Evidence · Trials · continued

How to read a forest plot, properly, from scratch posts 61–82

Evidence › Trials randomised trial intention to treat surrogate endpoints estimand Solved

This is a continuation of a long topic, addressed by post number rather than by page. Start at post 1 · go to the accepted answer.

s.antonsenTL226 Nov 2025#61

Risk of bias: structured appraisal of internal validity. Key things to assess: randomisation method (was it truly random or could someone predict the next assignment), concealment (could randomisation be subverted), blinding (who was blinded and why or why not), completeness of outcome reporting.

Where I would look next, rather than where I would stop.

1 like 8mo

forest_plotTL3Evidence synthesis28 Nov 2025 · edited#62

i.lehtinen, post #56: I had written a reply contradicting post #55 and deleted it. Here is what survived. Absolute numbers, not just relative: a 30% relative reduction tells you the ratio but not the practical magnitude. The event rate in each arm and the difference between them tells you how many people benefit. Go to post

A treatment-policy estimand asks what happens to people assigned to a strategy, including those who abandon it. A hypothetical estimand asks what would have happened had everyone continued. Both are legitimate and they give different numbers.

0 likes in reply to #56 8mo

a.pereiraTL230 Nov 2025#63

Collapsed as off-topic by two members at trust level 3 or above

Answering the question post #59 raises rather than the one it answers.

Composite endpoints should be read component by component. A composite driven entirely by its softest component is a different finding from one where the components move together.

The step people skip is the one I have spelled out.

ppm_errorTL3Analytical chemist2 Dec 2025#64

Worth checking against a second source before it gets quoted onward.

6 likes 8mo

y.ibarraTL24 Dec 2025#65

Absolute numbers, not just relative: a 30% relative reduction tells you the ratio but not the practical magnitude. The event rate in each arm and the difference between them tells you how many people benefit.

3 likes 8mo

d.moreauTL2Regular6 Dec 2025#66

s.cardoso, post #1: The question in the title: How to read a forest plot, properly, from scratch I will give what I have already checked below so nobody repeats it. Reading STEP 4 ( JAMA , 2021) for the population rather than the effect, which I have not done properly before. The baseline table is more restrictive than the way the trial gets discussed… Go to post

Post #64 is the version of this I will quote in future. One addition.

Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers.

0 likes in reply to #1 8mo

r.lundgrenTL28 Dec 2025#67

unit_conversion, post #53: Adding the measurement that post #52 says would settle it. A treatment-policy estimand asks what happens to people assigned to a strategy, including those who abandon it. A hypothetical estimand asks what would have happened had everyone continued. Both are legitimate and they give different numbers. Go to post

I would put a moderate confidence on that and no more.

23 likes in reply to #53 8mo

q.zhao_qaTL3Quality assurance10 Dec 2025#68

I checked the source rather than the summary, and they differ.

10 likes 8mo

v.bruunTL212 Dec 2025 · edited#69

Worth separating two things that post #67 runs together.

6 likes 8mo

nl_translatorTL2Translator · NL14 Dec 2025#70

Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers.

I have left out the parts I could not verify.

1 like 7mo

m.duarteTL216 Dec 2025#71

3 likes 7mo

o.vogelTL218 Dec 2025#72

r.frisk, post #35: Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers. Go to post

Grateful for the specificity. Vague answers to this question are what sent me looking.

11 likes in reply to #35 7mo

a.zamoraTL220 Dec 2025#73

f.haddad, post #52: Where I part company with post #50, and it is a narrow parting. Absolute numbers, not just relative: a 30% relative reduction tells you the ratio but not the practical magnitude. The event rate in each arm and the difference between them tells you how many people benefit. I would treat the number as indicative rather than as a… Go to post

Stating my assumptions rather than smuggling them in.

33 likes in reply to #52 7mo

d.tammTL222 Dec 2025#74

0 likes 7mo

a.nwosuTL224 Dec 2025 · edited#75

Picking up post #74: that is the part I would want checked first.

I am describing what is, rather than arguing for what should be.

1 like 7mo

a.batistaTL226 Dec 2025#76

cohort_drift, post #45: I read post #41 twice before replying, because I had assumed the opposite. Composite endpoints should be read component by component. A composite driven entirely by its softest component is a different finding from one where the components move together. Take the reasoning and check the arithmetic; I do not always get it right. Go to post

On post #73 — agreed on the reasoning, with one qualification.

Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers.

That is all the detail I have. Someone else will have more.

7 likes in reply to #45 7mo

k.perrinTL228 Dec 2025#77

25 likes 7mo

bench_notesTL4 Moderator30 Dec 2025#78

0 likes 7mo

e.vargaTL21 Jan 2026#79

y.mensah, post #5: A treatment-policy estimand asks what happens to people assigned to a strategy, including those who abandon it. A hypothetical estimand asks what would have happened had everyone continued. Both are legitimate and they give different numbers. Go to post

That is my reading. Someone else read the same page differently and was reasonable.

11 likes in reply to #5 7mo

r.aldana_pharmdTL4Pharmacist2 Jan 2026#80

plateau_notes, post #22: Sensible. I would want the same detail before I acted on it either. Go to post

Discontinuation handling is the methodological detail that most changes a result and gets the least attention. Read how missing data was imputed before reading the effect size.

I would not lead a decision with this, but I would not ignore it either.

24 likes in reply to #22 7mo

forest_plotTL3Evidence synthesis4 Jan 2026#81

v.kirchner, post #16: Absolute numbers, not just relative: a 30% relative reduction tells you the ratio but not the practical magnitude. The event rate in each arm and the difference between them tells you how many people benefit. Second-hand, so weight it accordingly. Go to post

Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers.

If the premise is wrong, everything after it is decoration.

7 likes in reply to #16 7mo

n.villalobosTL26 Jan 2026#82

e.varga, post #79: A treatment-policy estimand asks what happens to people assigned to a strategy, including those who abandon it. A hypothetical estimand asks what would have happened had everyone continued. Both are legitimate and they give different numbers. That is my reading. Someone else read the same page differently and was reasonable. Go to post

Adding the measurement that post #81 says would settle it.

Correct me on the arithmetic if it is wrong; I would rather know.

1 like in reply to #79 7mo

This topic was referenced in

[2026 update] How to read a forest plot, properly, from scratchEvidence › Trials · 53 replies

Promoted into the documentation commons. The content of this topic is maintained at LEADER — trial digest, with named maintainers and a review date. The promotion was discussed in doc review. Corrections are best raised against the document, which is the version that gets kept current.

Topic	Participants	Replies	Views	Activity
Interim analyses and stopping rules Evidence › Trials intention to treat surrogate endpoints open label discontinuation & dropout Interim analyses and stopping rules Writing it up because I had to work it out twice and would rather nobody else did. Asking about interim analyses and stopping rules on behalf of the question I keep seeing…	+25	29	35k	16d
Interim analyses and stopping rules — does this still hold? Evidence › Trials estimand discontinuation & dropout open label surrogate endpoints Solved Interim analyses and stopping rules — does this still hold? — that is the question, and I have not found it answered plainly anywhere I have looked. Interim analyses and stopping rules, from the point of view…	+30	34	29k	13mo
[2026 update] How to read a forest plot, properly, from scratch Evidence › Trials estimand discontinuation & dropout surrogate endpoints Solved How to read a forest plot, properly, from scratch — that is the question, and I have not found it answered plainly anywhere I have looked. Reading SURMOUNT-4 ( JAMA , 2024) for the population rather than the…	+49	53	643	19mo
Reading a trial's population section before its results — does this still hold? Evidence › Trials randomised trial open label intention to treat discontinuation & dropout Closed Reading a trial's population section before its results — does this still hold? I have a specific reason for asking rather than idle curiosity, and the context is below. Reading STEP 1 ( N Engl J Med , 2021)…	+140	154	5.4k	14mo
Trial registration and comparing the protocol with the paper Evidence › Trials randomised trial open label intention to treat Trial registration and comparing the protocol with the paper — setting out what I have, and where I think it stops being reliable. I have spent a fortnight trying to pin trial registration down and I want to…	+38	44	4.8k	10mo

Topic	Participants	Replies	Views	Activity
Journal club: STEP-HFpEF and symptom endpoints Evidence › Journal club journal club effect size estimand risk of bias Journal club: STEP-HFpEF and symptom endpoints — setting out what I have, and where I think it stops being reliable. The question about step-hfpef and symptom endpoints that I actually want answered is the…	+18	22	2.5k	11mo
Composite endpoints and the component doing the work Evidence › Trials randomised trial surrogate endpoints open label discontinuation & dropout Solved Under review Composite endpoints and the component doing the work Writing it up because I had to work it out twice and would rather nobody else did. Collecting what is known about composite endpoints in one place, because…	+3	7	33k	13h
Second pass at: Historical amylin analogues and what happened to them Compounds › Cagrilintide & amylin analogues amylin analogues effect size cagrilintide cagrisema Closed Second pass at: Historical amylin analogues and what happened to them — setting out what I have, and where I think it stops being reliable. The question about Historical amylin analogues that I actually want…	+15	19	22k	18mo
Journal club: SUSTAIN 6 and its retinopathy signal Evidence › Journal club estimand effect size risk of bias journal club Under review On the subject in the title: Journal club: SUSTAIN 6 and its retinopathy signal Working notes rather than a conclusion. Reading SUSTAIN 6 ( N Engl J Med , 2016) for the population rather than the effect,…	+110	122	46k	1d
Journal club: STEP 8 and the fairness of the comparator dose Evidence › Journal club estimand number needed to treat journal club Journal club: STEP 8 and the fairness of the comparator dose — setting out what I have, and where I think it stops being reliable. Session topic: STEP 8 ( JAMA , 2022). Please read it before posting; the…	+2	6	39k	12mo

How to read a forest plot, properly, from scratch posts 61–82

This topic was referenced in

Suggested topics

Related topics — sharing the tags randomised trial, intention to treat, surrogate endpoints