The out-of-range flag is a statistical convention with a printer attached
A twelve-analyte panel in a perfectly healthy person has roughly even odds of producing at least one flagged result.
Tamsin Okonjo
Page 2 of 2 of this archive, newest first.
Archive — page 2
37 piecesMicronutrient monitoring: which panel, how often, and on whose evidence
A monitoring schedule requires evidence about incidence. For this population, that evidence does not exist.
Which results are findings and which are consequences of the weight loss
The estimated glomerular filtration rate is calculated from creatinine, creatinine comes from muscle, and muscle mass is falling. The arithmetic is unforgiving.
The reference change value: the number that tells you whether a delta is real
A flag is a probability statement about a population. It is not a statement about the person holding the printout.
Priming a pen is not optional
Aseptic technique here is four steps and none of them are difficult. They are, however, routinely compressed to two.
The convention nobody explains before handing over the vial
The graduation interval differs between barrel sizes, and a 1 mL barrel is frequently marked in two-unit steps. Reading one as though it were marked in single units halves…
What the placebo arms of the withdrawal trials actually tell us
Three randomised withdrawal designs have tested what happens when treatment stops. Their results are consistent and they are consistently misreported.
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.
Four millimetres is enough, and the data says so
A 4 mm needle at ninety degrees without a skin pinch is adequate for essentially all adults. The persistence of 12.7 mm needles in this market is habit, not reasoning.
Which results are findings and which are consequences of the weight loss
The panel drawn during a week of vomiting is measuring the vomiting.
The reference change value: the number that tells you whether a delta is real
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.
Why people actually stop, in the order they actually stop
The trials studied planned withdrawal. Almost nobody stops that way.
Why people actually stop, in the order they actually stop
A supply interruption is a discontinuation with no notice, no taper and no plan. That is a distinct clinical situation.
Why people actually stop, in the order they actually stop
Cost is the modal reason for discontinuation in every dataset we have seen, and it is absent from the clinical literature.
Randomised withdrawal is the cleanest design in this field and the least discussed
The regain trajectories, arm by arm, with the estimands named.
What has to change before a change has happened
A flag is a probability statement about a population. It is not a statement about the person holding the printout.
What a 1 mL syringe does that a 0.3 mL syringe does not
Dead space in the needle hub is a real and calculable loss, and it matters more at small volumes than anybody expects.