Statistical review that thinks like a biostatistician.
Surface the estimand gaps, unjustified sample-size assumptions, missing-data hand-waving, and uncontrolled multiplicity that undermine a trial's primary result — before the protocol locks.
Power calculation assumes an effect size not justified by prior data
The protocol powers on a 12% absolute difference, but the cited pilot suggests 6–8%. At the cited variance the study is materially underpowered.
MAJORMultiplicity
Five secondary endpoints, no testing hierarchy
No alpha-allocation or hierarchical testing strategy is described, so positive secondary results cannot support confirmatory claims.
MAJOREstimand
Intercurrent-event strategy not specified
The estimand names the population and endpoint but not how treatment discontinuation is handled — treatment-policy vs. hypothetical changes the estimate.
MINORMissing data
MMRM stated without a stated missingness assumption
The primary analysis is MMRM but the protocol doesn't state the MAR assumption or a sensitivity analysis under MNAR.
The review work that eats your week
Sample-size calculations whose assumptions aren't justified or don't match the primary analysis
Estimands missing the intercurrent-event strategy required by ICH E9(R1)
Secondary endpoints with no pre-specified multiplicity control
Missing-data approaches that are stated but not reconciled with the analysis model
What TrialScope checks for you
Estimand integrity
Checks the five estimand attributes per ICH E9(R1) and flags undefined intercurrent-event handling.
Sample-size audit
Verifies the calculation states effect size, variability, alpha, power, and dropout — and that they're consistent with the primary endpoint and analysis.
Multiplicity & analysis
Checks for a pre-specified testing hierarchy and alpha allocation across endpoints, and that the primary analysis population and model are unambiguous.
Missing-data coherence
Confirms the missing-data strategy is defined and consistent with the analysis model, with sensitivity analyses.
SEE IT ON YOUR OWN DOCUMENT
Upload a protocol, CSR, manuscript, or briefing document and get an audit calibrated to biostatistician — findings, severities, and a verdict in under 90 seconds. Free to use.
Frequently asked questions
What does TrialScope check in a statistical analysis plan?
It reviews the estimand framing against ICH E9(R1), the sample-size calculation and its assumptions, the primary analysis population and model, multiplicity control across endpoints, and the missing-data strategy — flagging gaps and internal inconsistencies, ranked by impact on the primary result.
Does it replace a statistical review?
No. TrialScope is decision-support that surfaces issues quickly and consistently; the statistical and clinical judgment stays with you. It's most useful as a fast first pass before a protocol or SAP locks.
How is it calibrated for biostatistics?
Audits run through a biostatistician persona, and corrections from verified statisticians sharpen how it weights and prioritizes statistical findings over time.