Clinical Decision Support
Generate professional clinical decision support documents with GRADE evidence grading and statistical analysis for cardiology content.
Triggers
- User needs evidence-based treatment recommendations
- User is creating clinical guideline summaries
- User wants to analyze patient cohort data
- User needs to present evidence with GRADE grading
- User is developing clinical algorithms
Document Types
1. Treatment Recommendation Reports
Structure:
- Clinical question (PICO format)
- Evidence summary
- GRADE assessment
- Recommendation statement
- Implementation considerations
GRADE Evidence Levels: | Grade | Certainty | Meaning | |-------|-----------|---------| | 1A | High | Strong recommendation, high-quality evidence | | 1B | Moderate | Strong recommendation, moderate evidence | | 2A | High | Weak recommendation, high-quality evidence | | 2B | Moderate | Weak recommendation, moderate evidence | | 2C | Low | Weak recommendation, low-quality evidence |
2. Patient Cohort Analysis
Components:
- Demographics and baseline characteristics
- Biomarker stratification
- Outcome comparisons with statistics
- Subgroup analyses
- Clinical implications
3. Guideline Summaries
Elements:
- Recommendation class (I, IIa, IIb, III)
- Level of evidence (A, B, C)
- Key supporting trials
- Clinical context
- Special populations
Cardiology-Specific Applications
Heart Failure Management
- GDMT optimization pathways
- Device therapy eligibility
- Risk stratification (MAGGIC, Seattle HF Model)
- Stage-based recommendations
Coronary Artery Disease
- Revascularization decisions
- Medical therapy optimization
- Risk scores (SYNTAX, HEART, TIMI)
- Secondary prevention
Arrhythmia Management
- Anticoagulation decisions (CHA₂DS₂-VASc)
- Rate vs rhythm control
- Device therapy indications
- Ablation candidacy
Valvular Heart Disease
- Intervention timing
- Surgical vs transcatheter approach
- Risk assessment (STS, EuroSCORE)
- Surveillance recommendations
Statistical Presentation
Required Elements
- Hazard ratios with 95% CI
- Absolute risk differences
- Number needed to treat (NNT)
- P-values (exact, not just thresholds)
- Forest plots for multiple comparisons
Survival Analysis Display
- Kaplan-Meier curves
- Number at risk tables
- Median survival with CI
- Landmark analyses if appropriate
Evidence Synthesis Framework
For Single Trial
- Study design and population
- Intervention and comparator
- Primary endpoint results
- Key secondary endpoints
- Safety profile
- Limitations
- Clinical implications
For Multiple Trials
- Consistency of findings
- Magnitude of effect across studies
- Population differences
- Statistical heterogeneity
- Overall certainty assessment
- Synthesized recommendation
GRADE Assessment Process
Factors That Lower Certainty
- Risk of bias (unblinded, high dropout)
- Inconsistency (heterogeneous results)
- Indirectness (surrogate outcomes, different population)
- Imprecision (wide CIs, few events)
- Publication bias
Factors That Raise Certainty
- Large effect (RR >2 or <0.5)
- Dose-response gradient
- All plausible confounders would reduce effect
Output Formatting
Executive Summary (Always First)
- 3-5 key findings highlighted
- Primary recommendation
- Evidence grade
- Clinical bottom line
Recommendation Statement Format
We recommend [intervention] for [population] with [condition]
to [outcome] (GRADE 1B: strong recommendation, moderate certainty).
Supporting evidence: [Key trials with effect sizes]
Best Practices
- Specify patient population precisely
- Use standardized outcome definitions (RECIST, CTCAE, etc.)
- Report both relative and absolute effects
- Include number at risk for survival data
- Acknowledge funding sources of cited trials
- Note guideline concordance/discordance
- Address special populations (elderly, renal impairment, etc.)
NOT For
This skill is NOT for individual patient treatment decisions. For that, clinical judgment integrating patient preferences, comorbidities, and circumstances is required beyond evidence synthesis.