article-appraisal - SKILL.md Agent Skill

name: article-appraisal description: Clinical Article Appraisal - Structured critical appraisal of clinical-pharmacology papers. Produces Highlights section + 8-section narrative appraisal (Abstract, Introduction, Methods, Results, Discussion, Limitations, Funding/COI, Critical Appraisal). Includes validity/bias assessment. Output formatted for Google Docs. Always responds in English regardless of article language. argument-hint:

Clinical Article Appraisal

🎯 TASK OVERVIEW

Perform a structured critical appraisal of the provided clinical-pharmacology paper PDF. Produce:

A brief 'Highlights' section summarizing the key takeaways with internal references
Narrative appraisal (target ≈ 600–800 words) organised under Sections 1-8
Final output formatted suitable for export to Google Docs

⚠️ CRITICAL RULES

Language Rule

Respond ONLY in English, even if the user or the article is in another language.

Source Limitation

Base every statement strictly and solely on the content within the provided article PDF. Do NOT access external websites, databases, or sources to supplement or verify information.

Missing Information

If required information is not present in the article, explicitly write "Not reported" and continue the appraisal.

Numerical Results

Quote numerical results exactly as printed in the article (e.g., effect estimate ± 95% CI, p-value). Use LaTeX formatting for all mathematical/scientific notation:

Enclose in $...$ or $$...$$
Example: $HR = 0.75; 95\% CI, 0.60-0.90$
Example: $p < 0.001$

Objectivity

Keep personal opinion or external knowledge out of the Highlights section and Sections 1–7. Objective critique belongs only in Section 8 (Critical Appraisal).

Clarity & Brevity

Use concise sentences and standard abbreviations (e.g., RCT, CI, HR, OR).

📋 OUTPUT STRUCTURE

Highlights

Purpose: Provide a very concise (3-5 key bullet points) overview of the most critical findings.

Content (based strictly on the article):

The core objective or primary research question (concisely stated)
The key population studied and main intervention tested
The most important quantitative result for primary outcome (effect estimate ± 95% CI, p-value)
(Optional) One or two other critical findings (e.g., major safety signal, key secondary outcome)

Format: Bullet points with internal references:

• [Finding] (Results, Table 2)
• [Finding] (Abstract)
• [Finding] (Discussion, p. 5)

📑 SECTION HEADINGS (Narrative Appraisal)

1. Abstract

Primary objective / research question as stated
Key result for primary endpoint (effect ± CI or p-value)
One-sentence author conclusion

2. Introduction

Clinical background and knowledge gap described
Explicit research question or hypothesis stated

3. Methods

Study design (e.g., RCT, cohort, PK crossover)
Setting & enrolment dates (country/region, single- vs multi-centre, dates)
Population: Key inclusion/exclusion criteria, final N analysed
Intervention or exposure: Drug(s), dose(s), route(s), duration(s)
Comparator / control group (if any)
Outcome definitions: Primary & key secondary outcomes, time-points
Study registration: Reported? (Yes/No/Not Reported; ID if stated)
Ethics: Committee approval reported? Participant consent reported?
Statistical approach:
- Effect measures (RR, OR, HR, IRR, ARR, NNT/NNH, means, medians)
- Statistical tests for primary endpoint & covariate adjustments
- Power calculation inputs ($\alpha$, $\beta$, expected effect, planned N)
- Missing data handling method (LOCF, MI, complete case, none reported)

4. Results

Participant flow: Screened → Eligible → Randomised → Allocated → Analysed
Primary outcome result(s): Effect estimate ± 95% CI, p-value
Key secondary outcomes: As for primary
Safety results: SAEs, AEs leading to withdrawal (counts & percentages)
Subgroup / sensitivity analyses: Pre-specified or post-hoc?
Unexpected findings highlighted by authors
Data presentation quality: Clarity of tables/figures

5. Discussion & Conclusions

Authors' main interpretation
Clinical/scientific implications claimed
Authors' discussion of consistency with previous literature
Future research directions proposed

6. Limitations

Limitations explicitly acknowledged by authors
Additional limitations identified by appraiser (based only on article content)

7. Funding & Conflicts of Interest

Funding source(s) disclosed
Author COI disclosures
Assessment of whether funding/COI could have influenced study

8. Critical Appraisal (Validity & Bias)

Internal validity: Randomisation, allocation concealment, blinding, outcome measurement accuracy, missing data handling
Risk of bias: Selection, performance, detection, attrition, reporting bias
External validity (Generalisability): Population, intervention, setting applicability
Overall evidence strength: Rate as High / Moderate / Low with one-sentence justification

📊 STATISTICAL & TABLE GUIDANCE

Effect Measures

Extract and report: RR, OR, HR, IRR, ARR, NNT/NNH, PAR, SMR, mean differences, etc.

Effect Estimates

Always give point estimate and 95% CI if provided:

$HR = 0.75; 95\% CI, 0.60-0.90$

p-values

Report exact value if given:

$p = 0.03$
$p < 0.001$
Avoid "NS" without actual p-value or CI

Tables/Figures

Cite clearly: "Results are shown in Table 2"
May recreate small summary table if needed

Graph-only Data

If results only in graphs without numbers:

"Numeric values for [outcome] were only presented graphically, limiting precise interpretation. Estimated effect $\approx X$ from Figure 1."

Multiple Testing

Note if authors mention adjustments (Bonferroni, FDR). If many outcomes tested without adjustment, note potential for inflated Type I error.

Power

State planned sample size, assumptions ($\alpha$, $\beta$, expected effect), achieved N. Comment if study appears underpowered.

🔬 OPTIONAL ADD-ON MODULES

Include only when clearly relevant based on paper's content:

Pharmacological Relevance (PK/PD)

If major focus: Discuss PK parameters ($AUC$, $C_{max}$, $t_{1/2}$), dose rationale, clinical implications

Cost-Effectiveness

If reported: Summarise ICER, economic results, budget impact

Population Applicability/Subgroups

If discussed in detail: Expand on specific subgroup effects (paediatrics, CKD, etc.)

Ethical Considerations

If discussed beyond basic approval/consent: Elaborate on ethical issues

Placement: Integrate into most appropriate section (Methods, Results, Discussion)

📄 OUTPUT FORMAT

Begin with Highlights section (with internal references)
Follow with structured Sections 1-8
Use clear headings (bold or Markdown)
Use paragraph breaks for readability
Use LaTeX formatting for all math/scientific notation
Format suitable for direct paste into Google Docs

📦 OUTPUT ARTIFACTS

At the end of each appraisal, offer to create exportable files:

Artifact	Format	Purpose
`appraisal-[AuthorYear].md`	Markdown	Complete structured appraisal
`appraisal-[AuthorYear].txt`	Plain text	Google Docs compatible
`appraisal-[AuthorYear].html`	HTML	Formatted with LaTeX rendering
`appraisal-summary.csv`	CSV	Key data for comparison across articles

Template: appraisal-[AuthorYear].md

# Clinical Article Appraisal

**Article:** [Full citation]
**Appraiser:** [User name if provided]
**Date:** [Today's date]

---

## Highlights

• [Key finding 1] (Section reference)
• [Key finding 2] (Section reference)
• [Key finding 3] (Section reference)
• [Key finding 4] (Section reference)

---

## 1. Abstract

**Primary Objective:** [Objective as stated]

**Key Result:** $[Effect estimate ± CI, p-value]$

**Author Conclusion:** [One sentence]

---

## 2. Introduction

**Clinical Background:** [Knowledge gap described]

**Research Question:** [Explicit hypothesis/question]

---

## 3. Methods

| Element | Details |
|---------|---------|
| **Design** | [e.g., Double-blind RCT] |
| **Setting** | [Country, centers, dates] |
| **Population** | [N, key criteria] |
| **Intervention** | [Drug, dose, duration] |
| **Comparator** | [Control description] |
| **Primary Outcome** | [Definition, timepoint] |
| **Registration** | [ID or Not Reported] |
| **Ethics** | [Approval + consent status] |

**Statistical Approach:**
- Effect measures: [RR, OR, HR, etc.]
- Primary analysis: [Test used]
- Power: $\alpha = [X], \beta = [Y]$, planned N = [Z]
- Missing data: [Method]

---

## 4. Results

**Participant Flow:** Screened (N) → Eligible (N) → Randomised (N) → Analysed (N)

**Primary Outcome:** $[Effect = X; 95\% CI, Y-Z; p = W]$

**Key Secondary Outcomes:**
- [Outcome 1]: $[Result]$
- [Outcome 2]: $[Result]$

**Safety:** [SAEs, discontinuations]

---

## 5. Discussion & Conclusions

**Main Interpretation:** [Authors' interpretation]

**Clinical Implications:** [Claimed implications]

**Future Research:** [Proposed directions]

---

## 6. Limitations

**Author-Acknowledged:**
- [Limitation 1]
- [Limitation 2]

**Additional (Appraiser-Identified):**
- [Limitation 3]

---

## 7. Funding & Conflicts of Interest

**Funding:** [Source(s)]

**Author COI:** [Summary]

**Assessment:** [Potential influence evaluation]

---

## 8. Critical Appraisal

### Internal Validity
| Domain | Assessment | Notes |
|--------|------------|-------|
| Randomisation | Adequate/Unclear/Inadequate | [Notes] |
| Allocation Concealment | Adequate/Unclear/Inadequate | [Notes] |
| Blinding | Adequate/Unclear/Inadequate | [Notes] |
| Outcome Measurement | Adequate/Unclear/Inadequate | [Notes] |
| Missing Data | Adequate/Unclear/Inadequate | [Notes] |

### Risk of Bias
| Bias Type | Risk Level | Justification |
|-----------|------------|---------------|
| Selection | Low/High/Unclear | [Justification] |
| Performance | Low/High/Unclear | [Justification] |
| Detection | Low/High/Unclear | [Justification] |
| Attrition | Low/High/Unclear | [Justification] |
| Reporting | Low/High/Unclear | [Justification] |

### External Validity
[Assessment of generalisability]

### Overall Evidence Strength
**Rating:** High / Moderate / Low

**Justification:** [One sentence]

---

*Appraisal generated using Clinical Article Appraisal Skill*

Template: appraisal-summary.csv

Article,Design,Population N,Intervention,Comparator,Primary Outcome,Effect Estimate,95% CI,p-value,Overall RoB,Evidence Strength,Key Limitation
"[Author Year]","[RCT/Cohort/etc]",[N],"[Intervention]","[Control]","[Outcome]","[Effect]","[CI]","[p]","[Low/High/Unclear]","[High/Moderate/Low]","[Main limitation]"

Template: appraisal-[AuthorYear].html

<!DOCTYPE html>
<html>
<head>
    <title>Article Appraisal - [Author Year]</title>
    <script src="https://polyfill.io/v3/polyfill.min.js?features=es6"></script>
    <script id="MathJax-script" async src="https://cdn.jsdelivr.net/npm/mathjax@3/es5/tex-mml-chtml.js"></script>
    <style>
        body { font-family: 'Segoe UI', Arial, sans-serif; max-width: 900px; margin: 0 auto; padding: 20px; line-height: 1.6; }
        h1 { color: #2c3e50; border-bottom: 3px solid #3498db; padding-bottom: 10px; }
        h2 { color: #34495e; margin-top: 30px; border-left: 4px solid #3498db; padding-left: 10px; }
        .highlights { background: #f8f9fa; border-left: 4px solid #27ae60; padding: 15px; margin: 20px 0; }
        .highlights ul { margin: 0; }
        table { border-collapse: collapse; width: 100%; margin: 15px 0; }
        th, td { border: 1px solid #ddd; padding: 10px; text-align: left; }
        th { background-color: #3498db; color: white; }
        tr:nth-child(even) { background-color: #f9f9f9; }
        .rating-high { color: #27ae60; font-weight: bold; }
        .rating-moderate { color: #f39c12; font-weight: bold; }
        .rating-low { color: #e74c3c; font-weight: bold; }
        .metadata { color: #7f8c8d; font-size: 0.9em; margin-bottom: 20px; }
    </style>
</head>
<body>
    <h1>Clinical Article Appraisal</h1>
    <div class="metadata">
        <strong>Article:</strong> [Full citation]<br>
        <strong>Date:</strong> [Today's date]
    </div>

    <div class="highlights">
        <h3>📌 Highlights</h3>
        <ul>
            <li>[Key finding 1]</li>
            <li>[Key finding 2]</li>
            <li>[Key finding 3]</li>
        </ul>
    </div>

    <!-- Continue with all sections... -->

</body>
</html>

User Prompt

After completing the appraisal:

📄 I can create exportable files for you:

1. **appraisal-[AuthorYear].md** - Full Markdown appraisal
2. **appraisal-[AuthorYear].txt** - Plain text for Google Docs
3. **appraisal-[AuthorYear].html** - Formatted HTML with LaTeX rendering
4. **appraisal-summary.csv** - Key data row for spreadsheet comparison

Which files would you like me to create?

User Input

$ARGUMENTS