Revision aid, not clinical advice. This page is educational material written for exam preparation. It is not a substitute for the current guideline or for clinical judgement, and must not be relied on for the care of any individual. Always check up-to-date national guidance and verify independently before acting on anything here.

AKT Statistics Guide · Part 5 · Chapter 16

Forest Plots & Meta-Analysis

How a single picture pools dozens of studies — reading the diamond, the line of no effect, and the heterogeneity behind it.

Where this shows up in the AKT

Interpreting a forest plot — deciding whether a pooled result is significant from where the diamond sits.
Reading heterogeneity (I²) and saying what a high value means for the pooled estimate.
Spotting publication bias on a funnel plot, and resisting the trap that every study must individually show benefit.

A meta-analysis sits at the top of the evidence hierarchy: it statistically combines the results of several studies into one pooled estimate. The forest plot is how that estimate is drawn.

Each study is a box (sized by its weight) with its 95% CI; the diamond is the pooled estimate; the dashed vertical is the line of no effect (RR = 1).

Anatomy of a forest plot

Read it row by row, then drop to the diamond. Every element on the plot is telling you one specific thing.

What each element of a forest plot represents.
Element	What it means
Each horizontal line	One study — its 95% CI (the line) and its point estimate (the box)
Box size	That study’s weight in the analysis — bigger box = more weight
The diamond	The pooled effect across all studies (its width is the pooled CI)
Vertical line	The line of no effect — at 1 for ratios, at 0 for differences

The line of no effect is 1 for ratios (RR, OR, HR) and 0 for differences (mean difference). Mixing those up is a classic slip.
If the diamond crosses the line of no effect, the pooled result is not significant. If it sits entirely to one side, it is significant.

Read the diamond first: it is the answer to “what did all these studies, taken together, find?” A diamond clear of the line of no effect = a significant pooled effect; a diamond straddling it = no significant effect.

Heterogeneity (I²) — the fruit-salad test

I² measures how much the individual studies disagree with one another — whether you are pooling apples with apples, or apples with oranges.

Interpreting I² on a meta-analysis.
I²	Interpretation
<25%	Low — “apples with apples”, safe to pool
25–75%	Moderate — interpret with caution
>75%	High — “fruit salad”, pooling is unreliable

A high I² means the studies are too different to combine meaningfully. You can still report the pooled estimate, but the AKT expects you to flag it as unreliable.
Fixed-effect vs random-effects in one line: a fixed-effect model assumes every study estimates the same true effect; a random-effects model allows the true effect to vary between studies and is the safer choice when heterogeneity is present.

Publication bias and the funnel plot

Publication bias: positive (statistically significant) studies are more likely to be published than negative ones, which can inflate a pooled estimate.
The funnel plot is the visual check. A symmetrical funnel suggests no publication bias; an asymmetrical one — typically a missing bottom corner — suggests the negative studies were never published.

Worked example

A meta-analysis asks whether a new antihypertensive reduces stroke risk. The pooled effect (the diamond) is a relative risk of 0.85, 95% CI 0.78–0.92, lying entirely to the left of the line of no effect.

Pooled estimate from the diamond. Ratio outcome → line of no effect at 1.
Pooled measure	Value	95% CI	Crosses 1?
Relative risk	0.85	0.78–0.92	No

The diamond sits entirely to the left of 1 and the CI does not cross 1 → the treatment significantly reduces stroke risk.
RR 0.85 means a 15% relative reduction in stroke. Individual studies in the plot may scatter either side of 1 — the pooled estimate is what the question is testing.

The traps that cost marks

Trap 1 — thinking every study must show benefit. A meta-analysis pools results; some individual studies may favour placebo. The diamond is the verdict, not any single line.

Trap 2 — confusing box size with effect size. A bigger box means more weight (usually a larger or more precise study), not a bigger effect.

Trap 3 — ignoring heterogeneity, or the wrong reference line. A high I² means the studies disagree, so interpret the pool cautiously. And the line of no effect is 1 for ratios, 0 for differences — never both. “Statistically significant” is not the same as “clinically important”.

AKT-style question

A meta-analysis of an antiplatelet drug for recurrent stroke reports a pooled odds ratio of 0.94 with a 95% CI of 0.84–1.05, so the diamond crosses the line of no effect. Two of the eight individual studies showed a significant benefit. Which interpretation is correct?

The pooled result shows no statistically significant effect
The drug significantly reduces recurrent stroke
The analysis is invalid because two studies were positive
The line of no effect should be at 0 for this outcome
A bigger box would have changed the pooled result

Reveal answer

A — no statistically significant effect. The diamond crosses the line of no effect (1 for a ratio) and the pooled CI includes 1, so the overall result is not significant — even though two individual studies were positive. Individual studies routinely scatter either side of the line; the pooled estimate is what counts. The line of no effect is correctly at 1 for an odds ratio (it would be 0 only for a difference), and box size reflects weight, not effect.

Lock it in

Each line = one study (CI + box); box size = weight; the diamond = pooled effect.
Line of no effect: 1 for ratios, 0 for differences. Diamond crossing it → not significant.
I²: <25% low, 25–75% moderate, >75% high (“fruit salad” — pool unreliable). Random-effects when heterogeneity is present.
Not every study must show benefit; an asymmetrical funnel plot suggests publication bias.

Source concepts: standard evidence-synthesis methodology (e.g. Cochrane Handbook; Centre for Evidence-Based Medicine). A revision summary, not a clinical protocol.

Want more questions like this one?

Try the free 20-question AKT taster: sixteen clinical, two evidence-based practice and two admin questions, with the reasoning after each and a marked report. No card needed. The full AKT Pack has 2,000+ more.

Note: Independent revision resource for the MRCGP AKT and the MSRA. Educational content for exam preparation only — not clinical advice, and not affiliated with the RCGP. A study aid, not a substitute for the official curriculum or clinical judgement.

← Back to the AKT Statistics Guide

by Core Revision · corerevision.com