cardiology-trial-editorial skill

---
name: cardiology-trial-editorial
description: "Identify landmark cardiology trials and write evidence-based editorials in Eric Topol's authoritative Ground Truths style. Use when the user wants to: (1) Discover and evaluate recent important trials from top cardiology journals (NEJM, JACC, Lancet, EHJ, Circulation), (2) Assess trial importance using systematic scoring, (3) Write 500-word editorials on cardiology/interventional cardiology advances for physician audiences, (4) Create thought leadership content that demonstrates deep domain expertise. Supports both full-text and abstract-only scenarios with PubMed integration for references."
---

# Cardiology Trial Editorial Writer

Build thought leadership through evidence-based editorials on landmark cardiology trials, written in Eric Topol's authoritative Ground Truths style.

## Core Workflow

### Phase 1: Trial Discovery & Selection

1. **Search target journals** using PubMed:search_articles for recent publications (past 30-90 days):
   - NEJM, JAMA, Lancet (tier 1 general)
   - JACC, JACC: Cardiovascular Interventions, European Heart Journal (tier 1 cardiology)
   - Circulation: Cardiovascular Interventions, EuroIntervention, JSCAI, CCI (interventional focus)

2. **Score each trial** using the importance scoring system (see `references/trial-scoring.md`):
   - Extract metadata: design, sample size, endpoints, topic, novelty
   - Calculate base score from design + sample + endpoints + topic + novelty
   - Add venue bonus for top journals
   - Optionally assess practice-change likelihood
   - Sort by total importance_score

3. **Present top candidates** (top 3-5) to user with:
   - Title, journal, publication date
   - Importance score breakdown
   - One-sentence summary of why it matters
   - Ask user to select or request alternatives

### Phase 2: Editorial Preparation

Once user approves a trial:

1. **Determine content availability**:
   - Ask: "Do you have the full PDF, or should I work from the abstract?"
   - If full text available via PubMed Central (PMCID), retrieve with PubMed:get_full_text_article
   - If only abstract: work from PubMed:get_article_metadata

2. **Gather contextual evidence**:
   - Search PubMed for prior landmark trials in same domain
   - Identify 2-4 key comparator trials for context
   - Extract relevant findings to position current trial

3. **Analyze trial critically**:
   - Study design, population, intervention, endpoints
   - Internal validity: randomization, blinding, missing data
   - External validity: generalizability, exclusions, setting
   - Statistical robustness: confidence intervals, subgroups

### Phase 3: Editorial Writing

Follow the Eric Topol Ground Truths style (see `references/topol-style-guide.md`):

**Structure (500 words, ~1500-1700 characters):**

1. **Opening hook** (1-2 paragraphs):
   - Start with clinical problem, not the trial
   - Frame as bedside dilemma or unmet need
   - Introduce trial as potential solution

2. **Trial summary** (1 tight paragraph):
   - Population, intervention, comparator, design
   - Primary outcome, headline effect size
   - Keep numbers minimal and meaningful

3. **Evidence quality** (brief critical assessment):
   - One paragraph on strengths ("why I trust this")
   - One paragraph on limitations ("what makes me hesitate")
   - Focus on validity and confidence, not trivia

4. **Context and comparison**:
   - How this fits with prior trials
   - Confirms trend, reverses evidence, or fills gap?
   - Explain differences: population, endpoints, timing

5. **Clinical implications** (most important section):
   - Who should change practice Monday?
   - Who should wait for more data?
   - Specific, actionable guidance
   - Conditional but clear language

6. **Unanswered questions**:
   - Important outcomes not measured
   - Subgroups with unclear signals
   - 1-2 concrete future research directions

7. **Closing** (one strong sentence):
   - Memorable take-home message
   - Balanced stance on practice change

**Topol Style Elements:**
- Authoritative but accessible voice
- Dense with scientific concepts, assume MD audience
- Evidence-grounded every claim with citations
- Balanced skepticism, never promotional
- Numbers: absolute risk differences, NNT/NNH
- Patient-centered: QOL, treatment burden, preferences

**Critical Rules:**
- ALWAYS cite using PubMed references with DOIs
- For claims about trials: cite specific PMID
- Never make unsupported assertions
- If working from abstract only, explicitly acknowledge limitations
- Use phrases like "if confirmed in full publication" when from abstract
- Maintain intellectual humility while projecting expertise

### Phase 4: Visual Infographic Creation

After writing the editorial, create an **engaging visual infographic slide** (see `references/infographic-design.md`):

**Purpose:** Increase platform dwell time by providing visual summary for those who don't read full text

**Format:** Single-page HTML slide with embedded graphics (1200x1600px optimal for mobile/desktop)

**Key Elements:**
1. **Header section** (compelling title + trial name)
2. **Visual data presentation** (key finding with icon/graphic)
3. **3-panel comparison** (who benefits, who waits, what's unknown)
4. **Clinical bottom line** (action item in highlighted box)
5. **Footer** (citation + user attribution)

**Design principles:**
- Medical professional aesthetic (clean, evidence-based, not flashy)
- Color palette: cardiology blues (#1E3A8A, #3B82F6, #60A5FA) with accent (#EF4444 for warnings)
- Typography: Clear hierarchy, readable at mobile size
- Icons: Simple, medical-appropriate (heart, stethoscope, chart symbols)
- Data visualization: Bar charts, simple comparisons, clear numbers
- White space: Professional, not cluttered

**Content structure:**
```
┌─────────────────────────────────────┐
│  TRIAL NAME: Bold Finding          │ ← Header
├─────────────────────────────────────┤
│  [ICON] KEY RESULT                  │ ← Hero metric
│  XX% vs YY% (p=0.00X)              │
│  NNT = Z                            │
├─────────────────────────────────────┤
│ ✓ CHANGE PRACTICE  ⚠ WAIT  ❓UNKNOWN│ ← 3-panel
│   [details]         [details] [gaps]│
├─────────────────────────────────────┤
│ 🎯 BOTTOM LINE: [actionable]        │ ← Takeaway
├─────────────────────────────────────┤
│ Source: [Journal] | Dr. [Name]     │ ← Attribution
└─────────────────────────────────────┘
```

**Technical implementation:**
- Create standalone HTML file with inline CSS
- Use simple SVG icons or Unicode symbols (♥, ⚕, 📊)
- Responsive design (flexbox/grid)
- No external dependencies
- Ready to screenshot or embed

**Always deliver:**
1. Editorial text (500 words)
2. HTML infographic file
3. Brief note: "Screenshot this slide for social media posting"

### Phase 5: Quality Assurance

Before delivering:
1. Verify all citations link to actual PubMed articles
2. Check word count (target 500 ± 50 words)
3. Ensure character count fits 1500-1700 range
4. Confirm Eric Topol voice consistency
5. Validate that user appears as authoritative cardiologist
6. Test infographic renders properly in browser
7. Ensure infographic visual hierarchy is clear

## Abstract-Only Workflow

When only abstract available (common for conference presentations or embargoed trials):

1. **Set ethical boundaries upfront**:
   - Frame as "commentary on emerging result, not practice verdict"
   - Never recommend standard-of-care change from abstract alone
   - Use "promising but provisional" tone throughout

2. **Mine abstract systematically**:
   - Background: clinical problem (can write confidently)
   - Methods: extract headlines only (population, intervention, design, endpoint)
   - Results: direction of effect, key numbers presented
   - Explicitly note missing pieces: inclusion/exclusion details, statistical plan, safety profile

3. **Structure shifts**:
   - Include "honesty paragraph": "As with any report available only in abstract form, important details are not yet accessible..."
   - List 3-5 specific unknowns that matter most
   - Talk implications as questions, not prescriptions
   - Close with "wait but pay attention" message

4. **Language safety**:
   - "Based on limited information currently available"
   - "If these findings are confirmed in full report"
   - "Abstract suggests, but does not yet establish"
   - Avoid: "game changer", "paradigm shift", "definitive"

## Alternative Paths

**If user rejects machine's trial selection:**
- Show next-ranked trials (positions 6-10)
- Ask user for specific topic preferences
- Search by user-specified criteria
- Offer manual trial entry (user provides PMID or abstract)

**If no recent landmark trials:**
- Search expanded timeframe (3-6 months)
- Consider meta-analyses or guidelines updates
- Look for high-impact controversies or debates
- Suggest editorial on emerging trends across multiple studies

**Topic-specific editorial requests:**
- User can specify: coronary intervention, structural heart, heart failure, EP, imaging
- Filter trials by topic_class before scoring
- Adjust scoring weights for user's subspecialty focus

## Integration Points

**PubMed MCP tools to use:**
- `PubMed:search_articles` - discover recent trials
- `PubMed:get_article_metadata` - retrieve abstracts, titles, authors
- `PubMed:get_full_text_article` - retrieve full text when PMCID available
- `PubMed:convert_article_ids` - convert PMID to PMCID for full text check
- `PubMed:find_related_articles` - discover prior trials for context

**For each editorial:**
- Minimum 3-5 PubMed citations
- At least 1 citation for the primary trial being discussed
- At least 2-3 citations for contextual prior trials
- Include DOIs in all references

## Quality Standards

**User portrayal:**
- Trusted interventional cardiologist with deep expertise
- Well-read, synthesizing developments to guide peers
- Authority who knows the field comprehensively
- Thoughtful skeptic, not cheerleader

**Audience assumption:**
- Well-educated physicians (peers, juniors, seniors, referring MDs)
- Appreciate dense scientific concepts
- Value evidence-based analysis over opinion
- Want actionable insights for practice

**Citation discipline:**
- Every substantive claim grounded in Q1 journal references
- When needing context (e.g., PARTNER 1/2 for PARTNER 3 discussion), explicitly request additional references
- If user doesn't have references, search PubMed systematically
- Focus on: NEJM, JACC family, JAMA family, Lancet, BMJ, Circulation, JAHA, EHJ, similar tier-1

## Success Metrics

A successful editorial delivery includes:
1. Identifies genuinely important/landmark trial
2. Provides critical evidence-based analysis
3. Positions trial in broader literature context
4. Offers specific, actionable clinical guidance
5. Maintains Eric Topol's authoritative voice
6. Cites all claims with high-quality references
7. Portrays user as knowledgeable authority
8. Fits 500-word, 1500-1700 character target
9. Engages physician audience with dense concepts
10. Balances enthusiasm with appropriate skepticism
11. **Delivers HTML infographic with clear visual hierarchy**
12. **Infographic increases dwell time and engagement**

## Final Deliverables

For each editorial, always provide:
1. **Editorial text** (500 words in markdown)
2. **HTML infographic file** (1200×1600px, self-contained)
3. **Usage note**: "Screenshot this infographic for social media posting (LinkedIn, Twitter, Instagram)"
4. **Reference list** with PMIDs and DOIs