Case Studies

Three problems sponsors brought us, what we did, and what held up afterward. Sponsors are described, not named.

Case study 01 · Biostatistics

We checked the digital twin.

The claim. A vendor claimed roughly 20% variance reduction from digital-twin controls: prognostic models that predict each participant's trajectory and are then used as covariates. Sponsors are being pitched exactly this right now.

What we did. We ran it against a completed Alzheimer's study whose pre-specified model already carried treatment-by-baseline-by-visit interactions and a second baseline covariate, and compared the variance explained with and without the twin.

What we found. The incremental benefit was close to zero. The claimed gain was real only against a model specified with baseline alone. A well-specified analysis had already captured what the twin was selling.

Why it matters. The question is never whether a method works in the abstract. It is whether it adds anything to the analysis you have already specified. That is a judgment call, and it is verifiable.

Case study 02 · Clinical Programming

Good vendors, broken trail.

The problem. A rare-disease sponsor was using well-qualified vendors on traditional scales. At inspection, the FDA found gaps in the audit trail: not in the data itself, but in the record of how it moved from raw capture to the analysis datasets.

What we did. We remediated the raw → SDTM → ADaM chain end to end, so that every value in the analysis could be traced back to its origin and every transformation was documented.

What held up. The agency re-reviewed clean.

The point. Vendor quality and trail legibility are different things, and the second one is what an inspection tests.

Case study 03 · Study Design

A trial that shouldn't have been powerable.

The problem. A tau-targeted program with a fixed budget and a sample size below what a conventional design required. On paper, the clinical-endpoint hypotheses could not be tested.

What we did. A Global Statistical Test (GST) allowed the clinical-endpoint hypotheses to be tested anyway, pre-specified, within the sample the budget could fund.

What held up. The sponsor now uses GST as standard across its programs.

Next step

Let's Talk About Your Gray Zones

Thirty minutes on your endpoint, or a fixed-scope read of your SAP before it locks. Both start with one message.

Contact Pentara