routine |
| **Related codes** | `monitoring_target_too_narrow` |
| **Batch** | VTE / DVT |
| **Current status** | Recorded for later review. High priority if repeated. |

**Example (DVT batch)**

DVT monitoring item keyed **anti-Xa level** for routine follow-up on **low-molecular-weight heparin**, without renal impairment, extremes of body weight, pregnancy, bleeding concern, or other special indication. Repair also mislabelled **full blood count → prothrombin time** as style-only.

**Why it matters**

A specialist or exceptional monitoring test may be incorrectly presented as routine follow-up, creating a clinically misleading key.

**Reference pathway (faculty notes — not a rules engine)**

Use to align stem, key, and monitoring with the clinical stage:

| Clinical stage | Typical management focus |
|----------------|-------------------------|
| **Imaging delayed** | **Interim therapeutic anticoagulation** |
| **Confirmed DVT, no contraindications** | **DOAC** (e.g. apixaban / rivaroxaban) may be appropriate depending on local pathway |
| **Renal impairment, pregnancy, cancer, APS, or contraindication** | **Alternative anticoagulant pathway** may apply |

Routine LMWH follow-up monitoring is usually **clinical review, renal function, bleeding, adherence, weight-appropriate dosing** — **not** routine anti-Xa unless a special indication is in the stem.

---

## VTE / DVT reference pathway (faculty notes)

Lightweight anchor for generation, review, and safety checks on venous thromboembolic disease items. **Not** enforced automatically.

- **If imaging is delayed:** interim **therapeutic anticoagulation**
- **If confirmed DVT and no contraindications:** **DOAC** (e.g. apixaban / rivaroxaban) may be appropriate depending on **local pathway**
- **If renal impairment, pregnancy, cancer, APS, or contraindication:** **alternative anticoagulant pathway** may apply

When writing **Monitoring** or **Prescribing** SBAs, match the keyed answer to the **stage in the stem** (suspected vs confirmed vs special population). Do not key **anti-Xa**, **INR**, or **drug levels** as routine follow-up without stem justification.

---

### Anticoagulation pathway ambiguity

| Field | Detail |
|-------|--------|
| **Short code** | `anticoagulation_pathway_ambiguity` |
| **Severity** | Medium |
| **Affected skills** | Prescribing, Management, Monitoring |
| **Affected condition** | Venous thromboembolic disease (DVT / PE) |
| **Related codes** | `specialist_monitoring_test_keyed_routinely`; VTE / DVT reference pathway (faculty notes) |
| **Batch** | VTE / DVT |
| **Current status** | Recorded for later review. Likely worth tutor comments or condition overlay later. |

**Example (DVT batch)**

DVT prescribing item keyed **enoxaparin** as first treatment after **confirmed DVT**, but **DOACs** such as apixaban / rivaroxaban may be suitable first-line options depending on current / local guidance and patient factors.

**Why it matters**

The best anticoagulant depends on whether the case is **suspected vs confirmed** DVT, whether **imaging is delayed**, **renal function**, **pregnancy**, **cancer**, **APS**, **bleeding risk**, and **local formulary**.

---

### Anatomical classification contradiction

| Field | Detail |
|-------|--------|
| **Short code** | `anatomical_classification_contradiction` |
| **Severity** | Medium |
| **Batch** | VTE / DVT (ultrasound interpretation) |
| **Affected skills** | Investigation, Diagnosis |
| **Affected conditions so far** | Venous thromboembolic disease |
| **Related codes** | `result_key_contradiction`, `repair_content_change_labelled_style` |
| **Current status** | Recorded for later review. Fix if repeated in VTE items. |

**Example (DVT batch)**

Ultrasound item described a thrombus in the **popliteal vein** but stated **no proximal DVT**, even though popliteal DVT is commonly classified as **proximal DVT**. Related repair swaps (e.g. **proximal DVT → acute limb ischaemia**) must not be labelled style-only.

**Why it matters**

The stem may contradict standard anatomical classification, making the keyed interpretation unfair.

---

### Wrong emergency pathway imported into non-infective emergency

| Field | Detail |
|-------|--------|
| **Short code** | `wrong_emergency_pathway_imported` |
| **Severity** | High (if repeated) |
| **Affected skill** | Emergency Management |
| **Affected conditions so far** | Aortic dissection, acute coronary syndrome |
| **Potentially affected conditions** | Pulmonary embolism, stroke, DKA, anaphylaxis, other non-infective emergencies |
| **Current status** | Recurring. High priority for later fix. |

**Examples**

- ACS emergency-management item keyed IV fluids and broad-spectrum IV antibiotics despite a stem suggesting acute coronary syndrome and no infection cues.
- Aortic dissection emergency item keyed IV fluids and broad-spectrum antibiotics despite sudden tearing chest pain radiating to the back and no infection cues.

**Why it matters**

Generator may import sepsis-style emergency management into another acute presentation, creating an unsafe key.

---

### Justification invents unsupported diagnosis

| Field | Detail |
|-------|--------|
| **Short code** | `justification_invents_unsupported_diagnosis` |
| **Severity** | High |
| **Affected stage** | Generation / justification writing |
| **Affected conditions so far** | Acute coronary syndrome emergency items; aortic dissection emergency items |
| **Current status** | Recurring. High priority. |

**Examples**

- ACS emergency item justification described potential sepsis/infectious shock, but the stem had no fever, infection source, or sepsis features.
- Aortic dissection emergency item justification said fluids and antibiotics were needed to address potential sepsis, but the stem had no fever, infection source, or sepsis features.

**Why it matters**

The justification can rationalise an unsafe key by adding diagnoses not present in the stem.

---

### Unstable ACS keyed as antiplatelet-only management

| Field | Detail |
|-------|--------|
| **Short code** | `unstable_acs_antiplatelet_only_key` |
| **Severity** | High |
| **Affected skill** | Emergency Management |
| **Affected conditions so far** | Acute coronary syndrome (unstable / high-risk presentation) |
| **Potentially affected conditions** | NSTEMI, STEMI pathway items, other chest-pain emergencies where haemodynamic instability is in the stem |
| **Related codes** | `acute_management_missing_essential_component`, `wrong_emergency_pathway_imported` |
| **Current status** | Caught by Reviewer 1 and Reviewer 2. **High-priority generator issue** — do not downgrade; monitor recurrence. |

**Example**

Emergency Management ACS item keyed **aspirin and clopidogrel** in a patient with **BP 90/60** and **pulse 120**, where urgent escalation, resuscitation, and reperfusion assessment (and haemodynamic support as appropriate) are needed — not antiplatelet therapy alone as the immediate emergency bundle.

**Why it matters**

Antiplatelets are important in ACS but are not a complete immediate management answer for an **unstable** presentation with hypotension and tachycardia. The key under-tests resuscitation and pathway escalation and may imply it is safe to give antiplatelets without addressing shock — clinically unsafe and unfair if a fuller emergency option exists among distractors.

---

### Repeated key concept across skills

| Field | Detail |
|-------|--------|
| **Short code** | `batch_repetition_same_treatment_family` |
| **Validator code** | `batch_repetition.management_emergency_same_treatment_family` (also related: `batch_repetition.management_emergency_same_answer`, `batch_repetition.same_correct_answer_concept`, `batch_repetition.acs_batch_duplicate_aspirin`) |
| **Severity** | Medium |
| **Affected stage** | Multi-skill batch generation |
| **Affected skills** | Management + Emergency Management (same batch) |
| **Affected conditions so far** | Acute coronary syndrome |
| **Potentially affected conditions** | Sepsis, PE, asthma, DKA — any condition where Management and Emergency Management could share the same drug class or bundle |
| **Related codes** | `unstable_acs_antiplatelet_only_key` |
| **Current status** | **Validator working. Keep.** |

**Example**

In the same ACS batch, **Management** and **Emergency Management** were both keyed on **antiplatelet therapy** (e.g. aspirin ± second antiplatelet), so candidates see the same treatment family twice instead of distinct skill-appropriate decisions (e.g. ongoing pathway step vs immediate resuscitation/escalation).

**Why it matters**

Batches should spread learning across skills — diagnosis, investigation, management step, emergency action, monitoring — not repeat the same keyed concept. Repetition reduces educational value and can mask missing emergency-bundle or management-depth content.

---

### Repair overclassifies title changes as educational content

| Field | Detail |
|-------|--------|
| **Short code** | `repair_title_change_overclassified` |
| **Severity** | Low |
| **Affected stage** | AI repair pipeline (change classification) |
| **Affected conditions so far** | ACS / chest pain items |
| **Current status** | Recorded. Low priority. |

**Example**

Title change from “Initial investigation for chest pain” to “Suspected acute coronary syndrome” was labelled educational content, although the question’s clinical content and key were unchanged.

**Why it matters**

The repair log may overstate risk and make tutors less confident in harmless style repairs.

---

### Monitoring priority ambiguity

| Field | Detail |
|-------|--------|
| **Short code** | `monitoring_priority_ambiguity` |
| **Severity** | Watchlist |
| **Affected stage** | Generation (monitoring question framing) |
| **Affected conditions so far** | Aortic dissection monitoring items |
| **Current status** | Recorded. Watchlist. |

**Example**

Aortic dissection monitoring item keyed pulse pressure differential, but neurological status and signs of rupture were also plausible monitoring priorities.

**Why it matters**

Monitoring questions can become unfair when several parameters are clinically important and the stem does not specify the monitoring aim.

---

### Over-specific diagnosis unsupported by stem

| Field | Detail |
|-------|--------|
| **Short code** | `overspecific_diagnosis_unsupported_by_stem` |
| **Severity** | Medium |
| **Affected stage** | Generation / key selection |
| **Affected skills** | Diagnosis, Investigation / interpretation |
| **Affected conditions so far** | Upper GI bleeding (peptic ulcer disease context) |
| **Potentially affected conditions** | Any diagnosis item where the stem supports a broader syndrome or category but the key names a specific subtype without discriminating evidence |
| **Related codes** | `repair_introduces_competing_answer`, `distractor_key_overlap`, `multiple_defensible_answers` (placeholder) |
| **Current status** | Recorded for later review. Monitor recurrence before adding hard validator rules. |

**Example**

Stem supports peptic ulcer bleed, but the key specifies **gastric ulcer** while **duodenal ulcer** remains equally plausible from the information provided.

**Why it matters**

When the stem does not contain discriminating findings, a subtype-specific key can create multiple defensible answers and reduce one-best-answer fairness.

---

### Asthma step mismatch

| Field | Detail |
|-------|--------|
| **Short code** | `asthma_step_mismatch_relief_only_vs_ics_laba` |
| **Severity** | Medium–high |
| **Affected stage** | Generation / key selection / justification alignment |
| **Affected skills** | Prescribing, Management |
| **Affected conditions so far** | Asthma |
| **Potentially affected conditions** | Other stepwise long-term condition frameworks where baseline treatment must be explicit before escalation |
| **Current status** | Recorded for later review. Monitor recurrence before adding a strict validator rule. |

**Pattern**

Stem describes **reliever-only** asthma treatment, but the key or justification assumes the patient is already on regular **low-dose ICS** and recommends **ICS/LABA** escalation.

**Fix**

Either add regular low-dose ICS to the stem, or key starting an ICS / ICS-containing controller.

**Why it matters**

Escalation choices in asthma depend on the patient’s current maintenance step. If baseline controller therapy is not stated, a step-up key can become clinically inconsistent and unfair.

---

### Asthma step mismatch (missing current controller)

| Field | Detail |
|-------|--------|
| **Short code** | `asthma_step_mismatch_missing_current_controller` |
| **Severity** | Medium–high |
| **Affected stage** | Generation / key selection / justification alignment |
| **Affected skills** | Prescribing, Management, Monitoring |
| **Affected conditions so far** | Asthma |
| **Related codes** | `asthma_step_mismatch_relief_only_vs_ics_laba` |
| **Current status** | Recorded for later review. Monitor recurrence before adding strict rule checks. |

**Pattern**

The stem says asthma is poorly controlled but does not specify current controller treatment. The key recommends LABA / ICS-LABA escalation and the justification assumes regular ICS use that is not stated.

**Fix**

State current asthma therapy, inhaler technique and adherence if the question tests step-up treatment. Otherwise key review of adherence/technique or initiation of ICS-containing treatment.

**Why it matters**

Step-up decisions require baseline treatment and modifiable factors (especially adherence and inhaler technique). Without that context, escalation keys may be premature and reduce one-best-answer fairness.

---

### Acute asthma prescribing–management mismatch

| Field | Detail |
|-------|--------|
| **Short code** | `asthma_acute_prescribing_management_mismatch` |
| **Severity** | Medium–high |
| **Affected stage** | Generation / key selection / skill alignment |
| **Affected skills** | Prescribing (misapplied to acute severe presentation); Emergency Management (if stem stays severe) |
| **Affected conditions so far** | Asthma |
| **Potentially affected conditions** | COPD exacerbation, anaphylaxis — any acute respiratory item where Prescribing is selected but the stem implies immediate resuscitation |
| **Related codes** | `acute_management_missing_essential_component`, `acute_management_bundle_incompleteness_watchlist`, `condition_stage_ambiguity` |
| **Current status** | Recorded for later review. Monitor recurrence before adding strict validator or prompt rules. |

**Pattern**

Stem describes **acute asthma** with **hypoxia** and/or **tachypnoea** (or other markers of severity), but **Current skill** is **Prescribing** and the keyed answer is **salbutamol alone** (or another single reliever step). This under-specifies **emergency management** and may omit **oxygen**, **nebulised bronchodilator delivery**, and **systemic corticosteroids** that tutors expect for the severity painted in the stem.

**Example**

Acute asthma vignette: low SpO₂, tachypnoea, distress. Skill = Prescribing. Key = salbutamol MDI alone; distractors include oxygen or a bundled acute treatment without making one best answer clear.

**Fix**

Either **make the scenario mild** so it is genuinely prescribing-focused (stable or mild exacerbation, no hypoxia, clear maintenance or step-up decision), **or** change **Current skill** to **Emergency Management** and key a **complete immediate management action** (e.g. controlled oxygen plus nebulised bronchodilator and systemic corticosteroids, aligned with UK acute asthma guidance and the stem severity).

**Why it matters**

Prescribing items should test a defensible prescribing choice for the **stage** in the stem. A severe acute presentation with a single reliever key is often clinically incomplete and creates unfair competition with options that include standard acute bundle components.

---

## Placeholder headings (future entries)

- Unsafe clinical sequencing
- Stability/treatment mismatch
- Ruled-out condition still treated
- Contraindication ignored
- Option specificity overlap
- Multiple defensible answers
- Skill drift
- Severity overclaim