skills/generate-clinical-scenario/references/clinical-evidence.md

# FILE: references/clinical-evidence.md

# Clinical Evidence and Accuracy

Use this reference whenever a scenario depends materially on:

- clinical guidelines
- current recommendations
- medication information
- clinical thresholds
- resuscitation algorithms
- professional standards
- healthcare policy
- educational simulation standards
- external evidence or citations

# General Principles

Do not invent clinical facts, recommendations, citations, publication titles or URLs.

Distinguish between:

- established clinical knowledge
- current guideline recommendations
- local or organisational practice
- educational assumptions made for simulation design

If current guidance cannot be verified, state that limitation.

Do not imply that an educational scenario has been formally validated or approved by a professional body unless that is genuinely established.

# UK Evidence Sources

For UK scenarios, prefer authoritative primary sources where relevant, including:

- NICE
- NHS
- Resuscitation Council UK
- British National Formulary
- UK professional regulators
- Relevant Royal Colleges
- National specialty guidance
- UK Health Security Agency where relevant
- Other appropriate national clinical bodies

For simulation practice, recognised sources may include:

- ASPiH
- INACSL

Use JRCALC where appropriate only when the relevant recommendation can actually be verified from material available to the model or supplied by the user.

Do not claim exact JRCALC alignment solely from general pre-hospital knowledge.

# Current Guidance

When a scenario depends materially on information that may have changed, verify current guidance when suitable tools or authoritative sources are available.

Examples include:

- medication dosing
- resuscitation guidance
- treatment thresholds
- clinical pathways
- national policy
- infection-control guidance
- professional standards

Prefer primary authoritative sources over summaries, blogs or secondary commentary.

When current evidence is not material to the learning objective, avoid unnecessary research that does not improve the scenario.

# User-Supplied Material

When the user provides:

- learning outcomes
- curriculum documents
- local clinical guidance
- an existing scenario
- simulation templates
- assessment criteria
- organisational policies

use those materials as the source of truth for the requested educational structure or local requirements.

Do not silently replace user-supplied requirements with generic alternatives.

If supplied material conflicts with current authoritative guidance in a way that materially affects the scenario, identify the discrepancy rather than silently resolving it.

# Medications

Where medications are relevant, use:

- Generic medication name
- Indication
- Route
- Dose where educationally appropriate
- Relevant cautions or contraindications
- Expected response

Medication content is for simulation and education.

It must not be framed as personalised prescribing advice for a real patient.

Where useful and verifiable, link to the BNF or another appropriate authoritative source.

Do not fabricate BNF links, doses or contraindications.

# Clinical Consistency

Before finalising a scenario, check that:

- symptoms fit the intended condition
- observations are physiologically plausible
- severity is internally consistent
- examination findings fit the presentation
- investigations are plausible
- medication effects are plausible
- clinical progression follows logically from the condition and interventions
- improvement or deterioration occurs at a realistic pace for the educational scenario

Do not add abnormal findings merely to make a case more complex.

# Scoring Systems

When using systems such as NEWS2, GCS or other clinical scores:

- include only where relevant
- ensure the component values and total score are internally consistent
- avoid presenting a calculated score unless confident it is correct
- verify current scoring rules where they materially affect the scenario

# Simulation Standards

Where relevant, scenario design should reflect recognised simulation principles such as:

- clear learning objectives
- appropriate learner level
- psychological and physical safety
- purposeful fidelity
- structured facilitation
- appropriate pre-briefing
- meaningful debriefing
- accessibility and inclusion

Do not overstate compliance with ASPiH or INACSL standards where a full standards-based review has not been undertaken.

Prefer wording such as:

"designed with recognised simulation principles in mind"

rather than:

"fully compliant with ASPiH/INACSL"

unless compliance has actually been assessed.

# Citations and Resources

Provide educational resources only when they add value to the scenario.

Prefer a short, relevant set of authoritative resources rather than a long generic bibliography.

When citing a resource:

- ensure the organisation or publication exists
- ensure the title is accurate
- use a verified URL where possible
- do not fabricate publication dates
- do not invent guideline numbers
- make clear when a source could not be directly verified

# Visual and Moulage Evidence

When generating or describing wounds, moulage or clinical visual material:

- ensure findings are consistent with the intended pathology
- distinguish simulated appearances from diagnostic images of real patients
- avoid adding clinical signs that contradict the scenario
- keep visual detail proportionate to the learning objective
- use an appropriate UK healthcare environment where relevant

If image generation is available and the user requests an image, generate the visual rather than limiting the response to a textual description when appropriate.

# Final Evidence Check

Before completing evidence-dependent scenario content, check:

- Are clinical claims plausible?
- Are medication details appropriate?
- Are current recommendations verified where necessary?
- Are sources authoritative?
- Have any citations or URLs been invented?
- Is the content clearly educational rather than real-patient advice?
- Have local assumptions been identified where relevant?