The Statistical Analysis Plan (SAP) is a detailed document that precisely describes all statistical methods, analysis populations, endpoint definitions, and evaluation strategies of a clinical study. It is developed on the basis of the study protocol but is considerably more detailed and contains all technical specifications required for the analysis of the study data. The SAP must be finalized and dated before database lock and unblinding. It is a binding document of the Trial Master File (TMF) and is subject to the essential document requirements in accordance with ICH E6(R3). Only a pre-specified, finalized SAP protects the analysis from the accusation of outcome-driven analysis decisions.
Contents and Structure of the SAP
A complete SAP contains all relevant statistical aspects of the study. These include: the precise definition of all primary, secondary, and exploratory endpoints, including measurement time points and measurement instruments; the description of the analysis populations (intent-to-treat, per-protocol, safety population); the statistical method for the primary analysis with all model parameters; and the strategy for handling missing data.
Further mandatory contents are: the pre-specified sensitivity analyses, subgroup analyses, and exploratory analyses; the description of multiplicity control (e.g., hierarchical testing procedure, Bonferroni correction); a detailed list of tables and figures (shell tables); and the definition of outliers and protocol violators. The SAP is a living document that can be updated via amendment – as long as the changes are made before unblinding.
Timeline and Regulatory Requirements
The timeline is clearly regulated: The SAP is based on the final study protocol and is usually created and finalized during study conduct. The EMA and FDA expect the SAP to be available and dated before database lock. Changes after database lock are considered post-hoc and are critically assessed by authorities. An exception is minor amendments that are purely editorial in nature.
According to ICH E9(R1), the SAP must contain an explicit description of the estimand-based analysis strategy. The estimand defines which target population, which intervention, which variable, and which handling of intercurrent events are relevant for the primary analysis. This requirement has significantly refined SAP development: In addition to the statistical method, the clinical interpretation of the analysis result must now also be clearly communicated. Typical intercurrent events are study discontinuation, initiation of rescue therapy, or death before the endpoint measurement time point – for each, the SAP must specify how it is handled in the analysis (e.g., composite strategy, hypothetical strategy, while-on-treatment).
SAP and Clinical Study Report
The SAP forms the analytical basis for the Clinical Study Report (CSR) according to ICH E3. In the CSR, all analyses described in the SAP are performed and their results are fully reported. Deviations between the SAP and the analyses actually performed must be explicitly explained and justified in the CSR. GCP inspectors check the consistency between protocol, SAP, and CSR as a core component of a GCP inspection.
For regulatory submission (e.g., MAA to the EMA), the SAP is a mandatory component of the eCTD dossier in Module 5 (clinical study reports). Together with the study protocol and the CSR, it forms the three-part core document for the assessment of the clinical efficacy and safety of a medicinal product by the competent authority.
Relevance for clinical trials
A well-written SAP protects the scientific integrity of the study. It prevents outcome-driven analysis decisions (data dredging), ensures the reproducibility of the evaluation, and enables independent verification of the results by authorities and external reviewers. For sponsors, CROs, and biostatisticians, the SAP is the central working document for the entire analysis phase of a study.
The development of the SAP requires close collaboration between biostatisticians, clinical scientists, data managers, and regulatory experts. Full-service CROs such as mediconomics take on the complete SAP development on behalf of the sponsor and ensure that all regulatory requirements – in particular the estimand terminology according to ICH E9(R1) – are correctly implemented. A formal SAP review by an independent second biostatistician is best practice and is anchored in the SOPs of many sponsors and CROs as a quality assurance step. Particular attention is paid to the consistency between the endpoint definitions in the protocol, the SAP, and the wording later used in the CSR.
Frequently Asked Questions (FAQ)
What is the difference between the study protocol and the SAP?
The study protocol describes the design, objectives, inclusion and exclusion criteria, treatment, and endpoints of a study at a conceptual level. The SAP translates these specifications into a technically precise statistical language: It contains the exact model specification, covariates, missing data strategy, multiplicity control, and shell tables. The protocol is the medical-scientific basis, the SAP the statistical blueprint for the analysis.
Can the SAP be changed after database lock?
Generally not. Changes after database lock are considered post-hoc and can jeopardize the credibility of the analysis. In exceptional cases, purely administrative corrections are possible, which must be documented in a traceable manner. Authorities critically assess any deviation between the planned and actual SAP. Adherence to the timeline – SAP finalized before database lock – is a key point in GCP inspections.
Who is responsible for the development of the SAP?
The primary responsibility lies with the biostatistician of the sponsor or the CRO. The SAP is developed in close coordination with the clinical lead, the study physician, and, for regulatorily sensitive endpoints, also with regulatory affairs experts. Before finalization, the SAP usually undergoes a formal internal review process, which is defined in the QM SOPs of the sponsor or the CRO.