PLANREFERENCE OUTLINE

Statistical Analysis Plan

Analysis plan built on the estimand rather than the method: treatment, population, variable, summary measure, and a strategy chosen for every enumerated intercurrent event; analysis populations defined so membership needs no judgement at analysis time; sensitivity analyses each naming the assumption they probe; and a version table recording whether each change preceded unblinding. Maps to ICH E9(R1).

What a template is not

A template is a document baseline to adapt inside your own quality system. SPEQ does not approve, validate, or take responsibility for what you issue from it, and using one is not evidence of compliance.

CHECKING ACCESS

Checking your Professional access…

REGULATIONS MAPPED
ICH E9(R1)
DOCUMENT TYPE
Plan
LAST UPDATED
August 2026
PURPOSE

The analysis plan removes discretion at the point where discretion is most dangerous — after the data exist. Its centre of gravity is the estimand: a precise statement of the treatment effect to be estimated, including what happens when participants stop treatment, take rescue medication, or die. Choosing a strategy for each intercurrent event is a clinical question the sponsor owns, not a statistical convenience.

What's Inside

✓Plan identification with the date finalised and its stated relation to unblinding
✓The primary estimand set out attribute by attribute, so the question being answered is unambiguous
✓Intercurrent events enumerated, each with a chosen strategy and the rationale for choosing it
✓Analysis populations defined so membership can be decided without judgement at analysis time
✓The primary analysis method, stated as following from the estimand rather than preceding it
✓Sensitivity analyses, each naming the assumption it probes, alongside the multiplicity strategy
✓A version table recording, for each change, whether it preceded unblinding and why it was made

How to Use It

1Define the estimand before choosing a method — a method without an estimand describes a calculation rather than a question anyone asked.
2Enumerate intercurrent events and choose a strategy for each; leaving them unaddressed delegates the decision to whatever the software does with missing values.
3Define analysis populations so membership needs no judgement at analysis time, because judgement exercised after unblinding is indefensible.
4Name the assumption each sensitivity analysis probes, so it cannot later be read as an alternative result to select among.
5Finalise and date the plan before unblinding, and version any later change with its reason and its timing relative to unblinding.
6Have the clinical team agree the intercurrent event strategies, since these are clinical questions being answered inside a statistical document.
DOCUMENT CONTENTS

The full section structure of this template — every section and sub-section, so you can use it as a baseline for your own site document.

—Document Control
› Document Information› Approval Signatures› Revision History› Distribution List
1Plan Identification and Timing
2Estimands
3Intercurrent Events
4Analysis Populations
5Primary Analysis, Sensitivity and Multiplicity
6Version Control and Approval
REGULATORY CONTEXT

ICH E9(R1) introduces the estimand framework and the treatment of intercurrent events, extending the established statistical principles for clinical trials and setting the expectation that the analysis is pre-specified. It does not choose your estimand, your intercurrent event strategies, or which sensitivity analyses are informative for your trial — those follow from the clinical question and belong to the sponsor’s clinical and statistical teams jointly. The plan structure and guidance are SPEQ practitioner synthesis.

MAPPED STANDARDS
ICH E9(R1)
Browse the standards catalog →

Where to next

All 112 SPEQ templates →Filter the library by document type, product area, and format.