Contents
Academy · Concepts & Definitions · Before database lock
Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed
Database lock is the point of no return for a trial's analysis plan. Four specific items need to be nailed down and written before that point — not understood informally, written down.
Before a clinical trial's database is locked, four items specifically need to be finalised in writing: the statistical analysis plan itself, the trial's estimands (what exactly is being estimated for each outcome), derived-variable definitions, and the blinding arrangements for the analysis — each of which becomes much harder to change credibly once the data are unblinded.
What you will learn
- Why these four items specifically need to be written down before lock, not just understood by the team.
- What an 'estimand' is, and why defining it matters before, not after, seeing the data.
- Why derived-variable definitions are a common, avoidable source of late-stage disagreement.
What to unlearn
- “The statistical analysis plan can be finalised informally, through team discussion rather than a written document.” A written, dated, finalised statistical analysis plan — ideally itself version-controlled — is what lets anyone later verify that the pre-specified analysis, not a post-hoc choice, was actually followed.
- “An 'estimand' is just another word for the outcome being measured.” An estimand specifies more than the outcome — it specifies the population, the variable, the handling of intercurrent events (such as treatment discontinuation or rescue medication), and the summary measure, per the ICH E9(R1) addendum's estimand framework, and these specifics need to be agreed before unblinding, not inferred afterward.
1. The statistical analysis plan
A finalised, written statistical analysis plan — specifying the planned analyses for every outcome, including any planned subgroup or sensitivity analyses — should be signed off and dated before database lock, so that any deviation from it during the actual analysis can be identified and disclosed as a deviation, rather than silently blending into the primary analysis.
2. Estimands
Per the ICH E9(R1) addendum's estimand framework, each outcome's estimand specifies the population, the variable, how intercurrent events (such as treatment discontinuation, rescue medication use, or death before the assessment) will be handled analytically, and the summary measure used to compare groups — and these specifics should be agreed before unblinding, since the choice of intercurrent-event strategy can materially change the resulting estimate.
3. Derived-variable definitions
Any variable derived from raw collected data (such as a composite endpoint, a change-from-baseline score, or a categorised version of a continuous measure) needs its exact derivation rule specified and finalised before lock — a common and avoidable source of late-stage disagreement is discovering, after unblinding, that two team members assumed different derivation rules for the same named variable.
4. Blinding arrangements
The specific arrangements for how blinding will be maintained during the analysis itself (who will see treatment-group labels, and when) should be agreed and documented before lock, since confirming this only once the analysis is already underway risks an inadvertent unblinding that can't be fully undone.
Take-home message
Take-home message
The statistical analysis plan, the trial's estimands, derived-variable definitions, and the analysis blinding arrangements all need to be finalised in writing before database lock — each becomes far harder to settle credibly, or to change without raising bias concerns, once the data are unblinded.
How to apply this to your work
How to apply this to your own work
- Finalise and date the statistical analysis plan as a standalone, version-controlled document. Before lock, not during the analysis itself.
- Write out each outcome's estimand explicitly, including its intercurrent-event strategy. Per the ICH E9(R1) estimand framework, rather than leaving this implicit.
- Document every derived variable's exact derivation rule. Confirm the whole analysis team agrees on the same rule before lock.
- Document and agree the specific blinding arrangements for the analysis itself. Before the analysis is underway, not reactively once a blinding question comes up.
Common questions
The statistical analysis plan details the planned analyses. The clinical study report presents the statistical methods, results and relevant deviations from the planned analyses.
A genuine, well-justified correction can sometimes be made and transparently disclosed as a deviation, but changing an estimand after unblinding specifically to achieve a more favourable result is not acceptable practice.
Typically the biostatistician, the clinical team, and anyone responsible for the data management plan, since derivation rules often depend on how the raw data were actually collected.
References and sources
- ICH Harmonised Guideline. Addendum to ICH E9: Statistical Principles for Clinical Trials — Estimands and Sensitivity Analysis in Clinical Trials (E9(R1)). read 10 October 2026. View ICH E9(R1) guideline
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 10). Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed. Pubrica. https://pubrica.com/insights/checklist/before-database-lock-checklist/
Pubrica Academy. “Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed.” Pubrica, 10 Oct. 2026, pubrica.com/insights/checklist/before-database-lock-checklist/.
Pubrica Academy (2026) Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed. Pubrica. Available at: https://pubrica.com/insights/checklist/before-database-lock-checklist/ (Accessed: 15 October 2026).
Pubrica Academy. “Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed.” Pubrica. October 10, 2026. https://pubrica.com/insights/checklist/before-database-lock-checklist/
Pubrica Academy, ‘Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed’ (Pubrica, 10 October 2026) <https://pubrica.com/insights/checklist/before-database-lock-checklist/> accessed 15 October 2026.
Pubrica Academy. Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/insights/checklist/before-database-lock-checklist/
Pubrica Academy, “Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/insights/checklist/before-database-lock-checklist/
@misc{before_database_lock_checklist_2026,
author = {{Pubrica Academy}},
title = {Before Database Lock: Analysis Plan, Estimands, Derivations, and Blinding — Finalised, Not Assumed},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/insights/checklist/before-database-lock-checklist/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Pre-lock finalisation check
Confirm all four before database lock.
[ ] Statistical analysis plan finalised, dated, version-controlled [ ] Each outcome's estimand specified (population, variable, intercurrent-event strategy, summary measure) [ ] Every derived variable's derivation rule documented and agreed by the full analysis team [ ] Analysis blinding arrangements documented and agreed
Pubrica AcademyWritten and maintained by the subject matter experts of Pubrica who do this work. Articles are revised when the standard underneath them moves, and each one carries the date it was last checked.
How Pubrica helps
We help trial teams finalise and document the statistical analysis plan, estimands, derived-variable definitions, and blinding arrangements before database lock.