Repository navigation
Do not read an allocation ratio as a sample size - #83
Merged
Merged
Conversation
The screening-count fix shipped with the retatrutide page reordered the sample-size patterns, and that reordering broke five rows while fixing four. Moving the noun pattern below the bare verb pattern let "randomized 1:1" win: kisspeptin went from 62 to 1, survodutide from 387 to 1, thymosin-alpha-1 from 194 to 1, semax from 43 to 2 and tesamorelin from 152 to 3. None of those reached a record, because the drafter never regenerates a record that has guide sections, but the next re-draft would have written them. The noun pattern goes back above the verb pattern, keeping the ordering that was right before, with assignment phrasing still ahead of both so CONSORT abstracts read correctly. A number immediately followed by a colon is refused outright. Both context checks now look at what follows the captured number rather than the whole match. Anchoring them to the match end rejected the very case the change existed to fix, because ASSIGNED ends after "randomly assigned" and the next character in "assigned: 134 to retatrutide 4 mg" is a colon. Two more screening counts corrected on retatrutide: the body-composition substudy stored 534 where 534 were screened and 281 enrolled, and the cardiovascular trial stored 367 where 367 were screened and 146 randomised. A full re-check of all 223 stored sample sizes says this class is now closed. Of 64 rows where the reader disagrees with what is stored, 55 disagree with the pre-existing reader too, which means they were corrected by hand and must not be overwritten; the audit cannot tell a drafter bug from a human correction, so it is not a list of fixes to apply. The 9 rows this change moved were each checked against their abstract: 4 were genuine screening counts, all on retatrutide, and 5 were the regressions above. 36 drafter tests, up from 29. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The screening-count fix that shipped with the retatrutide page reordered the sample-size patterns. That reordering fixed four rows and broke five. This corrects it.
The regression
Moving the noun pattern below the bare verb pattern let
randomized 1:1win, becauserandomi[sz]ed\s+(\d+)matches the allocation ratio.None reached a record, because the drafter never regenerates a record that has guide sections. The next re-draft would have written them.
The fix
The noun pattern goes back above the verb pattern, with assignment phrasing still ahead of both so CONSORT abstracts read correctly. A number immediately followed by a colon is refused.
Both context checks now look at what follows the captured number rather than the whole match. Anchoring to the match end rejected the very case the change existed to fix:
ASSIGNEDends after "randomly assigned", and the next character inassigned: 134 to retatrutide 4 mgis a colon.Two more genuine corrections
pmid-40609566pmid-41160422The site-wide re-check, and why it is not a fix list
All 223 stored sample sizes were re-read against their abstracts. 64 disagree with the reader, but 55 of those disagree with the pre-existing reader too, which means they were corrected by hand at some point and must not be overwritten. The audit cannot distinguish a drafter bug from a human correction, so it is evidence, not a patch.
The 9 rows this change actually moved were each checked against their abstract by hand: 4 genuine screening counts, all on retatrutide, and the 5 regressions above.
Tests
36 drafter tests, up from 29. New cases cover allocation ratios of both
1:1and1:1:1:1:1, the colon that follows an assignment verb, and both phrasings of a screening count.🤖 Generated with Claude Code