Platform data populations and control pools

Track epochs and comparator eligibility

Concurrent controls must be eligible for the relevant comparison during its active epoch. Nonconcurrent borrowing requires separate assumptions and sensitivity analyses; it is not equivalent to randomized concurrent evidence.

Flow diagram

Steps

Version participant eligibility at entry

Retain cohort, active arms, allocation rule and epoch.

Define concurrent controls

Define controls eligible for each arm during its active recruitment epochs, preserving eligibility and calendar dates. Shared controls are not independent observations across comparisons.

Define any nonconcurrent borrowing separately

Specify any nonconcurrent-control borrowing separately, with calendar-trend, comparability and model assumptions plus sensitivity analyses.

Construct comparison-specific datasets

Preserve shared-control membership and dependence without double-counting participants.

Freeze populations and provenance

Record amendments, data cuts and original versus borrowed control membership.