The out-of-range flag is a statistical convention with a printer attached
A flag is a probability statement about a population. It is not a statement about the person holding the printout.
Clinical Trials in 2025
Page 6 of 11 of this archive, newest first.
STEP 1 extension data: what happens after the trial stops
A design note rather than a result: what the comparator was, and what that permits you to conclude.
Percentage of loss, absolute kilograms, and the sleight of hand between them
What was pre-specified, what was exploratory, and what was calculated afterwards by people who did not run the trial.
SURMOUNT-4 extension data: what happens after the trial stops
A design note rather than a result: what the comparator was, and what that permits you to conclude.
Every fortnight, every third week, seasonally: an unstudied practice
An absence of evidence is not evidence of harm. It is also not a licence.
SURPASS-4 misses on a secondary endpoint the coverage has not mentioned
A design note rather than a result: what the comparator was, and what that permits you to conclude.
A reference interval is not a target, and a result outside one is not a diagnosis
Almost every misreading of a laboratory panel is a misunderstanding of what a reference interval is and how much a result has to move before the movement means anything.
STEP 4 extension data: what happens after the trial stops
A design note rather than a result: what the comparator was, and what that permits you to conclude.
SURMOUNT-OSA misses on a secondary endpoint the coverage has not mentioned
A design note rather than a result: what the comparator was, and what that permits you to conclude.
What the older diet-and-exercise trials found in the hip
A plausible mechanism, a measurable change, and no outcome data. This is what an open question looks like.
LEADER misses on a secondary endpoint the coverage has not mentioned
A design note rather than a result: what the comparator was, and what that permits you to conclude.
SURMOUNT-2 was built to answer the stopping question, and it did
Three randomised withdrawal designs have tested what happens when treatment stops. Their results are consistent and they are consistently misreported.
STEP 8 misses on a secondary endpoint the coverage has not mentioned
A design note rather than a result: what the comparator was, and what that permits you to conclude.
The three-times-upper-limit convention, and where it came from
What the renal and hepatic outcome programmes actually measured, and over what duration.
What the protein literature actually shows, and in whom it was shown
Where the familiar figures come from, what populations they were measured in, and how far the extrapolation reaches.
SURMOUNT-4 extension data: what happens after the trial stops
A design note rather than a result: what the comparator was, and what that permits you to conclude.
REWIND misses on a secondary endpoint the coverage has not mentioned
A design note rather than a result: what the comparator was, and what that permits you to conclude.
STEP 1 extension data: what happens after the trial stops
A design note rather than a result: what the comparator was, and what that permits you to conclude.
Coming off: what the pharmacology permits
What a slow reduction could plausibly buy, and what it certainly cannot prevent.
SURPASS-2 extension data: what happens after the trial stops
A design note rather than a result: what the comparator was, and what that permits you to conclude.