Cardiology Trial Editorial

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

Editorial Templates and Examples

Concrete templates for different scenarios: full text available vs abstract only.

Template 1: Full-Text Editorial (500 words)

Example: PARTNER 3 Trial (Full Text Available)

---

Title: TAVR Reaches Low-Risk Patients: Evolution or Revolution?

Every heart team now confronts the question: for a 70-year-old with isolated severe aortic stenosis and preserved ventricular function, has surgical aortic valve replacement become the second choice? For years, we navigated transcatheter aortic valve replacement's expansion from compassionate use to standard care in high-risk and intermediate-risk patients, always cognizant that durability questions limited applicability to younger, healthier patients. The PARTNER 3 trial, reported in this issue, tests whether TAVR's benefits extend to those at low surgical risk—a population where surgery has been the default for decades.

Mack and colleagues randomly assigned 1,000 patients with severe aortic stenosis and low surgical risk (mean STS score 1.9%) to receive TAVR with a balloon-expandable valve via transfemoral approach or surgical replacement. The primary composite endpoint—death, stroke, or rehospitalization at one year—occurred in 8.5% of TAVR patients versus 15.1% with surgery (absolute risk reduction 6.6 percentage points, hazard ratio 0.54, 95% CI 0.37-0.79, P=0.001 for superiority). This translates to one event prevented for every 15 patients treated with TAVR, a clinically meaningful difference by any standard. Secondary outcomes favored TAVR across the board: lower stroke rates at 30 days (0.6% vs 2.4%), shorter index hospitalization (3 vs 7 days), and better quality of life at early time points.

Why trust these results more than previous signals? The trial's strengths are substantial: adequate power, pre-specified endpoints, independent core laboratory adjudication, and low loss to follow-up (2.8%). The population represents real-world practice—not the highly selected cohorts that often populate pivotal device trials. Importantly, the margin of benefit remained robust across sensitivity analyses.

Yet several factors warrant caution. First, one-year durability differs from 10-year durability. For a 65-year-old patient with 20-year life expectancy, valve longevity beyond this trial's timeframe remains incompletely characterized. The PARTNER 2 intermediate-risk cohort showed durable benefits to five years, but extrapolating further requires faith, not data. Second, the higher pacemaker implantation rate with TAVR (17.4% vs 6.1%)—while not affecting the primary endpoint—carries long-term implications for right ventricular function that a one-year study cannot capture. Third, paravalvular regurgitation, though mild in most cases, occurred more frequently with TAVR (0.6% moderate or severe vs 0.1% with surgery). These are not trivial trade-offs.

How does this fit with prior evidence? The trajectory from PARTNER 1 (high-risk, 2011), through PARTNER 2 (intermediate-risk, 2016), to PARTNER 3 represents TAVR's systematic march down the risk spectrum. The NOTION trial hinted at this possibility in low-risk patients, but PARTNER 3's size and methodological rigor make the case more definitively. Concurrent data from Evolut Low Risk, using a self-expanding valve, showed similar superiority, suggesting the finding is valve-agnostic.

For patients matching the trial profile—adults under 80 with isolated severe AS, low surgical risk, and anatomy favorable for transfemoral TAVR—the intervention should now be offered alongside surgery as a guideline-supported option. The decision requires transparent discussion of the one-year advantage versus unknown long-term durability. For patients younger than 65, where valve longevity matters most, many heart teams will reasonably continue to favor surgery until longer-term data mature. The paradigm has shifted, but it has not been replaced.

---

Word count: 567 | Character count (with spaces): ~3,200 | Citations needed: 5-6

Required citations: 1. Mack MJ, et al. PARTNER 3. NEJM 2019. PMID: 30883058 2. Leon MB, et al. PARTNER 1. NEJM 2010. PMID: 20961243 3. Leon MB, et al. PARTNER 2. NEJM 2016. PMID: 27040324 4. Thyregod HG, et al. NOTION. Lancet 2015. PMID: 25788066 5. Popma JJ, et al. Evolut Low Risk. NEJM 2019. PMID: 30883053

---

Template 2: Abstract-Only Editorial (500 words)

Example: Hypothetical New TAVR Trial (Abstract Only)

---

Title: Extended TAVR Indications: Promise Meets Caution

The boundaries of transcatheter aortic valve replacement continue to expand, with each trial extending eligibility to younger, lower-risk patients. For clinicians navigating shared decision-making about valve intervention, knowing where the evidence stops and extrapolation begins matters profoundly. The abstract presented by Chen and colleagues at this year's ACC conference reports one-year outcomes from ADVANCE-TAVR, a randomized trial comparing TAVR to medical management in asymptomatic patients with severe aortic stenosis—a population for whom surgical timing remains intensely debated.

Based on the limited information currently available, the trial enrolled approximately 800 patients with asymptomatic severe AS and low surgical risk, randomizing them to either TAVR or guideline-directed medical management with watchful waiting. The primary endpoint—composite of cardiovascular death, heart failure hospitalization, or progression to symptomatic AS requiring intervention—occurred less frequently in the TAVR group (reported relative reduction of 42%, p<0.001). At face value, this suggests benefit from early intervention in a population traditionally managed conservatively.

However, several critical details remain inaccessible in abstract form. We cannot judge the exact inclusion and exclusion criteria that define "asymptomatic" in this context—were patients with subtle, easily dismissed symptoms excluded? How was disease progression defined and adjudicated? What threshold of valve area or gradient was required? The abstract provides no information on crossover rates, treatment discontinuation, or serious adverse events beyond the primary composite. Patterns of missing data, pre-specified versus post-hoc analyses, and the robustness of sensitivity analyses are similarly unavailable. These omissions are not the authors' fault—abstracts cannot contain the detail required for confident practice change—but they prevent definitive judgment about generalizability and internal validity.

If confirmed in full publication, these findings would extend the provocative signals from observational studies suggesting earlier intervention in asymptomatic severe AS improves outcomes. The mechanistic rationale is compelling: prolonged left ventricular pressure overload causes irreversible myocardial fibrosis, and relieving obstruction before symptoms emerge might preserve cardiac function. Yet previous attempts to intervene early in asymptomatic valve disease—from mitral stenosis to aortic regurgitation—have taught us that physiology and intuition don't always translate to clinical benefit. The distinction between asymptomatic imaging abnormality and truly asymptomatic clinical disease remains slippery.

Several practical questions will arise if these results hold. Can health systems deliver systematic echocardiographic surveillance to identify candidates before symptoms develop? How will costs balance against benefits when "watchful waiting" successfully avoids intervention in many patients for years? Most importantly, will patients accept procedural risk (stroke, pacemaker, paravalvular leak) to prevent a future event that might never occur without intervention?

The abstract provides a compelling glimpse of what may represent an important shift in how we approach asymptomatic severe AS. Until the complete dataset, peer-reviewed publication, and longer-term follow-up are available, caution is warranted in drawing firm conclusions for practice. Nevertheless, clinicians and guideline writers should watch closely, as confirmation of these results could fundamentally reshape timing strategies for aortic valve intervention. The possibility that earlier truly is better—when delivered thoughtfully—deserves the rigorous scrutiny that only full data disclosure can provide.

---

Word count: 556 | Character count (with spaces): ~3,400 | Citations needed: 3-4

Required citations: 1. Observational study on early AS intervention (hypothetical) 2. Natural history study of asymptomatic AS 3. Prior surgical timing trial in valve disease 4. Guideline recommendation on asymptomatic AS

Key phrases for abstract-only safety:

• "Based on the limited information currently available"
• "We cannot judge..."
• "The abstract provides no information on..."
• "If confirmed in full publication"
• "Until the complete dataset...is available"
• "Caution is warranted"

---

Template 3: Negative Trial Editorial (Full Text)

When Null Results Matter

Structure: 1. Opening: Frame the clinical equipoise that motivated the trial 2. Why we expected benefit: Mechanistic rationale, observational data 3. What the trial showed: Null result with confidence intervals 4. Why it's important: Negative trials prevent futile treatments 5. Implications: What should stop, who benefited from not knowing 6. Context: How this fits with prior conflicting signals 7. Close: Value of well-done negative trials

Example opening:

The allure of pharmacologic preconditioning before percutaneous coronary 
intervention has persisted for two decades, sustained by compelling 
mechanistic studies and tantalizing observational signals. Yet rigorous 
randomized trials have repeatedly failed to demonstrate benefit. The 
PROTECT-PCI trial, reported in this issue, administered high-dose statins 
immediately pre-PCI in patients not previously on statin therapy, testing 
whether acute pleiotropic effects could reduce periprocedural myocardial 
injury. The answer, definitively, is no. And that answer matters.

---

Template 4: Meta-Analysis Editorial

When Individual Trials Need Synthesis

Structure: 1. Opening: Landscape of conflicting or underpowered individual trials 2. Meta-analysis methods: Patient-level vs study-level, search strategy quality 3. Pooled findings: Main results with heterogeneity assessment 4. Subgroup analyses: Where effects differ by population/intervention 5. Limitations: Publication bias, quality of included trials, heterogeneity 6. Implications: Stronger conclusion than any single trial 7. Future: What RCT should be done based on this synthesis

Example opening:

When no single trial achieves definitive power, meta-analysis offers a path 
to clarity—provided the individual studies are sufficiently homogeneous to 
permit pooling. For the question of optimal P2Y12 inhibitor duration after 
drug-eluting stent implantation, we've accumulated a dozen modestly sized 
trials testing everything from one month to 24 months, each showing trends 
but none individually conclusive. The meta-analysis by Wang and colleagues 
synthesizes data from 15,000 patients across these trials, finally achieving 
the statistical power to parse bleeding risk from ischemic benefit.

---

Template 5: Guideline-Changing Trial

When Practice Must Shift

Structure: 1. Opening: Current guideline recommendation and its evidence basis 2. What was uncertain: Gap that motivated this trial 3. Trial results: Clear benefit with tight confidence intervals 4. Evidence quality: Why this trial is more definitive than predecessors 5. Guideline implications: Specific anticipated changes (Class I vs IIa) 6. Implementation: Practical aspects of practice change 7. Close: Timeline for guideline update and what doesn't change

Example opening:

Current ACC/AHA guidelines assign a Class IIb recommendation to transcatheter 
edge-to-edge mitral valve repair for functional mitral regurgitation in heart 
failure—essentially saying "might be reasonable in selected patients." This 
tepid language reflects prior trials' mixed signals and methodological 
limitations. The RESHAPE-HF2 trial, a rigorously conducted 1,500-patient 
randomized study demonstrating 35% reduction in cardiovascular death and heart 
failure hospitalization at two years, will almost certainly elevate this to 
Class I. Guideline writers are surely already drafting.

---

Word Budget Allocation (500-word target)

Opening hook: 80-100 words

• Clinical problem: 40-50 words
• Trial introduction: 40-50 words

Trial summary: 60-80 words

• Population/design: 30-40 words
• Results: 30-40 words

Evidence quality: 100-120 words

• Strengths: 50-60 words
• Limitations: 50-60 words

Context: 80-100 words

• Prior trials: 40-50 words
• How this fits: 40-50 words

Clinical implications: 120-140 words

• Who changes practice: 50-60 words
• Nuances/caveats: 40-50 words
• Patient perspective: 30-40 words

Future questions: 60-80 words

• Unanswered questions: 40-50 words
• Research agenda: 20-30 words

Closing: 20-30 words

• One-sentence take-home

Total: 520-650 words → edit down to 450-550 word final

---

Citation Integration Examples

Inline citation style:

"The ISCHEMIA trial demonstrated no benefit of routine invasive strategy 
over medical management for stable coronary disease (Maron et al., NEJM 
2020; PMID: 32227755), fundamentally challenging decades of practice."

Multi-trial context:

"The evolution from SYNTAX (Serruys et al., NEJM 2009; PMID: 19252140) 
through EXCEL (Stone et al., NEJM 2016; PMID: 27792200) to NOBLE (Mäkikallio 
et al., Lancet 2016; PMID: 27765474) has traced the expanding—and now 
contracting—enthusiasm for PCI in left main disease."

Mechanistic grounding:

"Colchicine's anti-inflammatory effects in coronary disease (Nidorf et al., 
NEJM 2020; PMID: 32865380) represent a proof-of-concept for the inflammatory 
hypothesis of atherosclerosis first articulated in landmark observational work 
(Libby et al., Nature 2002; PMID: 12490958)."

---

Quality Assurance Checklist

Before finalizing any editorial:

Structure:

• [ ] Opens with clinical problem, not trial
• [ ] Trial summary is tight (60-80 words)
• [ ] Evidence assessment includes strengths AND limitations
• [ ] Prior trials are contextualized
• [ ] Clinical implications are specific and actionable
• [ ] Closes with memorable take-home

Content:

• [ ] Every substantive claim has PubMed citation
• [ ] Absolute risk differences provided, not just relative
• [ ] Confidence intervals presented
• [ ] NNT/NNH calculated where appropriate
• [ ] Patient perspective incorporated
• [ ] Conflicts of interest acknowledged if relevant

Style:

• [ ] Topol voice: authoritative, evidence-dense, skeptical
• [ ] No hype words (game-changer, revolutionary, etc.)
• [ ] Appropriate certainty modulation
• [ ] User portrayed as knowledgeable cardiologist
• [ ] Physician-level audience (dense concepts welcomed)

Metrics:

• [ ] Word count: 450-550 words
• [ ] Character count: 1500-1700 with spaces
• [ ] Minimum 3-5 PubMed citations with PMIDs
• [ ] All citations from Q1 journals

Abstract-only specific (if applicable):

• [ ] Explicitly acknowledges working from abstract only
• [ ] Lists 3-5 specific unknowns
• [ ] Uses conditional language throughout
• [ ] Frames as "promising but provisional"
• [ ] Closes with "wait but watch" message
• [ ] Never recommends practice change

Infographic Design Guide

Create engaging visual summaries that increase platform dwell time and make complex trials accessible at a glance.

Design Philosophy

Goal: Professional medical infographic that works on LinkedIn, Twitter/X, Instagram, and newsletter platforms

Key principles:

• Evidence-based aesthetic (credible, not sensational)
• Mobile-first design (readable on phone screens)
• Screenshot-ready (no cropping needed)
• Accessible to physicians scrolling quickly
• Complements editorial (doesn't replace it)

Dimensions and Format

Optimal size: 1200px width × 1600px height (3:4 aspect ratio)

• Works well on Instagram, LinkedIn, Twitter
• Mobile-friendly vertical orientation
• Professional portrait layout

File format: HTML with inline CSS/SVG

• Single self-contained file
• No external dependencies
• Easy to render and screenshot
• Can be opened in any browser

Color Palette

Primary Colors (Cardiology Theme)

--navy-dark: #1E3A8A      /* Headers, emphasis */
--blue-primary: #3B82F6   /* Key metrics, icons */
--blue-light: #60A5FA     /* Backgrounds, accents */
--blue-pale: #DBEAFE      /* Section backgrounds */

Accent Colors

--red-warning: #EF4444    /* Cautions, limitations */
--green-positive: #10B981 /* Benefits, improvements */
--amber-neutral: #F59E0B  /* Wait/uncertain */
--gray-text: #374151      /* Body text */
--gray-light: #F3F4F6     /* Subtle backgrounds */

Usage Guidelines

• Headers: Navy dark (#1E3A8A)
• Key metrics: Blue primary (#3B82F6), large and bold
• Benefits/Change practice: Green positive (#10B981)
• Limitations/Wait: Amber neutral (#F59E0B)
• Unknowns/Gaps: Red warning (#EF4444)
• Body text: Gray text (#374151)
• Backgrounds: White primary, blue pale for sections

Typography

Font Stack

font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', 'Roboto', 
             'Helvetica Neue', Arial, sans-serif;

Size Hierarchy

--h1: 36px / 2.25rem   /* Trial name */
--h2: 28px / 1.75rem   /* Section headers */
--h3: 20px / 1.25rem   /* Subsection labels */
--body: 16px / 1rem    /* Details */
--metric: 48px / 3rem  /* Key numbers */
--small: 14px / 0.875rem /* Citations */

Weight Hierarchy

• Trial name: 700 (bold)
• Key metric: 700 (bold)
• Section headers: 600 (semibold)
• Body text: 400 (regular)
• Citations: 400 (regular)

Layout Structure

Six-Section Vertical Layout

┌─────────────────────────────────────────┐
│ 1. HEADER SECTION                       │ ← 200px
│    • Trial acronym/name                 │
│    • One-line hook                      │
├─────────────────────────────────────────┤
│ 2. HERO METRIC                          │ ← 300px
│    • Large visual of key finding       │
│    • Icon + number + context           │
│    • NNT if applicable                 │
├─────────────────────────────────────────┤
│ 3. TRIAL DETAILS                        │ ← 200px
│    • Design, N, endpoint               │
│    • Brief description                 │
├─────────────────────────────────────────┤
│ 4. THREE-PANEL COMPARISON               │ ← 500px
│    ┌──────┬──────┬──────┐              │
│    │ ✓    │ ⚠    │ ❓    │              │
│    │CHANGE│ WAIT │UNKNOWN│              │
│    │      │      │       │              │
│    └──────┴──────┴──────┘              │
├─────────────────────────────────────────┤
│ 5. BOTTOM LINE                          │ ← 200px
│    • Key takeaway box                  │
│    • Actionable insight                │
├─────────────────────────────────────────┤
│ 6. FOOTER                               │ ← 200px
│    • Citation                          │
│    • Dr. [Your Name]                   │
└─────────────────────────────────────────┘
Total: ~1600px

Section-by-Section Design

Section 1: Header (200px)

<div class="header">
  <h1>TRIAL ACRONYM</h1>
  <p class="hook">Compelling one-liner about what changed</p>
</div>

Design specs:

• Background: White or blue-pale gradient
• Trial name: Navy dark, 36px, bold, uppercase
• Hook: Gray text, 20px, regular
• Padding: 40px top/bottom, 30px sides

Example:

PARTNER 3
TAVR Moves to Low-Risk Patients with Superior Outcomes

Section 2: Hero Metric (300px)

<div class="hero-metric">
  <div class="icon">❤️</div>
  <div class="comparison">
    <div class="result">8.5%</div>
    <div class="vs">vs</div>
    <div class="comparator">15.1%</div>
  </div>
  <div class="context">Death, Stroke, or Rehospitalization at 1 Year</div>
  <div class="nnt">NNT = 15</div>
</div>

Design specs:

• Background: Blue pale (#DBEAFE)
• Icon: 64px, centered, heart or relevant medical symbol
• Numbers: 48px, bold, blue primary for result, gray for comparator
• VS: 24px, gray, between numbers
• Context: 18px, gray text
• NNT: 20px, navy dark, highlighted box

Visual layout:

       ❤️
    
   8.5%  vs  15.1%
   
Death, Stroke, or Rehospitalization at 1 Year

    NNT = 15

Section 3: Trial Details (200px)

<div class="trial-details">
  <div class="detail-row">
    <span class="label">Design:</span>
    <span class="value">Multicenter RCT</span>
  </div>
  <div class="detail-row">
    <span class="label">Patients:</span>
    <span class="value">N=1,000 (low surgical risk, severe AS)</span>
  </div>
  <div class="detail-row">
    <span class="label">Intervention:</span>
    <span class="value">TAVR vs Surgical AVR</span>
  </div>
</div>

Design specs:

• Background: White
• Layout: Two-column (label left, value right)
• Label: 16px, semibold, gray text
• Value: 16px, regular, navy dark
• Border: Subtle gray divider between rows
• Padding: 20px all sides

Section 4: Three-Panel Comparison (500px)

<div class="three-panel">
  <div class="panel change-practice">
    <div class="panel-icon">✓</div>
    <h3>Change Practice</h3>
    <ul>
      <li>Low-risk severe AS</li>
      <li>Age < 80 years</li>
      <li>Transfemoral access</li>
    </ul>
  </div>
  
  <div class="panel wait">
    <div class="panel-icon">⚠</div>
    <h3>Wait for Data</h3>
    <ul>
      <li>Age < 65 years</li>
      <li>Bicuspid valves</li>
      <li>Need for longevity</li>
    </ul>
  </div>
  
  <div class="panel unknown">
    <div class="panel-icon">❓</div>
    <h3>Still Unknown</h3>
    <ul>
      <li>Durability > 5 years</li>
      <li>RV impact of PPM</li>
      <li>Cost-effectiveness</li>
    </ul>
  </div>
</div>

Design specs:

• Layout: 3 equal columns (33% each)
• Panel backgrounds:
• Change Practice: Light green tint (#ECFDF5)
• Wait: Light amber tint (#FEF3C7)
• Unknown: Light red tint (#FEE2E2)
• Icons: 32px, matching panel color
• Headers: 20px, semibold, matching panel color (darker shade)
• List items: 15px, regular, gray text
• Bullets: Panel-colored circles
• Padding: 20px all sides
• Border radius: 8px

Visual layout:

┌──────────┬──────────┬──────────┐
│    ✓     │    ⚠     │    ❓     │
│  CHANGE  │   WAIT   │  UNKNOWN │
│ PRACTICE │          │          │
│          │          │          │
│ • Item 1 │ • Item 1 │ • Item 1 │
│ • Item 2 │ • Item 2 │ • Item 2 │
│ • Item 3 │ • Item 3 │ • Item 3 │
└──────────┴──────────┴──────────┘

Section 5: Bottom Line (200px)

<div class="bottom-line">
  <div class="icon">🎯</div>
  <p class="takeaway">For low-risk severe AS patients, TAVR is now a 
  guideline-supported option alongside surgery, with superior short-term 
  outcomes balanced against uncertain long-term durability.</p>
</div>

Design specs:

• Background: Navy dark (#1E3A8A)
• Text color: White
• Icon: 32px, gold/yellow emoji
• Takeaway: 18px, regular, white text
• Padding: 30px all sides
• Border: 3px solid blue primary

Visual emphasis:

• This is the "screenshot gold" - what people remember
• Should be quotable and actionable
• Max 2 sentences

Section 6: Footer (200px)

<div class="footer">
  <div class="citation">
    <strong>Source:</strong> Mack MJ, et al. NEJM 2019;380(18):1695-1705. 
    PMID: 30883058
  </div>
  <div class="attribution">
    <div class="author">Dr. [Your Name]</div>
    <div class="credentials">Interventional Cardiologist</div>
  </div>
</div>

Design specs:

• Background: Gray light (#F3F4F6)
• Citation: 14px, regular, gray text
• Author: 18px, semibold, navy dark
• Credentials: 14px, regular, gray text
• Layout: Two-column (citation left, author right)
• Padding: 30px all sides

Icon Library

Medical Symbols (Unicode)

• Heart: ❤️ (U+2764)
• Medical symbol: ⚕️ (U+2695)
• Check mark: ✓ (U+2713)
• Warning: ⚠ (U+26A0)
• Question: ❓ (U+2753)
• Target: 🎯 (U+1F3AF)
• Chart: 📊 (U+1F4CA)
• Syringe: 💉 (U+1F489)
• Pills: 💊 (U+1F48A)
• Microscope: 🔬 (U+1F52C)

Simple SVG Icons (When Unicode Insufficient)

Heart with EKG Line:

<svg width="64" height="64" viewBox="0 0 64 64">
  <path d="M32 54L10 32C4 26 4 16 10 10C16 4 26 4 32 10C38 4 48 4 54 10C60 16 60 26 54 32L32 54Z" 
        fill="#3B82F6"/>
  <path d="M10 32L20 32L24 28L28 36L32 32L54 32" 
        stroke="#FFFFFF" stroke-width="2" fill="none"/>
</svg>

Stethoscope:

<svg width="64" height="64" viewBox="0 0 64 64">
  <circle cx="48" cy="48" r="8" fill="#3B82F6"/>
  <path d="M20 10C20 6 24 6 24 10L24 30C24 40 34 50 48 50" 
        stroke="#3B82F6" stroke-width="3" fill="none"/>
  <circle cx="20" cy="10" r="4" fill="#1E3A8A"/>
  <circle cx="24" cy="10" r="4" fill="#1E3A8A"/>
</svg>

Data Visualization

Comparison Bars

For showing differences between groups:

<div class="comparison-bars">
  <div class="bar-group">
    <div class="bar-label">TAVR</div>
    <div class="bar" style="width: 8.5%;">
      <span class="bar-value">8.5%</span>
    </div>
  </div>
  <div class="bar-group">
    <div class="bar-label">Surgery</div>
    <div class="bar" style="width: 15.1%;">
      <span class="bar-value">15.1%</span>
    </div>
  </div>
</div>

Design specs:

• Bar height: 40px
• TAVR bar: Green (#10B981)
• Surgery bar: Gray (#6B7280)
• Values: 16px, bold, white text inside bar
• Labels: 16px, regular, gray text on left
• Normalize to make visual difference clear

Simple Risk Reduction

Visual representation of NNT:

<div class="nnt-visual">
  <div class="people-grid">
    <!-- 15 person icons, 1 highlighted -->
    <div class="person highlighted">👤</div>
    <div class="person">👤</div>
    <!-- ... repeat 13 more times -->
  </div>
  <p class="nnt-text">1 in 15 patients benefits from TAVR</p>
</div>

Complete HTML Template

<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>[Trial Name] - Infographic</title>
<style>
* {
  margin: 0;
  padding: 0;
  box-sizing: border-box;
}

body {
  font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', 'Roboto', 
               'Helvetica Neue', Arial, sans-serif;
  background: #ffffff;
  width: 1200px;
  height: 1600px;
  margin: 0 auto;
}

.infographic {
  width: 100%;
  height: 100%;
  background: white;
  display: flex;
  flex-direction: column;
}

/* Header Section */
.header {
  background: linear-gradient(135deg, #DBEAFE 0%, #ffffff 100%);
  padding: 40px 30px;
  text-align: center;
  border-bottom: 4px solid #3B82F6;
}

.header h1 {
  font-size: 36px;
  font-weight: 700;
  color: #1E3A8A;
  margin-bottom: 10px;
  text-transform: uppercase;
  letter-spacing: 1px;
}

.header .hook {
  font-size: 20px;
  color: #374151;
  line-height: 1.4;
}

/* Hero Metric Section */
.hero-metric {
  background: #DBEAFE;
  padding: 40px 30px;
  text-align: center;
}

.hero-metric .icon {
  font-size: 64px;
  margin-bottom: 20px;
}

.hero-metric .comparison {
  display: flex;
  justify-content: center;
  align-items: center;
  gap: 20px;
  margin-bottom: 15px;
}

.hero-metric .result {
  font-size: 48px;
  font-weight: 700;
  color: #3B82F6;
}

.hero-metric .vs {
  font-size: 24px;
  color: #6B7280;
}

.hero-metric .comparator {
  font-size: 48px;
  font-weight: 700;
  color: #9CA3AF;
}

.hero-metric .context {
  font-size: 18px;
  color: #374151;
  margin-bottom: 15px;
}

.hero-metric .nnt {
  display: inline-block;
  background: #1E3A8A;
  color: white;
  font-size: 20px;
  font-weight: 600;
  padding: 10px 30px;
  border-radius: 25px;
}

/* Trial Details Section */
.trial-details {
  background: white;
  padding: 30px;
}

.detail-row {
  display: flex;
  padding: 12px 0;
  border-bottom: 1px solid #E5E7EB;
}

.detail-row:last-child {
  border-bottom: none;
}

.detail-row .label {
  font-size: 16px;
  font-weight: 600;
  color: #6B7280;
  width: 150px;
  flex-shrink: 0;
}

.detail-row .value {
  font-size: 16px;
  color: #1E3A8A;
  flex-grow: 1;
}

/* Three-Panel Section */
.three-panel {
  display: flex;
  gap: 20px;
  padding: 30px;
  background: #F9FAFB;
}

.panel {
  flex: 1;
  padding: 25px;
  border-radius: 12px;
  box-shadow: 0 2px 8px rgba(0, 0, 0, 0.1);
}

.panel.change-practice {
  background: #ECFDF5;
  border: 2px solid #10B981;
}

.panel.wait {
  background: #FEF3C7;
  border: 2px solid #F59E0B;
}

.panel.unknown {
  background: #FEE2E2;
  border: 2px solid #EF4444;
}

.panel-icon {
  font-size: 32px;
  text-align: center;
  margin-bottom: 15px;
}

.panel h3 {
  font-size: 20px;
  font-weight: 600;
  text-align: center;
  margin-bottom: 20px;
}

.panel.change-practice h3 {
  color: #059669;
}

.panel.wait h3 {
  color: #D97706;
}

.panel.unknown h3 {
  color: #DC2626;
}

.panel ul {
  list-style: none;
}

.panel li {
  font-size: 15px;
  color: #374151;
  line-height: 1.6;
  padding: 8px 0;
  padding-left: 25px;
  position: relative;
}

.panel li::before {
  content: "●";
  position: absolute;
  left: 0;
  font-size: 18px;
}

.panel.change-practice li::before {
  color: #10B981;
}

.panel.wait li::before {
  color: #F59E0B;
}

.panel.unknown li::before {
  color: #EF4444;
}

/* Bottom Line Section */
.bottom-line {
  background: #1E3A8A;
  padding: 35px 40px;
  border: 3px solid #3B82F6;
  margin: 20px 30px;
  border-radius: 8px;
  display: flex;
  align-items: center;
  gap: 20px;
}

.bottom-line .icon {
  font-size: 32px;
  flex-shrink: 0;
}

.bottom-line .takeaway {
  font-size: 18px;
  color: white;
  line-height: 1.6;
  font-weight: 400;
}

/* Footer Section */
.footer {
  background: #F3F4F6;
  padding: 30px;
  margin-top: auto;
  display: flex;
  justify-content: space-between;
  align-items: center;
}

.footer .citation {
  font-size: 14px;
  color: #6B7280;
  max-width: 60%;
}

.footer .citation strong {
  color: #374151;
}

.footer .attribution {
  text-align: right;
}

.footer .author {
  font-size: 18px;
  font-weight: 600;
  color: #1E3A8A;
  margin-bottom: 4px;
}

.footer .credentials {
  font-size: 14px;
  color: #6B7280;
}
</style>
</head>
<body>
<div class="infographic">
  
  <!-- Header Section -->
  <div class="header">
    <h1>[TRIAL ACRONYM]</h1>
    <p class="hook">[Compelling one-liner about key finding]</p>
  </div>
  
  <!-- Hero Metric Section -->
  <div class="hero-metric">
    <div class="icon">❤️</div>
    <div class="comparison">
      <div class="result">[X.X%]</div>
      <div class="vs">vs</div>
      <div class="comparator">[Y.Y%]</div>
    </div>
    <div class="context">[Primary endpoint description]</div>
    <div class="nnt">NNT = [Z]</div>
  </div>
  
  <!-- Trial Details Section -->
  <div class="trial-details">
    <div class="detail-row">
      <span class="label">Design:</span>
      <span class="value">[Study design]</span>
    </div>
    <div class="detail-row">
      <span class="label">Patients:</span>
      <span class="value">[N and population]</span>
    </div>
    <div class="detail-row">
      <span class="label">Intervention:</span>
      <span class="value">[Treatment vs control]</span>
    </div>
  </div>
  
  <!-- Three-Panel Comparison -->
  <div class="three-panel">
    <div class="panel change-practice">
      <div class="panel-icon">✓</div>
      <h3>Change Practice</h3>
      <ul>
        <li>[Population characteristic]</li>
        <li>[Population characteristic]</li>
        <li>[Population characteristic]</li>
      </ul>
    </div>
    
    <div class="panel wait">
      <div class="panel-icon">⚠</div>
      <h3>Wait for Data</h3>
      <ul>
        <li>[Uncertain population]</li>
        <li>[Uncertain population]</li>
        <li>[Uncertain population]</li>
      </ul>
    </div>
    
    <div class="panel unknown">
      <div class="panel-icon">❓</div>
      <h3>Still Unknown</h3>
      <ul>
        <li>[Unanswered question]</li>
        <li>[Unanswered question]</li>
        <li>[Unanswered question]</li>
      </ul>
    </div>
  </div>
  
  <!-- Bottom Line -->
  <div class="bottom-line">
    <div class="icon">🎯</div>
    <p class="takeaway">[2-sentence actionable clinical takeaway]</p>
  </div>
  
  <!-- Footer -->
  <div class="footer">
    <div class="citation">
      <strong>Source:</strong> [First Author] et al. [Journal] [Year];[Vol]([Issue]):[Pages]. PMID: [PMID]
    </div>
    <div class="attribution">
      <div class="author">Dr. [Your Name]</div>
      <div class="credentials">Interventional Cardiologist</div>
    </div>
  </div>
  
</div>
</body>
</html>

Content Guidelines

What to Include

Header:

• Trial acronym or short name (e.g., "PARTNER 3", "ISCHEMIA", "COAPT")
• Hook must be outcome-focused, not process-focused
• Good: "TAVR Superior to Surgery in Low-Risk Patients"
• Bad: "New Study Compares TAVR and Surgery"

Hero Metric:

• Primary endpoint result as comparison
• Always show absolute numbers, not just relative
• Include NNT when meaningful (typically NNT 5-50 range)
• Use icons that relate to outcome (heart for cardiac, brain for stroke)

Trial Details:

• Design: Be specific (e.g., "Double-blind RCT" not just "RCT")
• Patients: Include N and key population feature
• Intervention: Show both arms clearly

Three Panels:

• Change Practice: Specific population characteristics where benefit is clear
• Wait: Populations where more data needed or benefit uncertain
• Unknown: Key unanswered questions (durability, long-term safety, cost)
• Each panel: 2-4 bullet points max

Bottom Line:

• Max 2 sentences
• Actionable and quotable
• Balance benefit and uncertainty
• Patient-centered when possible

Footer:

• Full citation with PMID for traceability
• Your name and credentials for authority

What to Avoid

Don't:

• Use more than 3 colors in data visualizations
• Include p-values unless truly necessary (focus on effect size)
• Crowd the design - white space is professional
• Use clipart or stock photos
• Include more than 1 chart/graph per infographic
• Make font sizes smaller than 14px
• Use all caps except for trial name
• Include your photo (keeps focus on science)

Responsive Considerations

While designed at 1200×1600px, ensure readability when scaled:

At 600px width (50% scale):

• Minimum 7px font size
• Icons still recognizable
• Three panels stack vertically on very narrow screens (not typical for this size)

Testing checklist:

• Open HTML in browser at 100%
• Take screenshot (should be 1200×1600)
• View at 50% zoom - still readable?
• Print to PDF - maintains quality?

Export and Usage

For LinkedIn:

• Screenshot the full HTML page
• Upload as image post
• Add editorial text as caption
• Tag relevant hashtags (#Cardiology #TAVR #MedEd)

For Twitter/X:

• Same screenshot
• Thread: Infographic as first tweet, editorial key points as subsequent tweets
• Alt text for accessibility

For Instagram:

• Screenshot works well in feed
• Consider adding branded border if posting regularly
• Use stories for swipe-up to full editorial

For Newsletter:

• Embed HTML directly OR
• Include screenshot with link to read full editorial
• Infographic increases open rate and engagement

Accessibility

Alt text template:

"[Trial Name] infographic showing [key finding]: [intervention] resulted in 
[X%] vs [Y%] for [comparator] in [population]. Change practice for 
[specific population], wait for data on [uncertain populations], 
still unknown: [key questions]."

Screen reader friendly:

• Use semantic HTML (header, section, footer)
• Ensure color contrast meets WCAG AA standards (4.5:1 for normal text)
• All data conveyed in text, not just visually

Quality Checklist

Before delivering infographic:

• [ ] Trial name/acronym is accurate
• [ ] Numbers match editorial (primary endpoint, NNT)
• [ ] Citation is complete with PMID
• [ ] User's name and credentials included
• [ ] Three panels have 2-4 items each
• [ ] Bottom line is quotable and actionable
• [ ] All fonts are readable at mobile size
• [ ] Color contrast is sufficient
• [ ] No spelling or grammar errors
• [ ] HTML renders correctly in Chrome/Safari/Firefox
• [ ] File is self-contained (no external dependencies)
• [ ] Screenshot dimensions are 1200×1600

Advanced: Animated Version (Optional)

For higher engagement on social media, consider subtle animations:

@keyframes fadeIn {
  from { opacity: 0; transform: translateY(10px); }
  to { opacity: 1; transform: translateY(0); }
}

.header { animation: fadeIn 0.6s ease-out; }
.hero-metric { animation: fadeIn 0.8s ease-out 0.2s backwards; }
.three-panel { animation: fadeIn 1s ease-out 0.4s backwards; }

Only use if:

• Posting on platforms that support HTML/CSS (rare)
• Creating video version (screen record the HTML)
• User specifically requests animated version

Default: Static screenshot is preferred for maximum compatibility

Target Journals and Search Strategies

Prioritized list of cardiology journals for trial discovery, with PubMed search strategies.

Journal Hierarchy by Impact Factor

Tier 1: General Medicine (Highest Impact)

1. New England Journal of Medicine (NEJM) - IF: ~158

• PubMed search: "N Engl J Med"[Journal]
• Focus: Landmark trials, often multi-specialty
• Cardiology frequency: 2-3 major trials/month

2. The Lancet - IF: ~168

• PubMed search: "Lancet"[Journal]
• Focus: International trials, public health angle
• Cardiology frequency: 1-2 major trials/month

3. JAMA - IF: ~120

• PubMed search: "JAMA"[Journal]
• Focus: Practice-changing trials, US-centric
• Cardiology frequency: 1-2 major trials/month

Tier 1: Cardiology Specialist (Very High Impact)

4. Journal of the American College of Cardiology (JACC) - IF: ~24

• PubMed search: "J Am Coll Cardiol"[Journal]
• Focus: Broad cardiology, high-quality trials
• Editorial sweet spot: 3-4 landmark trials/month

5. European Heart Journal (EHJ) - IF: ~39

• PubMed search: "Eur Heart J"[Journal]
• Focus: European trials, imaging, prevention
• Editorial sweet spot: 2-3 landmark trials/month

6. Circulation - IF: ~37

• PubMed search: "Circulation"[Journal]
• Focus: AHA journal, mechanistic + clinical
• Editorial sweet spot: 2-3 major trials/month

Tier 2: Subspecialty High Impact

7. JACC: Cardiovascular Interventions - IF: ~12

• PubMed search: "JACC Cardiovasc Interv"[Journal]
• Focus: Interventional cardiology trials
• Prime territory for interventional cardiologist
• Editorial sweet spot: 2-4 trials/month

8. Circulation: Cardiovascular Interventions - IF: ~7

• PubMed search: "Circ Cardiovasc Interv"[Journal]
• Focus: PCI, structural, peripheral interventions
• Editorial sweet spot: 1-2 trials/month

9. EuroIntervention - IF: ~4

• PubMed search: "EuroIntervention"[Journal]
• Focus: European interventional trials, devices
• Editorial sweet spot: 1-2 trials/month

Tier 3: Subspecialty Focused

10. JSCAI (Journal of the Society for Cardiovascular Angiography and Interventions) - IF: ~3

• PubMed search: "J Soc Cardiovasc Angiogr Interv"[Journal] OR "JSCAI"[Journal]
• Focus: SCAI society journal, emerging data
• Editorial opportunity: 1 trial/month

11. Catheterization and Cardiovascular Interventions (CCI) - IF: ~3

• PubMed search: "Catheter Cardiovasc Interv"[Journal]
• Focus: Technical interventional studies
• Editorial opportunity: 1 trial/month

Also Monitor (JAMA Family)

12. JAMA Cardiology - IF: ~18

• PubMed search: "JAMA Cardiol"[Journal]
• Focus: High-quality cardiology trials
• Editorial sweet spot: 2-3 trials/month

Search Strategies by Priority

Strategy 1: High-Impact Sweep (Daily/Weekly)

Prioritize journals most likely to publish practice-changing trials.

PubMed search query:
("N Engl J Med"[Journal] OR "Lancet"[Journal] OR "JAMA"[Journal] OR 
"J Am Coll Cardiol"[Journal] OR "Eur Heart J"[Journal] OR 
"Circulation"[Journal] OR "JACC Cardiovasc Interv"[Journal] OR 
"JAMA Cardiol"[Journal]) AND 
("randomized controlled trial"[Publication Type] OR "meta-analysis"[Publication Type]) AND 
("cardiovascular"[All Fields] OR "cardiac"[All Fields] OR "coronary"[All Fields] OR 
"heart"[All Fields] OR "aortic"[All Fields] OR "mitral"[All Fields])

Date filter: Past 30 days

Expected yield: 15-25 trials/month

Strategy 2: Interventional Focus (Weekly)

For interventional cardiology specialist audience.

PubMed search query:
("JACC Cardiovasc Interv"[Journal] OR "Circ Cardiovasc Interv"[Journal] OR 
"EuroIntervention"[Journal] OR "Catheter Cardiovasc Interv"[Journal] OR 
"J Soc Cardiovasc Angiogr Interv"[Journal]) AND 
("randomized controlled trial"[Publication Type] OR "prospective"[All Fields]) AND 
("percutaneous coronary intervention"[All Fields] OR "PCI"[All Fields] OR 
"TAVR"[All Fields] OR "transcatheter"[All Fields] OR "structural heart"[All Fields] OR 
"mitral clip"[All Fields] OR "left atrial appendage"[All Fields])

Date filter: Past 30 days

Expected yield: 10-15 trials/month

Strategy 3: Topic-Specific Deep Dive

When focusing on specific intervention or condition.

Example: TAVR/Structural Heart

("transcatheter aortic valve"[All Fields] OR "TAVR"[All Fields] OR 
"TAVI"[All Fields] OR "mitral valve repair"[All Fields] OR 
"MitraClip"[All Fields] OR "tricuspid valve"[All Fields]) AND 
("randomized controlled trial"[Publication Type] OR "prospective"[All Fields])

Example: Complex PCI

("chronic total occlusion"[All Fields] OR "CTO"[All Fields] OR 
"left main"[All Fields] OR "bifurcation"[All Fields] OR 
"multivessel PCI"[All Fields] OR "drug-eluting stent"[All Fields]) AND 
("randomized controlled trial"[Publication Type] OR "registry"[All Fields])

Example: Heart Failure Interventions

("heart failure"[All Fields] AND ("device"[All Fields] OR "intervention"[All Fields])) AND 
("cardiac resynchronization"[All Fields] OR "CRT"[All Fields] OR 
"left ventricular assist"[All Fields] OR "LVAD"[All Fields] OR 
"mitral regurgitation"[All Fields] AND "heart failure"[All Fields])

Strategy 4: Meta-Analyses and Guidelines

When individual trials are sparse, synthesize evidence.

("meta-analysis"[Publication Type] OR "systematic review"[Publication Type] OR 
"guideline"[Publication Type]) AND 
("cardiovascular"[All Fields] OR "cardiology"[All Fields]) AND 
("JACC"[Journal] OR "Eur Heart J"[Journal] OR "Circulation"[Journal] OR 
"N Engl J Med"[Journal] OR "Lancet"[Journal])

Date filter: Past 90 days

Expected yield: 5-10 high-quality reviews/month

Trial Discovery Workflow

Step-by-Step Process

1. Weekly high-impact sweep (Strategy 1)

• Run search for past 7 days
• Retrieve PMIDs for all results
• Extract abstracts using PubMed:get_article_metadata

2. Score all trials using trial-scoring system

• Run LLM metadata extraction on each abstract
• Calculate importance scores
• Rank order by score

3. Weekly interventional focus (Strategy 2)

• Run search for past 7 days
• Add to candidate pool
• Re-score and re-rank combined pool

4. Identify top 5 candidates

• Present to user with scores and summaries
• User selects or requests alternatives

5. If no strong candidates, expand:

• Increase date range to 30 days
• Run Strategy 4 (meta-analyses)
• Consider topic-specific deep dive (Strategy 3)

Seasonality and Conference Timing

Peak trial publication months:

• March: ACC conference → high JACC, JACC-CI output
• May: ESC conference → high EHJ, EuroIntervention output
• August: ESC congress → late-breaking trials
• November: AHA conference → high Circulation output

Adjust search strategy:

• During conference months: search daily, not weekly
• Post-conference (1-2 months later): expect embargoed trials to publish

Special Scenarios

Negative Trials

Don't overlook high-quality null results:

Additional filter: "no significant difference"[All Fields] OR "non-inferiority"[All Fields]

Negative trials can be just as important as positive ones for editorial purposes.

Device/Procedural Studies

May not use "randomized controlled trial" MeSH term:

Alternative filter: "prospective"[All Fields] AND "consecutive"[All Fields]

Registry Studies

Large-scale real-world data:

"registry"[All Fields] OR "database"[All Fields] AND "outcomes"[All Fields]
AND sample size filter: look for N > 10,000

Controversies and Retractions

Monitor for debates worth editorializing:

"retracted"[All Fields] OR "concern"[All Fields] OR "controversy"[All Fields]

Rare, but high editorial value when legitimate scientific debate emerges.

Automation Notes

For systematic trial discovery, ideal workflow:

1. Daily automated PubMed search (Strategy 1) 2. Extract metadata for all new results 3. Score trials using importance algorithm 4. Alert user when high-scoring trial (>12 points) appears 5. Present top 3-5 weekly for editorial selection

This ensures user never misses landmark trials while avoiding alert fatigue.

Quality Filters

Beyond journal and study design, consider:

Include if:

• Sample size > 300 for RCTs
• Sample size > 1000 for registries
• Hard clinical endpoints (death, MI, stroke)
• Pre-registered on clinicaltrials.gov
• Industry-sponsored but independently monitored

Exclude if:

• Case series (N < 10)
• Single-center studies (unless exceptionally novel)
• Pure imaging/diagnostic studies (unless breakthrough)
• Editorials/commentaries (we're writing those, not reading them)
• Retracted or major concerns raised

PubMed MCP Integration

Key tools to use:

1. PubMed:search_articles(query, date_from, date_to, sort="pub_date", max_results=50)

• Run searches with strategies above
• Retrieve recent trials

2. PubMed:get_article_metadata(pmids=["12345678", ...])

• Get abstracts, titles, authors, journals
• Feed into scoring algorithm

3. PubMed:convert_article_ids(ids=["12345678"], id_type="pmid")

• Check if PMCID exists (full text available)
• If PMCID present, can retrieve full text

4. PubMed:get_full_text_article(pmc_ids=["PMC1234567"])

• Retrieve complete article when available
• Enables deeper analysis for editorial

5. PubMed:find_related_articles(pmids=["12345678"], link_type="pubmed_pubmed")

• Discover prior trials in same domain
• Build context for editorial

Deliverable Format

When presenting trial candidates to user:

**Top 5 Editorial Candidates (Past 30 Days)**

1. **[Trial Name/First Author]** - NEJM, Oct 2024 | Score: 16/18
   - Design: Large RCT (N=5000) | Endpoints: Hard clinical
   - Finding: SGLT2i reduced HF hospitalization by 35% in CKD patients
   - Why it matters: Expands SGLT2i to non-diabetic CKD population
   - Full text: Available (PMCID: PMC12345678)

2. **[Trial Name/First Author]** - JACC-CI, Oct 2024 | Score: 14/18
   - Design: Large RCT (N=2000) | Endpoints: Hard clinical
   - Finding: Drug-coated balloon non-inferior to DES for small vessels
   - Why it matters: Potential paradigm shift in small vessel PCI
   - Full text: Abstract only

[Continue for remaining 3...]

Which trial would you like me to write an editorial on? Or would you like to see alternatives?

Eric Topol Ground Truths Style Guide

Comprehensive guide to mimicking Eric Topol's authoritative, evidence-dense writing style for cardiology editorials.

Voice Characteristics

Authority Without Arrogance

• Speak as a peer to other physicians, not lecturing down
• Confidence comes from evidence, not bombast
• Willing to say "we don't know yet" when appropriate
• Show intellectual humility while demonstrating deep expertise

Dense but Accessible

• Pack substantive content into every sentence
• Assume physician audience understands medical concepts
• Don't oversimplify, but don't use jargon gratuitously
• Balance technical precision with readability

Evidence-Grounded Skepticism

• Default stance: "show me the data"
• Question methodology, not just accept conclusions
• Point out conflicts of interest when relevant
• Distinguish association from causation rigorously

Patient-Centered Lens

• Always circle back to "what does this mean for patients?"
• Consider quality of life, not just mortality/morbidity
• Acknowledge treatment burden and patient preferences
• Question whether statistical significance equals clinical meaningfulness

Structural Patterns

Opening Gambits (First 1-2 Paragraphs)

Pattern 1: The Clinical Dilemma

Example: "Every interventional cardiologist faces this scenario weekly: 
an 82-year-old with severe aortic stenosis, multiple comorbidities, 
and a surgical risk score that makes the heart team uncomfortable. 
For years, we've navigated this with educated guesses about who 
truly benefits from transcatheter intervention. The NOTION-3 trial 
offers data where we've had only intuition."

Pattern 2: The Practice Gap

Example: "Despite decades of statin trials, we continue to debate 
LDL targets in secondary prevention. The paradigm of 'lower is better' 
bumps against concerns about polypharmacy, costs, and diminishing returns. 
Into this unsettled landscape comes CLEAR Outcomes, testing a question 
that practicing cardiologists need answered."

Pattern 3: The Provocative Question

Example: "Can we prevent Alzheimer's disease in patients with 
coronary artery disease by intensifying cardiovascular risk factor 
control? The connection between vascular health and cognitive 
decline has moved from hypothesis to mechanism, but intervention 
trials have disappointed. Now we have preliminary evidence that 
changes the conversation."

The "Why I Trust This" / "Why I Hesitate" Technique

Always include both in evidence assessment:

Trust signals:

• Pre-specified primary outcome, no changes mid-trial
• Adequate sample size and power for stated hypothesis
• Independent adjudication of events
• Intention-to-treat analysis
• Low loss to follow-up (<5%)
• Registry in clinicaltrials.gov before enrollment
• Transparent reporting of conflicts

Hesitation signals:

• Composite endpoints hiding null individual components
• Early stopping for benefit (inflates effect size)
• Post-hoc subgroup analyses
• Surrogate endpoints without long-term validation
• Industry sponsorship without independent data monitoring
• Highly selected population limiting generalizability
• Missing data >10% for primary outcome

Contextualizing with Prior Evidence

Template:

These findings [confirm/extend/contradict] the signals from [Trial A] 
and [Trial B], which demonstrated [brief prior result]. However, 
unlike those earlier studies, [current trial] enrolled [key difference], 
which may explain [why results differ or align].

Always answer: 1. What did we think before this trial? 2. How does this trial change our thinking? 3. What's the plausible mechanistic or methodological explanation?

Clinical Implications: The "Monday Morning" Test

Good practice guidance (Topol-style):

"For patients who resemble the trial population—adults under 75 
with isolated severe AS and low surgical risk—TAVR should now be 
discussed alongside surgery as a guideline-supported option. The 
8.5% vs 15.1% composite endpoint at one year represents a number 
needed to treat of 15, clinically meaningful by any standard."

What to avoid:

"This is a game-changer. Everyone should get TAVR now."

Nuanced stance:

"The durability question remains. While TAVR shows superior 
short-term outcomes, valve longevity beyond 5 years is less 
well established than for surgical bioprostheses. For a 65-year-old 
with 20+ year life expectancy, this uncertainty matters and merits 
shared decision-making."

Closing Statements

Pattern 1: The Balanced Summary

"This trial represents a major advance in how we approach [condition], 
but it is not the final word. [Specific unanswered question] will 
require longer follow-up and real-world data."

Pattern 2: The Practice Directive

"For the profile studied here—[specifics]—the evidence now supports 
[specific practice change], provided that [important caveat]. Outside 
this population, caution and individualization remain essential."

Pattern 3: The Research Agenda

"The natural next question is whether [extension or variation]. 
Until we have that data, clinicians will need to balance the clear 
gains in [outcome A] against uncertain effects on [outcome B]."

Numerical Presentation

Prefer Absolute Over Relative Risk

Weak:

"Treatment reduced events by 43%"

Strong (Topol-style):

"Treatment reduced events from 15.1% to 8.5%—an absolute risk 
reduction of 6.6 percentage points (HR 0.54, 95% CI 0.37-0.79). 
This translates to one event prevented for every 15 patients treated."

Number Needed to Treat/Harm

Calculate and present when possible:

"With an ARR of 6.6%, the NNT is 15 to prevent one composite event 
(death, stroke, or rehospitalization) at one year. Against this, 
the increased pacemaker rate of 17% vs 7% yields an NNH of 10."

Confidence Intervals Tell Stories

Good:

"The confidence interval (0.37 to 0.79) excludes unity and excludes 
trivial benefits, supporting both statistical and clinical significance."

Point out when CIs are problematic:

"While the point estimate suggests benefit, the wide confidence 
interval (0.58 to 1.12) includes both meaningful benefit and 
meaningful harm. This uncertainty should temper enthusiasm."

Citation Discipline

Always Cite Substantive Claims

Claim types requiring citations:

• Trial results or findings
• Historical context ("Previous trials showed...")
• Mechanistic statements ("The pathway involves...")
• Epidemiological data ("Affecting 5 million Americans...")
• Guideline recommendations

Citation Format

In-text:

"The PARTNER 3 trial demonstrated superior outcomes with TAVR 
compared to surgery in low-risk patients (Mack et al., NEJM 2019; 
PMID: 30883058)."

Reference list (end of editorial):

1. Mack MJ, Leon MB, Thourani VH, et al. Transcatheter Aortic-Valve 
Replacement with a Balloon-Expandable Valve in Low-Risk Patients. 
N Engl J Med. 2019;380(18):1695-1705. doi:10.1056/NEJMoa1814052 
PMID: 30883058

Multi-Trial Context Example

"The trajectory from PARTNER 1 (high-risk, 2011; PMID: 21696309) 
through PARTNER 2 (intermediate-risk, 2016; PMID: 27040324) to 
PARTNER 3 (low-risk, 2019; PMID: 30883058) traces TAVR's expansion 
across the surgical risk spectrum, each trial shifting the paradigm 
one step further."

Language Patterns

Certainty Modulation

High certainty (strong evidence):

• "The data demonstrate..."
• "This trial establishes..."
• "We can now conclude..."

Moderate certainty (good but imperfect evidence):

• "The findings suggest..."
• "This provides evidence that..."
• "It appears likely that..."

Low certainty (preliminary, abstract-only, or limited data):

• "Initial results hint at..."
• "If confirmed, this could..."
• "The abstract suggests, but does not prove..."

Avoiding Hype

Never use:

• "Game-changer"
• "Paradigm shift"
• "Revolutionary"
• "Breakthrough" (unless truly first-in-class)
• "Definitive proof"

Use instead:

• "Important advance"
• "Meaningful progress"
• "Practice-changing for [specific population]"
• "Strong evidence"
• "Valuable contribution"

Patient-Centered Language

Integrate throughout:

• "What patients actually experience..."
• "From the patient's perspective..."
• "The treatment burden includes..."
• "Quality of life considerations..."
• "Shared decision-making should weigh..."

Topol Signature Moves

The Comparative Framework

Position every trial against what came before:

"Unlike SURTAVI, which studied intermediate-risk patients, 
Evolut Low Risk enrolled patients with STS scores <3%, 
making direct comparison to PARTNER 3 possible for the 
first time across different valve platforms."

The Methodological Dive

Show you've read the methods carefully:

"The trial's adaptive design allowed early stopping for efficacy, 
reached at the pre-planned interim analysis. While this accelerates 
knowledge, it also tends to overestimate effect sizes—something 
to remember when projecting benefits forward."

The Cost/Access Reality Check

Don't ignore practical constraints:

"These results matter only if they translate to practice. At 
$30,000 per TAVR procedure, expansion to low-risk patients 
carries health system implications that extend beyond individual 
patient benefit. Cost-effectiveness analyses will be essential."

The Subgroup Scrutiny

When subgroups tell different stories:

"While the overall benefit is clear, the interaction p-value of 0.03 
for age suggests younger patients derive less benefit—or potentially 
more harm. This isn't a post-hoc fishing expedition; it's a 
pre-specified analysis that demands attention."

Word Economy

Topol packs information density into limited space:

Weak (65 words):

"The trial was a randomized controlled trial that included a total 
of 1000 patients. These patients had severe aortic stenosis and were 
considered to be at low risk for surgery. The patients were randomly 
assigned to receive either TAVR or surgical valve replacement. The 
main outcome that was measured was a composite endpoint consisting 
of death, stroke, or rehospitalization at one year."

Strong (Topol-style, 35 words):

"In this 1000-patient RCT, individuals with severe aortic stenosis 
and low surgical risk were randomized to TAVR versus surgery, with 
a primary composite endpoint of death, stroke, or rehospitalization 
at one year."

Target Metrics

For 500-word editorial:

• Opening hook: 80-100 words
• Trial summary: 60-80 words
• Evidence quality: 100-120 words
• Context: 80-100 words
• Clinical implications: 120-140 words
• Limitations/future: 60-80 words
• Closing: 20-30 words

Character count: 1500-1700 (including spaces, punctuation)

Quality Checklist

Before finalizing, verify:

• [ ] Every substantive claim has a PubMed citation
• [ ] Numbers include both absolute and relative risks
• [ ] Confidence intervals are presented for effect estimates
• [ ] Prior trials are contextualized with specific citations
• [ ] Clinical implications are specific and actionable
• [ ] Limitations are acknowledged without undermining the message
• [ ] Patient perspective is incorporated
• [ ] No hype words ("game-changer", "revolutionary")
• [ ] Voice is authoritative but not arrogant
• [ ] User appears as knowledgeable cardiologist authority
• [ ] Physician audience would find content dense and valuable
• [ ] Word count: 450-550 words
• [ ] Character count: 1500-1700

Common Pitfalls to Avoid

1. Press release tone: Don't simply relay trial conclusions 2. Excessive hedging: Balance skepticism with clarity 3. Ignoring prior evidence: Trials don't exist in vacuum 4. Forgetting patients: More than just statistics 5. Uncritical acceptance: Every trial has limitations 6. Vague implications: Be specific about who should change what 7. Missing the forest: What's the big picture message? 8. Citation gaps: Ground every claim in evidence 9. Wrong audience level: Assume MD-level sophistication 10. Losing Topol voice: Authoritative, evidence-dense, patient-centered

Trial Importance Scoring System

Hybrid rules + LLM approach to identify landmark cardiology trials worthy of editorial coverage.

Two-Layer Approach

Layer 1: Rules-Based Scoring (Transparent, Token-Efficient)

For each trial abstract, use an LLM call to extract structured metadata:

Prompt Template:

You are a cardiology trial methodologist.
You receive the title and abstract of a new cardiology article.
Output a JSON object with these fields only:

"design": one of ["large_RCT", "small_RCT", "observational", "registry", "meta_analysis", "case_series", "basic_science", "review", "editorial", "other"]
"sample_size": an integer estimate if clearly stated in the abstract, else null
"endpoints": one of ["hard_clinical", "surrogate", "procedural", "diagnostic", "other"]
"topic_class": one of ["coronary_intervention", "structural_intervention", "EP", "heart_failure", "prevention", "imaging", "other"]
"novelty": one of ["incremental", "moderate", "high"] based on whether the study seems to test a new strategy or device vs standard of care
"journal": the journal name from the article metadata

Do not hallucinate numbers or facts not clearly present in the abstract.

Scoring Algorithm:

def calculate_importance_score(metadata):
    score = 0
    
    # Design weight (5 points max)
    if metadata["design"] == "large_RCT":
        score += 5
    elif metadata["design"] == "small_RCT":
        score += 3
    elif metadata["design"] == "meta_analysis":
        score += 3
    elif metadata["design"] in ["observational", "registry"]:
        score += 1
    
    # Sample size weight (3 points max)
    if metadata["sample_size"]:
        if metadata["sample_size"] >= 3000:
            score += 3
        elif metadata["sample_size"] >= 1000:
            score += 2
        elif metadata["sample_size"] >= 300:
            score += 1
    
    # Endpoints weight (3 points max)
    if metadata["endpoints"] == "hard_clinical":
        score += 3
    elif metadata["endpoints"] == "surrogate":
        score += 1
    
    # Topic relevance (2 points max)
    high_value_topics = ["coronary_intervention", "structural_intervention", "heart_failure"]
    if metadata["topic_class"] in high_value_topics:
        score += 2
    
    # Novelty weight (3 points max)
    if metadata["novelty"] == "high":
        score += 3
    elif metadata["novelty"] == "moderate":
        score += 1
    
    # Venue bonus (2 points max)
    journal_lower = metadata.get("journal", "").lower()
    top_journals = ["jacc", "circulation", "european heart journal", "nejm", "jama", "lancet"]
    if any(j in journal_lower for j in top_journals):
        score += 2
    
    return score

Maximum possible score: 18 points

Layer 2: Practice-Change Likelihood (Optional, for Top Candidates)

After initial scoring, for the top 5-10 trials, add a second LLM assessment:

Prompt Template:

You are a senior cardiologist reviewing a new clinical trial.
Based only on the title and abstract, estimate how likely this study is to meaningfully influence clinical guidelines or everyday practice, if the results are confirmed.

Answer strictly in JSON with:
"practice_change_likelihood": one of ["low", "moderate", "high"]
"reason": one sentence explanation

Additional Points:

• "high" → +3 points
• "moderate" → +1 point
• "low" → 0 points

Final score = Layer 1 score + Layer 2 points

Score Interpretation

15+ points: Likely landmark trial, strong editorial candidate 10-14 points: Important contribution, good editorial candidate 7-9 points: Moderate interest, editorial if slow news cycle < 7 points: Usually skip unless special circumstances

Special Considerations

User Subspecialty Adjustment

If user specializes in specific area, adjust topic weights:

• Interventional cardiologist: coronary_intervention +3, structural_intervention +3
• Heart failure specialist: heart_failure +3
• Electrophysiologist: EP +3
• Imaging specialist: imaging +2

Contextual Factors (Not in Score)

Consider but don't score:

• First-in-class device or therapy
• Reverses prior dogma
• Addresses FDA "black box" warning
• Major guideline timing (e.g., ACC/AHA update pending)
• Media buzz or social media traction
• Controversy or debate in the field

Multi-Trial Synthesis

Sometimes no single trial scores high, but 3-4 moderate trials on same topic collectively matter:

• Meta-analysis opportunity
• "State of the field" editorial
• Trend analysis across studies

Implementation Workflow

1. Search PubMed for recent articles (30-90 days) from target journals 2. Extract abstracts for all results 3. Run Layer 1 LLM call for each abstract → get structured JSON 4. Calculate base scores using algorithm above 5. Sort by score, take top 10 6. Run Layer 2 LLM call on top 10 → practice-change assessment 7. Add Layer 2 points, re-sort 8. Present top 3-5 to user with score breakdown

Quality Checks

• Verify sample sizes match abstract (no hallucination)
• Confirm design classification (RCT vs observational)
• Check endpoint classification against actual primary outcome
• Validate novelty assessment against clinical knowledge

Edge Cases

Negative trials: High-quality null results can be landmark too

• If large RCT with hard endpoints shows no difference, still score high
• Novelty should account for "practice-changing negative result"

Subset analyses: Post-hoc or subgroup papers usually score lower

• Unless they fundamentally change interpretation of parent trial

Device/procedural studies: May have surrogate endpoints but high clinical relevance

• Consider pragmatic adjustment for interventional cardiology context

Registry studies: Can be landmark if massive scale or novel insights

• E.g., 100,000+ patients with real-world outcomes