How to Group Clinical Study Files by Source and Version Every Data Cut with a Checksum

Clinical study files rarely arrive as a neat bundle. An EDC export, an SDTM transfer, an ADaM package, a lab file and a revised protocol can land in the same week, each with its own naming habit. The practical response is not a deeper folder tree; it is a source-first filing rule and a versioning rule that every team member can follow. A clinical research data management software workspace can support that discipline because each study gets its own workspace that brings together CRF data, SDTM and ADaM datasets, define.xml, lab data, protocols, SAPs and TFL outputs. The workflow is deliberately simple enough to use on the next data cut.

Start with five source groups, not one growing inbox

Before uploading any file, agree on the source groups. Keep them stable across the study: EDC, SDTM, ADaM, lab and documents. The names can be plain; consistency matters more than clever abbreviations. Put CRF exports under EDC, tabulation datasets under SDTM, analysis datasets under ADaM, lab transfers under lab, and protocols, SAPs, annotated CRFs and similar study documents under documents. If a file could fit two groups, decide by the system that produced it, not by the person who last touched it. That rule prevents duplicate homes and makes later review predictable.

Then write a one-page filing convention. It should say what each group contains, which file types belong there, and who can move or retire a file. New team members can read it before their first upload. Existing team members can use it to challenge a misfiled item without a long debate. A workspace with tabs for forms, visit schedules, datasets and documents helps because the category is visible in the interface, not hidden in a folder path. It also gives subject filters a consistent place to work across the study.

Treat every data cut as a new version, never as a replacement

A data cut is a moment in the study, not a draft to overwrite. When a new EDC export or SDTM/ADaM delivery arrives, keep the prior cut intact and add the new one as its own revision. The workspace should make this the default: every upload is a new revision with its own checksum, nothing is overwritten, and each person chooses which cut they are looking at. That last detail matters in review meetings. One reviewer may be checking the current cut while another verifies a query against an earlier cut. Both need to see the label clearly.

Use a cut label that a new team member can decode without a key. Include the source, the sequence, and a month-year reference if your study uses dates, for example “SDTM cut 03, November 2026”. Avoid day-level dates in labels; they add precision without helping the reviewer. Record who uploaded the cut, which source it came from, and any known issue. If the platform keeps records rather than deleting them, retired or superseded files remain available for audit, which is exactly what a version history should do.

Use checksums and compare mode to prove what changed

A checksum is a short fingerprint of a file. It lets you confirm that two copies are identical, or that a new upload is genuinely different. In practice, store the checksum with the cut record and check it when a file is downloaded, moved, or shared outside the workspace. If a colleague sends a file with the same name but a different checksum, treat that as a version event, not a minor housekeeping detail. Do not rename it over the old file. Add it as a new revision and note why.

After a new cut is loaded, run a comparison before the team meeting. Filter to the subjects in question, open the dataset, and use compare mode between the new cut and the previous one. Look for changed rows, new records, missing records, and silent differences in variables that should not move. Then capture the result in a note attached to the exact row, field or document page. That keeps the discussion next to the evidence instead of in a separate spreadsheet. If your workspace supports page-by-page document search, use it to check whether a protocol or SAP revision actually supports the change you are seeing.

A simple pre-upload check:

  • Source group matches the file’s origin.
  • Cut label is unique and readable.
  • Checksum is recorded.
  • Prior cut remains available.

Keep documents, notes and access aligned with the same source map

Documents are part of the data chain, not background reading. Put protocols, SAPs and annotated CRFs in the documents group, and keep their versions tied to the cut they describe. When a SAP changes, the new document should sit beside the data cut it governs. Page-by-page search helps reviewers find the relevant section without downloading the whole file. A note or issue attached to a specific page gives the next reviewer a clear starting point.

Before a review, open the documents group and search page by page for the protocol section and the SAP definition that govern the endpoint in question. If the definition changed, attach a note to the exact page and connect it to the cut label you are reviewing. That small step prevents the common failure where reviewers quote an old SAP while looking at a new dataset. It also gives the study lead a defensible reason for why a row changed, beyond a recollection in a meeting.

Access should follow the source map too. Assign viewers, collaborators and owners by study, and keep unblinded content behind its own grant. If someone does not have that grant, the unblinded files, records and search results should stay hidden regardless of role. Review the append-only audit trail for membership, role, unblinding and record changes before any major review. The aim is not a perfect folder taxonomy; it is a reliable chain from source to cut to review, with a checksum and a note at every handoff. That chain is what makes a clinical data management workspace useful in daily study work, not just during setup.