Analysis starts with an export. Stable variable names, coded values with labels, a data dictionary and consistent formats make the difference between a day of programming and a week of cleaning. Capture exports all of them.
Free sandbox · No credit card · 21 CFR Part 11 aligned
| SUBJ | ECGPERF_CODE | ECGPERF_LABEL |
|---|---|---|
| 01-001 | Y | Yes |
| 01-002 | ND | Not done |
What statisticians get
Exports
Every export includes a data dictionary that lists variables, types and code lists, so a SAS or R program can read the files and apply formats without guesswork. Raw files carry coded values; decoded files carry labels, which is useful for data review and listings. Filters by date range and site support interim snapshots and site-level analyses.
Because forms are configured once and locked when approved, variable names are stable across the study. Amendments are deliberate and recorded, so a variable does not silently change meaning halfway through. Instrument items are exported as coded values, and totals and derived scores are calculated in analysis where the rules are written once.
Template forms carry SDTM domain labels and OIDs to make mapping easier. Creating SDTM datasets for a submission remains a programming task: plan the mapping specification early and validate it as you would with any source system.
| SUBJ | ECGPERF_CODE | ECGPERF_LABEL |
|---|---|---|
| 01-001 | Y | Yes |
| 01-002 | ND | Not done |
Blinding
Blinded roles never receive the treatment arm from the server, including in exports. An unblinded administrator role owns the randomisation and medication lists, keeping unblinded work separate from the blinded study team. Plan in advance how any unblinded analyses, for example for a data monitoring committee, will receive their data.
Study coordinator
Subject 01-004
Randomized · R-0012
KitK-10418
Arm columnNot sent
Unblinded administrator
Subject 01-004
Randomized · R-0012
KitK-10418
Arm columnA
Same subject, same screen. The arm never reaches blinded roles.
Enter test data in the free sandbox and export it with the data dictionary.
Outputs
| Output | Contents |
|---|---|
| Clinical data | CSV or Excel, wide format, one row per submission |
| Data dictionary | Variables, types, code lists and labels |
| Decoded files | Labels instead of codes, for review and listings |
| Query list | All queries with status, priority and age, as CSV |
| Safety | Safety listings and E2B export |
| Audit trail | Exportable, filterable by participant and date |
Planning
The cheapest time to fix an analysis problem is before the first participant. Review the forms against the statistical analysis plan: is every endpoint component captured as a coded field, are units fixed, are dates complete, are "not done" and "unknown" coded separately from missing? Statisticians can review the build directly in the sandbox and test exports with sample data.
The sample size calculation and randomisation design also belong in the build: stratification factors must be captured before randomisation, and block sizes set once at setup.
Interim analyses
Interim analyses and data monitoring committee reviews need a defined data cut-off. Agree the cut-off date and the cleaning standard in advance, freeze or lock the data that the snapshot will include, and export with a date-range filter so the dataset matches the cut-off exactly. Document what was included, when, and by whom.
Keep blinded and unblinded outputs separate. The blinded study team works with pooled or blinded data, and any unblinded outputs for the committee are produced by people with the appropriate role. See the database lock software page for freezing and locking data before a snapshot.
Before go-live
Every component captured as a coded field.
Not done, unknown and missing distinguished.
Fixed per field.
Captured before randomisation.
Read into SAS or R with the data dictionary.
Unblinded roles set up only where needed.
Agree a cut-off, freeze or lock the included data, and export with a date-range filter so the dataset matches the cut-off.
Calculated fields such as BMI, BSA and eGFR are shown during entry; other derivations are done in analysis.
Yes. CSV and Excel exports with a data dictionary read directly into SAS, R, Stata or SPSS.
Template forms carry SDTM domain labels and OIDs for mapping. Creating SDTM datasets remains a programming step.
Yes, as raw and decoded files.
Yes, by date range and site.
Yes. Blinded roles never receive the treatment arm, including in exports.
Yes, in the free sandbox with sample data.
Coded data with a data dictionary. Free sandbox.