Evidence · Trials · continued

Adjudicated events and why the definition matters — the long version posts 91–120

Evidence › Trials intention to treat estimand randomised trial discontinuation & dropout

This is a continuation of a long topic, addressed by post number rather than by page. Start at post 1.

methods_draftTL2Member10 Apr 2025 · edited#91

Post #90 is right about the mechanism and I think understates the practical bit.

Surrogate endpoints: an endpoint that is not the outcome that matters but is measured as a stand-in. HbA1c is a surrogate for long-term glucose control and the short-term complications it prevents. Weight loss is a surrogate for metabolic health and long-term outcomes. Surrogates are useful but not identical to the endpoint that matters.

Two sources, same conclusion, and I could not rule out that one copied the other.

1 like 16mo

t.vargaTL211 Apr 2025#92

in.guerrero, post #72: Seconded. It reads as careful rather than confident, which is the right register. Go to post

Coming back to post #88, because the follow-up matters more than the original answer.

Adding a small correction to the Adjudicated events summary above rather than a disagreement with it. The substance holds; one of the figures is out by a factor that matters.

6 likes in reply to #72 16mo

gradient_reviewTL2Member11 Apr 2025#93

What I want from this Adjudicated events thread is the list of things that would need to be true for the claim to hold. If we can write that list, we can check it.

22 likes 16mo

br.wikstromTL212 Apr 2025#94

Adding thanks rather than a view. I do not have a view worth the space.

0 likes 16mo

RodriguesTL3Regular12 Apr 2025#95

e.ferreira, post #43: Coming back to post #40, because the follow-up matters more than the original answer. On Adjudicated events: the maintained page in the documentation commons covers the general case with citations and a review date, which is more reliable than any reply here including this one. Go to post

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.

Reading it back, the second half matters more than the first.

3 likes in reply to #43 16mo

an.zamoraTL213 Apr 2025#96

p.diallo, post #65: Answering the question post #61 raises rather than the one it answers. The reason Adjudicated events keeps being re-asked is that the answer is conditional and people quote it without the condition. It is not that the answer is unknown. Go to post

I had written a reply contradicting post #92 and deleted it. Here is what survived.

The figure that circulates in coverage is almost always whichever estimand gives the larger effect. That is not fraud; it is selection, and it is why the paper matters more than the summary.

I would want to see it done twice before believing it once.

10 likes in reply to #65 15mo

SHermansenTL2Member13 Apr 2025#97

One caution on Adjudicated events: everything above assumes the underlying documentation is what it claims to be. That assumption is doing real work and is rarely stated.

30 likes 15mo

k.ogunleyeTL214 Apr 2025#98

I have been on both sides of the Adjudicated events argument in this category within eighteen months, which should tell you how strong the evidence for either side is.

0 likes 15mo

a.thorneTL2Wiki editor15 Apr 2025#99

Fine by me. I had wanted a stronger conclusion and there is not one available.

6 likes 15mo

j.sandvikTL215 Apr 2025#100

a.thorne, post #99: Fine by me. I had wanted a stronger conclusion and there is not one available. 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.

15 likes in reply to #99 15mo

an.zamoraTL216 Apr 2025#101

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

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

SHermansenTL2Member16 Apr 2025#102

Subgroup analyses are hypothesis-generating unless pre-specified and adequately powered, and almost none are the second. The interaction test matters more than the subgroup point estimate.

That matches what I was told, which is not the same as knowing it.

4 likes 15mo

k.ogunleyeTL217 Apr 2025#103

i.grimaldi, post #76: I read post #74 twice before replying, because I had assumed the opposite. Adding a null result on Adjudicated events. I looked, carefully, and found nothing, and null results deserve posting precisely because they never are. Go to post

The thing about Adjudicated events 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.

12 likes in reply to #76 15mo

methods_draftTL2Member17 Apr 2025#104

j.sandvik, post #100: 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 keep Adjudicated events and the decision it usually gets used for separate in this thread. They are related and they are not the same question, and merging them is why the last one went badly.

25 likes in reply to #100 15mo

j.sandvikTL218 Apr 2025#105

Post #104 answers the question as asked. The question underneath it is different.

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.

1 like 15mo

HadjipaterasTL1Member19 Apr 2025#106

I read post #102 twice before replying, because I had assumed the opposite.

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.

7 likes 15mo

p.trevinoTL219 Apr 2025 · edited#107

bench_entry, post #68: Subgroup analyses are hypothesis-generating unless pre-specified and adequately powered, and almost none are the second. The interaction test matters more than the subgroup point estimate. That is one dataset and I would not build a rule on it. Go to post

The version of Adjudicated events that I was taught turned out to be a teaching simplification. Useful, and not true in the way I had assumed it was.

17 likes in reply to #68 15mo

RodriguesTL3Regular20 Apr 2025#108

Adjudicated events is a good example of a question where the honest answer is boring and the interesting answers are unsupported. I would go with boring.

0 likes 15mo

g.radichTL220 Apr 2025#109

Written quickly, so the reasoning may be tighter than the wording.

0 likes 15mo

j.rasmussenTL221 Apr 2025#110

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

Adjudicated events is worth one more sentence than it usually gets, and the sentence is the one about how the number was arrived at.

l.sarkissianTL2Member21 Apr 2025#111

Multiplicity and multiple comparisons: if a trial tests many hypotheses, the chance of a false positive on at least one by random chance increases. This is why pre-specification of the primary endpoint matters and why secondary endpoints are weaker evidence.

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

3 likes 15mo

t.marchettiTL222 Apr 2025#112

gradient_review, post #93: What I want from this Adjudicated events thread is the list of things that would need to be true for the claim to hold. If we can write that list, we can check it. Go to post

Narrowing post #109, because the general version has more than one answer.

0 likes in reply to #93 15mo

IHollingworthTL2Member22 Apr 2025#113

t.vasquez, post #57: I have three months of notes on Adjudicated events and the honest summary is that the trend is real and the week-to-week numbers are noise. I nearly drew the opposite conclusion from the first fortnight. Go to post

Post #109 put the caveat in the right place and I want to underline it.

Checked the Adjudicated events claim against the primary source this morning. It survives, with a narrower scope than the version quoted here. Posting the narrower scope.

31 likes in reply to #57 15mo

p.novakTL223 Apr 2025#114

I read the earlier replies on Adjudicated events twice before writing this, because I had assumed the opposite and wanted to be sure I was disagreeing with what was said rather than what I expected.

16 likes 15mo

e.almeidaTL2Member23 Apr 2025#115

I read post #113 twice before replying, because I had assumed the opposite.

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.

6 likes 15mo

r.sobczakTL224 Apr 2025#116

Post #113 answers the question as asked. The question underneath it is different.

Dropout is information: high dropout rates can indicate tolerability problems or lower efficacy than the summary suggests. Where the analysis handled dropouts matters. An intention-to-treat analysis with many dropouts can give a smaller apparent effect than per-protocol analysis.

1 like 15mo

r.jhannsdttirTL3Regular25 Apr 2025 · edited#117

in.guerrero, post #72: Seconded. It reads as careful rather than confident, which is the right register. Go to post

On Adjudicated events I would separate what is worth knowing from what is worth acting on. The first list is long and the second is short, and conflating them is how threads get heated.

0 likes in reply to #72 15mo

ni.kravchenkoTL225 Apr 2025#118

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

22 likes 15mo

isotonic_sheetTL3Regular26 Apr 2025#119

an.zamora, post #101: This is the sort of exchange that makes the archive worth searching. Go to post

The bit of Adjudicated events that nobody enjoys is that the answer changes depending on what you are trying to decide with it. Say what the decision is and the thread will converge.

0 likes in reply to #101 15mo

p.krastevTL226 Apr 2025#120

erratum_file, post #75: Adjudicated events would be much easier to settle if anyone reported the denominator. Almost nobody reports the denominator. Go to post

Caveat: everything above assumes the paperwork is what it says it is.

32 likes in reply to #75 15mo