Underpowering is decided before the first patient enrolls, discovered after the last one leaves, and is indistinguishable at the end from the drug simply not working.
of published adjuvant phase III oncology trials had less statistical power at read-out than their own protocol specified. ESMO Open, 282 adjuvant phase III RCTs, 2013-2023
stopped recruitment without reaching the targeted sample size. ESMO Open, 282 adjuvant phase III RCTs, 2013-2023
is the gap in reaching a statistically significant result between trials without power failure and those with it. ESMO Open, 282 adjuvant phase III RCTs, 2013-2023
of superiority trials on intermediate survival endpoints reached a statistically significant result at all, and trials that lost power reached one far less often - 37.9% versus 21.9%, P = 0.04. ESMO Open, 282 adjuvant phase III RCTs, 2013-2023
The arithmetic that decides this is not difficult and it is not in dispute. It is a power calculation done early, with the dropout assumption written down, and re-checked when accrual disappoints instead of after the database locks.
Six that close the biggest gaps. 16 in the full pharma and biomedical library.
The gap it closes. Makes the assumption set explicit - effect size, variance, alpha, dropout - so an underpowered design is visible at protocol time rather than at analysis.
See it run in your browser →The gap it closes. The 80-125% window is pass or fail on the confidence interval, not the point estimate. Seeing the interval against the window is the whole determination.
See it run in your browser →The gap it closes. Whole-body compartment kinetics for exposure prediction, deterministic and reproducible, which is what an electronic-records context requires of a computation.
See it run in your browser →The gap it closes. Half-life, clearance and volume of distribution are linked; deriving one from assumed values for the others hides which was actually measured.
See it run in your browser →The gap it closes. Patent term, pediatric extension and regulatory exclusivity run on different clocks, and the binding one determines when a competitor can file.
See it run in your browser →The gap it closes. Reaction-diffusion growth with an explicit front speed, so a modelled invasion rate can be checked against the parameters that produced it.
See it run in your browser →Not medical, regulatory or statistical advice, and no substitute for a biostatistician, a clinical pharmacologist or your regulatory affairs group. These compute published methods over inputs you supply and show the assumptions used. A power calculation is only as good as the effect size you assumed, and the tool makes that assumption visible rather than burying it.
Every tool above runs entirely in this browser tab. Nothing is uploaded, nothing is sent to a model, and each result cites the rule it applied.
Free tier, no card, no install. An account saves your work and lets you export a citable record of a run. See all 16 pharma and biomedical tools.