Regulatory guideUpdated September 28, 2026

FDA's Bayesian guidance, and what it means for how you build a trial

On January 9, 2026, FDA published draft guidance on using Bayesian methods to support primary inference in pivotal drug and biologic trials, covering INDs, NDAs and BLAs. It sets out three core expectations: success criteria defined by posterior probabilities, principled prior specification, and prospective evaluation of a design's operating characteristics before the trial starts.

  • Draft guidance published Jan 9, 2026
  • Comment period closed Mar 13, 2026
  • Flexible arms, epochs and randomization to build on

Free sandbox · No credit card · 21 CFR Part 11 aligned

Schedule · arms, epochs and locations
ScreeningTreatmentFollow-upSCRD1D14D28D84EOSLocationOn-siteOn-siteRemoteHomeRemoteOn-site
Arm A
Arm B

Day 28 is an Arm B only visit, so Arm A participants never see it.

Jan 9, 2026

FDA published its Bayesian methodology draft guidance

3

core expectations: posterior success criteria, principled priors, pre-specified operating characteristics

Mar 13, 2026

public comment deadline on the draft

0

statistical analysis Capture performs for you; that stays with your biostatistics team

The guidance

What FDA's draft guidance actually asks for

The guidance focuses on Bayesian methods used for primary inference in pivotal trials meant to support effectiveness and safety findings. FDA lays out three expectations it wants addressed up front, not worked out after the data comes in: how success is defined through posterior probabilities, how any borrowed prior information is justified and safeguarded, and how a design's operating characteristics, like type I error and power, are evaluated before the trial starts.

  • Applies broadly across INDs, NDAs, BLAs and their supplements.
  • A published draft, open for comment through March 13, 2026, not yet a final guidance.
  • The statistical design and analysis work itself sits with your biostatistics team, not with any EDC.
AI Study Generator
phase2_protocol_v4.pdfStep 2 of 3
  • Reading protocol
  • Building visits and windows
  • Drafting forms

11

visits

9

draft forms

2

diaries

What the platform side needs

What an adaptive or Bayesian design actually needs from your data platform

A schedule that can flex by arm

Set up arms, epochs and visit schedules that differ by arm, including added or dropped visits mid-study.

Randomization that can change without a vendor ticket

Amend the randomization list and allocation ratios yourself as an interim analysis calls for it.

Clean, query-free data feeding the interim look

Edit checks and auto-queries keep the dataset an interim analysis pulls from as clean as the final one.

A record of every design amendment

Protocol and form amendments are tracked in the same audit trail as the clinical data, so a reviewer can see exactly what changed and when.

What Capture is not

Capture builds the study. Your biostatistics software runs the model

To be direct about scope: Capture does not run Bayesian posterior calculations, define your priors, or simulate operating characteristics. That is your biostatistics team's work, done in dedicated statistical software. What Capture does is give that team clean, exportable data and a study build flexible enough to support whatever adaptive or Bayesian design they land on.

  • CSV/Excel exports with an automatic data dictionary for your statistical software.
  • Randomization and arms/epochs that can be amended as the design calls for it.
  • One audit trail across every amendment, for when a reviewer asks how a mid-study change was made.
Data and exports
Clinical data (codes + labels)Data dictionaryAnnotated CRFSAS transport (XPT)E2B safetyAudit trail
SUBJECGPERF_CODEECGPERF_LABEL
01-001YYes
01-002NDNot done

Before you switch

What biostatistics and clinops teams ask before building on a new platform

Cost

Free sandbox to test a flexible arms/epochs build, no credit card required.

Compliance

Field-level audit trail across every design amendment, not just the original protocol.

Setup time

Amend the visit schedule and randomization list yourself, without a vendor change-order cycle.

Migration

Bring an existing adaptive design's CRFs across; the AI drafts a starting point from your protocol.

Support

Email support, typically under 24 hours, plus documentation on Enterprise.

Build your adaptive or Bayesian-ready study free

Free sandbox with every feature. No credit card, no sales call.

Build your adaptive design free

FAQ

Questions teams ask before they switch

Something not covered here? Ask us directly.

What did FDA's January 2026 Bayesian guidance actually change?

It set out draft expectations for using Bayesian methods in pivotal trials: posterior-probability success criteria, principled prior specification with safeguards, and pre-specified operating characteristics, published January 9, 2026 with comments closing March 13, 2026.

Does Capture run Bayesian statistical analysis?

No. Capture does not perform posterior calculations, define priors, or simulate operating characteristics. That work belongs to your biostatistics team in their own statistical software.

Can Capture support an adaptive trial design?

Yes, on the data-capture and build side: flexible arms and epochs, amendable randomization, and audit-tracked amendments. The statistical design itself is separate, done by your biostatistics team.

Is the Bayesian guidance final?

No, it was a draft as of publication, with its comment period closed March 13, 2026. Confirm current status with FDA before relying on it for a filing decision.

Does an adaptive design need a different EDC than a traditional trial?

Not necessarily. It needs a platform where the schedule, randomization and forms can be amended without a vendor change-order cycle, which Capture supports for any design.

Can I try building a flexible design before committing?

Yes. The free sandbox includes every feature with sample data, no credit card required.

Build your adaptive or Bayesian-ready study free

Free sandbox with every feature. No credit card, no sales call.

Build your adaptive design free