Evidence · Trials

[2026 update] Composite endpoints and the component doing the work

Evidence › Trials surrogate endpoints discontinuation & dropout estimand

curious_readerTL1Member9 Feb 2026#1

On the subject in the title: Composite endpoints and the component doing the work Working notes rather than a conclusion.

Comparing SURMOUNT-1 (N Engl J Med, 2022) with SURPASS-4 (Lancet, 2021) and finding the comparison harder than it looks.

Different populations, different durations, different endpoints defined slightly differently, and in one case a different estimand. People compare the headline percentages anyway, including me until recently.

Is there a defensible way to put these side by side, or is the honest answer that there is not and we should stop?

3 likes 6mo

sterile_fileTL3Regular11 Feb 2026#2

Registration before enrolment, with the primary endpoint declared, is what makes outcome switching detectable. Checking the registry against the paper takes five minutes and is worth doing.

7 likes 5mo

d.vestergaardTL213 Feb 2026#3

sterile_file, post #2: Registration before enrolment, with the primary endpoint declared, is what makes outcome switching detectable. Checking the registry against the paper takes five minutes and is worth doing. Go to post

Narrowing post #2, because the general version has more than one answer.

A definition problem is doing most of the work in this Composite endpoints discussion. Once the term is pinned down I suspect the disagreement mostly goes away and what is left is small.

18 likes in reply to #2 5mo

FairweatherTL2Member15 Feb 2026 · edited#4

Everything in the opening post holds. The case it does not cover is the one I have.

What I would check first on Composite endpoints is whether the thing being measured moved or whether the way of measuring it moved. Those look identical in a graph.

0 likes 5mo

j.palaciosTL216 Feb 2026#5

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 5mo

vial_tableTL217 Feb 2026#6

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

Composite endpoints was covered in the wiki last year and the page has a review date on it, which is a better starting point than my memory of a thread.

g.oyelaranTL219 Feb 2026#7

vial_table, post #6: Composite endpoints was covered in the wiki last year and the page has a review date on it, which is a better starting point than my memory of a thread. Go to post

The arithmetic in post #5 is right; the assumption feeding it is the part to check.

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.

Old habit: I write down the expected answer before I calculate it.

12 likes in reply to #6 5mo

integrator_logTL3Regular20 Feb 2026#8

That reframing is the whole thing. The facts I already had.

26 likes 5mo

b.friskTL221 Feb 2026#9

Composite endpoints has a well-known answer and a correct answer, and the interesting work is establishing that they are the same. Nobody has done that here yet.

7 likes 5mo

t.kulkarniTL3Regular22 Feb 2026#10

Where I part company with post #9, and it is a narrow parting.

Adding what did not work for me on Composite endpoints, since the failures never get written up and they are half the useful information.

17 likes 5mo

FairweatherTL2Member23 Feb 2026#11

Fairweather, post #4: Everything in the opening post holds. The case it does not cover is the one I have. What I would check first on Composite endpoints is whether the thing being measured moved or whether the way of measuring it moved. Those look identical in a graph. Go to post

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.

It is one reading of the data and not the only reasonable one.

32 likes in reply to #4 5mo

h.kimaniTL224 Feb 2026#12

Post #11 is the version of this I will quote in future. One addition.

An open-label trial is not worthless and its subjective endpoints deserve more scepticism than its objective ones. That is a graded judgement rather than a verdict.

That is a description of practice, not a recommendation of it.

17 likes 5mo

sterile_fileTL3Regular26 Feb 2026#13

I had read the opposite somewhere and cannot now find where, which tells me something.

3 likes 5mo

ka.batistaTL227 Feb 2026#14

integrator_log, post #8: That reframing is the whole thing. The facts I already had. Go to post

I keep a log for Composite endpoints specifically because my memory of it turned out to be systematically wrong in one direction. Six weeks of notes cost nothing and settled it.

0 likes in reply to #8 5mo

h.almeidaTL2Member28 Feb 2026#15

Small correction to my own earlier position on Composite endpoints. I had the units the wrong way round, which changes the conclusion by an order of magnitude and therefore changes it entirely.

0 likes 5mo

g.oyelaranTL21 Mar 2026#16

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

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.

Not disagreeing with anyone above, just adding the bit I keep having to look up.

BDraganovTL2Member2 Mar 2026 · edited#17

Post #15 describes the usual case. This is about the unusual one.

Where I have landed on Composite endpoints, having got it wrong once in public: the direction is clear, the magnitude is not, and anyone quoting a precise magnitude has borrowed it from somewhere that did not measure it.

6 likes 5mo

j.palaciosTL23 Mar 2026#18

integrator_log, post #8: That reframing is the whole thing. The facts I already had. Go to post

Adding the measurement that post #15 says would settle it.

The confident answers on Composite endpoints and the well-sourced answers are not the same answers, which is the most useful thing I have learned reading this category.

1 like in reply to #8 5mo

r.jhannsdttirTL3Regular4 Mar 2026#19

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 5mo

v.bergstromTL25 Mar 2026 · edited#20

A request rather than an answer: could whoever has the primary source for Composite endpoints post it? I have seen the claim three times this month and each version had lost a qualifier.

31 likes 5mo

n.zielinskiTL26 Mar 2026#21

Composite endpoints looks different depending on whether you are reading the primary literature or the summaries of it, and the difference is not in our favour.

13 likes 5mo

MJayawardenaTL3Regular7 Mar 2026#22

t.kulkarni, post #10: Where I part company with post #9, and it is a narrow parting. Adding what did not work for me on Composite endpoints, since the failures never get written up and they are half the useful information. Go to post

On post #18 — agreed on the reasoning, with one qualification.

27 likes in reply to #10 5mo

s.oyelaranTL27 Mar 2026#23

The practical version of Composite endpoints is three sentences long. The rigorous version is three pages and reaches the same conclusion with the conditions attached.

0 likes 5mo

OTeixeiraTL38 Mar 2026#24

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

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.

A modest claim, modestly supported.

ar.kravchenkoTL29 Mar 2026#25

Adding the measurement that post #23 says would settle it.

Two questions I would want answered before drawing anything from the Composite endpoints data above: how were the cases selected, and what happened to the ones that dropped out.

8 likes 5mo

e.dalgleishTL3Regular10 Mar 2026#26

ka.batista, post #14: I keep a log for Composite endpoints specifically because my memory of it turned out to be systematically wrong in one direction. Six weeks of notes cost nothing and settled it. Go to post

Post #22 describes the usual case. This is about the unusual one.

On Composite endpoints the community has more anecdote than the confidence in this thread implies, and I include my own contribution in that.

20 likes in reply to #14 5mo

b.teixeiraTL211 Mar 2026#27

The evidence for this is thinner than the way I have phrased it suggests.

0 likes 5mo

endpoint_marginTL2Member12 Mar 2026#28

This is the sort of exchange that makes the archive worth searching.

0 likes 5mo

s.beaulieuTL213 Mar 2026#29

The thing about Composite endpoints that took me longest to accept is that a plausible mechanism is not evidence of an effect. It is a reason to look, not a result.

5 likes 5mo

f.abrahamsenTL2Member14 Mar 2026#30

I have changed my mind on this once already, so take it as current rather than settled.

14 likes 4mo

Sponsored Janoshik Analytical — independent third-party testingHPLC purity and mass-spectrometric identity on submitted samples, reported with the method stated. The laboratory most often cited in this community’s verification records.Independent testing → Sponsored GL Biochem (Shanghai) Ltd. — synthesis, directFounded 1998. ISO 9001 and cGMP certified, over 1,500 staff and 200-plus patents. Custom peptide synthesis and amino acid derivatives, with a batch certificate on every order.Custom synthesis →