RevMan Funnel Plot

Overview

Generates a funnel plot for publication bias assessment. Includes pseudo-95% confidence interval lines and automatic Egger's regression test.

Command

revman funnel [--outcome <id>] [--output <file>]

Procedure:

1. If --outcome specified, read study data from the active review.

   • Determine active review from .revmanrc.
   • Read study files in {REVIEW_DIR}/.revman/studies/.
   • For each study with the specified outcome, extract TE and seTE.
   • Pooled estimate from fixed-effects inverse-variance.
   • If not, proceed with manual input.

2. If no outcome specified, ask:

   Funnel plot data source:
     1. Auto-generate from study outcome data in active review
     2. Manual entry
   

3. For manual entry, collect studies:

   • "Study label:" → label
   • "Effect estimate (TE):" → TE
   • "Standard error (seTE):" → seTE
   • "Add another study? (y/n):"

4. Build JSON input:

   {
     "studies": [
       {"label": "Smith 2020", "TE": -0.5, "seTE": 0.3},
       {"label": "Jones 2021", "TE": -0.8, "seTE": 0.4}
     ],
     "pooled_te": -0.52,
     "effect_measure": "OR",
     "scale": "log"
   }
   

5. Run:

   echo '<json>' | python3 scripts/revman-funnel.py -o {output_path}
   

6. Output defaults to funnel_plot.png in the review root.

7. Confirm and display Egger's test result:

   Funnel plot saved to reviews/{name}/funnel_plot.png
   Studies: {N}
   Egger's test: intercept = {x}, p = {y}
   

---

Script Location

scripts/revman-funnel.py — Python 3. Accepts JSON on stdin, writes PNG. Runs Egger's regression test automatically when ≥ 3 studies.

---

Error Handling

| Situation | Response | |-----------|----------| | < 3 studies | "Funnel plot requires at least 3 studies for meaningful interpretation. Generated without Egger's test." | | < 10 studies | "Warning: fewer than 10 studies. Egger's test has low power." | | Script not found | "Funnel plot script not found." |