Buy the GCP R3 course & get a FREE eBook— your complete ICH-GCP R3 reference guide. Book Now →

  • AI ClinicalAI ClinicalNewRight Arrow
  • Clinical PracticeClinical PracticeRight Arrow
  • PharmacovigilancePharmacovigilanceRight Arrow
  • ManufacturingManufacturingRight Arrow
  • Laboratory PracticeLaboratory PracticeRight Arrow
  • DistributionDistributionRight Arrow
  • Medical DevicesMedical DevicesRight Arrow
  • Regulatory AffairsRegulatory AffairsRight Arrow
  • Research ComplianceResearch ComplianceRight Arrow
  • Health & SafetyHealth & SafetyRight Arrow
  • Quality AssuranceQuality AssuranceRight Arrow
  • BiostatisticsBiostatisticsTrendingRight Arrow
  • Pharmaceutical MedicinePharmaceutical MedicineNewRight Arrow
  • All Bundles
  • Clinical Trials Development
  • GCP & GxP Standards
  • GMP & Quality Assurance
  • Pharmacovigilance
  • Role Based
  • Distribution
  • Medical Devices
  • Medical Writing
  • AI Clinical
  • Quality Assurance
  • Biostatistics
  • Pharmaceutical Medicine
  • Customize your bundleBuild your own path
View all 30 learning paths
  • Preclinical & Laboratory Foundations Learning Path
  • Phase I – First-in-Human Trials Learning Path
  • Phase II & III – Efficacy & Pivotal Trials Learning Path
  • Clinical Trials Foundation PathNew
  • Regulatory Submission & Approval
Preclinical & Laboratory Foundations Learning Path - courses at £199

About

A Statistical Analysis Plan (SAP) provides the detailed framework for analysing clinical trial data and translates the study objectives, endpoints, estimands, and statistical principles into prespecified analytical methods. A well-developed SAP supports scientific integrity, reduces analytical bias, promotes transparency, and ensures that statistical analyses are conducted consistently and in accordance with regulatory and protocol requirements.
This Biostatistics – Statistical Analysis Plans Training Course & Certification provides comprehensive knowledge of SAP structure and development, analysis populations, endpoints, estimands, hypothesis testing, statistical methods, missing data, intercurrent events, multiplicity, interim analyses, sensitivity analyses, subgroup analyses, protocol deviations, statistical programming alignment, SAP review and approval, version control, and regulatory expectations. Upon successful completion, learners receive a certification demonstrating their understanding of Statistical Analysis Plan development and best practices for clinical research.

Who Should Enrol?

  • Biostatisticians and Statistical Programmers
  • Clinical Research and Clinical Trial Professionals
  • Data Management and Clinical Data Scientists
  • Regulatory Affairs and Biostatistics Professionals
  • Statistical Analysis and Reporting Teams
  • Clinical Trial Managers and Study Team Members
  • Quality Assurance and Clinical Research Compliance Professionals
  • Anyone involved in developing, reviewing, implementing, or approving Statistical Analysis Plans for clinical studies
📢 Every purchase also includes our FREE companion Statistical Analysis Plans eBook, designed to help you apply principles in real-world clinical research settings.

What you will learn

Understand the purpose, structure, and principles of Statistical Analysis Plans (SAPs) and their role in ensuring consistent, transparent, and scientifically sound clinical trial analyses.

Learn how to define analysis populations, endpoints, estimands, hypotheses, treatment comparisons, and statistical methods in accordance with the clinical trial objectives and protocol.

Develop knowledge of missing data strategies, intercurrent events, multiplicity, interim analyses, sensitivity analyses, subgroup analyses, and handling of protocol deviations within an SAP.

Gain practical understanding of SAP development, review, version control, statistical programming alignment, regulatory expectations, and maintaining traceability throughout the SAP lifecycle.

Course Syllabus

  1. A. What a SAP is — and is not
  2. Defining the Statistical Analysis Plan
  3. What the SAP is NOT
  4. What the SAP adds beyond the protocol
  5. Who reads a SAP, and why the audience matters
  6. The SAP's scope, ring by ring
  7. Does every trial need a full SAP?
  8. Common SAP myths — true or false
  9. The cost of skipping a proper SAP
  10. B. Why pre-specification matters
  11. The problem: data-driven analysis
  12. Protecting the Type I error rate
  13. Credibility, reproducibility and regulatory trust
  14. Where multiplicity risk hides
  15. Trial registries reinforce pre-specification
  16. Bias can be unconscious, not just malicious
  17. Pre-specified vs post-hoc analysis, side by side
  18. Worked example — spot the data-driven red flag
  19. C. ICH E9 and the regulatory ecosystem
  20. ICH E9: the foundation
  21. ICH E9(R1): the estimands addendum
  22. ICH E3: the Clinical Study Report
  23. ICH E6 GCP and ICH E10
  24. The regulatory ecosystem at a glance
  25. FDA and EMA expectations
  26. Milestones in SAP-related guidance
  27. Where the guidance is silent: sponsor SOPs govern
  28. D. SAP vs protocol vs CSR
  29. Three documents, three jobs
  30. Side-by-side: protocol vs SAP vs CSR
  31. Why the boundaries matter
  32. A decision tool for borderline content
  33. Signs a document boundary has been violated
  34. SENTINEL-3's document map at a glance
  35. Why redundancy is still risky, even when consistent
  36. Resolving a document-boundary dispute
  37. E. Documents the SAP connects to
  38. The SAP's document neighbourhood
  39. SAP and the protocol: a two-way relationship
  40. SAP and the Data Management Plan (DMP)
  41. SAP and the CRF
  42. SAP and the TLF shells
  43. SAP and define.xml
  44. SAP document connections at a glance
  45. A pre-finalisation consistency checklist
  46. Worked example — classify SENTINEL-3 decisions by document
  47. F. Ownership: who writes, reviews and signs
  48. The lead trial statistician authors the SAP
  49. Who reviews the SAP
  50. Who signs the SAP
  51. Sign-off process flow
  52. Author, reviewer, signatory: three distinct roles
  53. When reviewer comments conflict
  54. A simple RACI for SAP authorship
  55. The iterative review cycle
  56. G. Timing and the golden rule
  57. The SAP lifecycle timeline
  58. Blinded data reviews
  59. THE GOLDEN RULE
  60. Why unblinding is the hard line
  61. SAP lifecycle milestones and typical lead times
  62. Firewalling the unblinded interim statistician
  63. Is this content still safe to add?
  64. Worked example — is this a golden-rule violation?
  65. H. Consequences of a late or amended-after-unblinding SAP
  66. Regulatory consequences
  67. Scientific and credibility consequences
  68. Why prevention, not remediation, is the strategy
  69. Consequences at a glance
  70. A cautionary pattern, generically described
  71. Where timing failures are usually first spotted
  72. Timing safeguards every trial should have

  1. A. Why a Standard Structure Matters
  2. Why SAPs converge on a common structure
  3. Where the standard section set comes from
  4. The reader's journey through a SAP
  5. The full section map, part 1: Sections 1-7
  6. The full section map, part 2: Sections 8-14
  7. A thorough section versus a thin section
  8. Level of detail: the Goldilocks problem
  9. SAP structure by the numbers
  10. Section numbering varies by sponsor — the content usually doesn't
  11. B. Section 1 — Administrative Front Matter
  12. What belongs in Section 1
  13. Title page anatomy
  14. Approvals and signatures: who signs, and why
  15. The version history / change log
  16. The abbreviations list
  17. Drafting, review and sign-off: the administrative workflow
  18. Administrative section: length and tone
  19. Common administrative-section mistakes
  20. Section 1 as an inspection entry point
  21. C. Sections 2-5 — Background, Objectives, Design, Sample Size
  22. Section 2: Introduction / Background
  23. Section 3: Study Objectives & Endpoints — a preview
  24. Section 4: Study Design Summary
  25. Section 5: Sample-Size Justification
  26. Why Sections 2-5 restate rather than redesign
  27. Decision guide: which section does this belong in?
  28. SENTINEL-3's Sections 2-5 at a glance
  29. Design factors reappear as methods covariates
  30. Wording differences versus content differences
  31. D. Sections 6-7 — Populations & General Analysis Conventions
  32. Section 6: Analysis Populations — a preview
  33. Section 7: General Conventions — visit windows
  34. Section 7: derived variables and baseline
  35. Section 7: date imputation rules
  36. Section 7: coding dictionaries and software
  37. Cross-referencing discipline inside Section 7
  38. SENTINEL-3's Section 7 at a glance
  39. Checklist: what a good general-conventions section contains
  40. Populations and conventions in the blinded data review
  41. E. Sections 8-11 — Methods, Missing Data, Interims, Safety
  42. Section 8: Statistical Methods — a preview
  43. Section 9: Missing Data & Intercurrent Events — a preview
  44. Section 10: Interim Analyses & Multiplicity — a preview
  45. Section 11: Safety Analyses
  46. Safety analyses: the table inventory
  47. Why the safety section is often underspecified
  48. The level-of-detail cascade, from objective to program
  49. The MedDRA connection between Section 7 and Section 11
  50. How Sections 8, 9 and 10 depend on each other
  51. F. Sections 12-13 — Changes from Protocol & TLF Shells
  52. Section 12: Changes from the Protocol-Planned Analysis
  53. Section 13: TLF Shells — what a 'shell' is
  54. The shell-first principle: why build shells before finalising methods
  55. Anatomy of a shell
  56. A SAP with shells attached versus one without
  57. SENTINEL-3's shell inventory
  58. Numbering shells to match the CSR
  59. Common shell mistakes
  60. Shells aren't only tables: listings and figures too
  61. G. Section 14 — Appendices, Cross-Referencing & Hands-On
  62. Section 14: Appendices
  63. Appendices feeding the SAP
  64. Cross-referencing discipline, revisited
  65. Level of detail, revisited: a preview of Module 7
  66. Diagnosing SAP structural readiness
  67. Final structure checklist: is the SAP complete?
  68. The SAP as a living reference during the trial
  69. The section-set's lifecycle across a trial
  70. Worked example — map SENTINEL-3 content to the correct sections
  71. Worked example — sketch a demographics ('Table 1') shell

  1. A. Endpoint, Variable and Estimand: The Vocabulary
  2. What is an endpoint?
  3. Endpoint vs variable: what's the difference?
  4. Endpoint vs estimand: a preview
  5. Why this precision matters: pre-specification
  6. Where this lives in the SAP
  7. SENTINEL-3's endpoint set at a glance
  8. Terminology pitfalls: endpoint vs outcome vs measure
  9. B. Endpoint Roles: Primary, Secondary, Exploratory
  10. The primary endpoint: one, pre-specified, decisive
  11. Key secondary endpoints
  12. Other secondary endpoints
  13. Exploratory endpoints
  14. Confirmatory vs exploratory: the practical difference
  15. SENTINEL-3's endpoint roles mapped
  16. C. Building the Primary Endpoint: Composite MACE
  17. Why composite endpoints in CV outcomes trials?
  18. Anatomy of SENTINEL-3's 3-point MACE
  19. Choosing components: the selection criteria
  20. The risk: one component can dominate the result
  21. Endpoint adjudication: the Clinical Events Committee
  22. Worked example — write a precise primary endpoint definition
  23. Expanding the composite: 3-point vs 4-point MACE
  24. Composite endpoint pitfalls — a checklist
  25. What the adjudication charter actually contains
  26. D. Endpoint Definition Precision
  27. Why 'time to first MACE' isn't precise enough
  28. Defining the event: diagnostic criteria
  29. Defining the timing: index date and event date
  30. Defining censoring precisely
  31. Defining the assessment schedule
  32. SENTINEL-3's fully precise primary endpoint definition
  33. Handling simultaneous or same-day events
  34. Precision and the estimand's variable attribute
  35. A precision checklist for any endpoint definition
  36. E. The Estimand as the Bridge
  37. Why endpoint precision still isn't the whole answer
  38. Introducing the estimand concept
  39. SENTINEL-3's primary estimand, at a glance
  40. Estimand vs endpoint: the SENTINEL-3 contrast
  41. How the estimand shapes the rest of the SAP
  42. The population-level summary attribute, previewed
  43. F. Hypotheses and the Testing Framework
  44. Null and alternative hypothesis basics
  45. Superiority hypotheses
  46. Non-inferiority hypotheses and the margin
  47. Equivalence hypotheses
  48. One-sided vs two-sided testing
  49. Setting the significance level
  50. Families of hypotheses in a SAP
  51. Writing testable hypotheses in the SAP
  52. G. Multiplicity: The Secondary Hierarchy
  53. Why testing multiple endpoints inflates Type I error
  54. Fixed-sequence gatekeeping: the concept
  55. SENTINEL-3's key secondary hierarchy, in order
  56. Worked example — order the secondary hierarchy
  57. What happens when the hierarchy stops
  58. Hierarchy vs alpha-splitting: a preview
  59. Choosing the hierarchy order: the underlying logic
  60. Documenting the hierarchy in the SAP
  61. H. Effect Measures and Model Choice
  62. Why the endpoint dictates the model
  63. Hazard ratio: for time-to-event endpoints
  64. Relative risk and odds ratio: for binary endpoints
  65. Mean difference: for continuous endpoints
  66. SENTINEL-3's endpoint-to-model map
  67. Reporting effect measures: estimate, CI, and p-value together
  68. Confidence intervals and clinical relevance

  1. A. Why Analysis Populations Matter
  2. Why 'the population' changes the answer
  3. Populations as an analysis convention
  4. The population menu, previewed
  5. The estimand connection, previewed
  6. SENTINEL-3's population overview
  7. Getting populations wrong: the consequences
  8. Who defines populations, and when
  9. Terminology: population vs analysis set vs dataset
  10. B. The Randomised Set and the ITT Principle
  11. Definition of the randomised (ITT) set
  12. The ITT principle: analyse as randomised
  13. Why ITT protects randomisation's bias protection
  14. ITT and the treatment-policy estimand
  15. What ITT does NOT exclude
  16. ITT as the default primary population for efficacy
  17. SENTINEL-3's ITT set definition, worked
  18. ITT's limits: what it can't fix
  19. C. Modified ITT (mITT): Variants and Risks
  20. What is mITT, and why do sponsors use it?
  21. Common mITT variants
  22. Risk: mITT can break randomisation balance
  23. Risk: differential dropout between arms
  24. When is mITT defensible?
  25. mITT vs ITT: the practical comparison
  26. Writing the mITT definition precisely in the SAP
  27. mITT and regulatory scrutiny
  28. D. The Per-Protocol (PP) Set
  29. What is the per-protocol set?
  30. PP exclusion criteria
  31. PP and non-inferiority trials
  32. Risks of relying on the per-protocol population
  33. PP as supportive, rarely primary, for superiority
  34. SENTINEL-3's per-protocol set definition
  35. ITT vs PP: a decision guide
  36. Reporting PP population size and attrition
  37. E. Safety, PK and Other Populations
  38. The safety population: the as-treated principle
  39. Why safety uses as-treated, not as-randomised
  40. Wrong-treatment-received subjects in safety analyses
  41. The pharmacokinetic (PK) population
  42. Evaluable populations for specific endpoints
  43. Population-endpoint matrix
  44. Choosing which population is primary, analysis by analysis
  45. F. Protocol Deviations: Defining and Classifying
  46. What is a protocol deviation?
  47. Deviation vs violation: a terminology note
  48. Important vs minor deviations
  49. Examples of important deviations for SENTINEL-3
  50. Examples of minor deviations for SENTINEL-3
  51. Who classifies deviations, and how
  52. The deviation log and tracking through the trial
  53. Worked example — classify SENTINEL-3 deviations
  54. G. The Blinded Review and Population Assignment
  55. What is the blinded data review?
  56. Timing: before unblinding, before database lock
  57. Who attends the blinded review
  58. The population-assignment output
  59. Randomisation errors: how they're handled
  60. Wrong-treatment-received: a decision guide
  61. Eligibility violations discovered post-randomisation
  62. Worked example — allocate SENTINEL-3 edge cases to populations
  63. H. Documenting Exclusions: CONSORT and Transparency
  64. Why transparent exclusion reporting matters
  65. The CONSORT flow diagram concept
  66. What the SAP must pre-specify for population reporting
  67. Building the population attrition table
  68. Reporting subject counts and percentages by reason
  69. Consistency checks across the SAP
  70. Population definitions and the amendment process

  1. A. The Addendum and Why Estimands Were Introduced
  2. ICH E9 (1998): a strong foundation with a gap
  3. The old failure mode: method chosen, question left implicit
  4. ICH E9(R1) (2019): the estimands addendum
  5. Estimand vs endpoint: closing the loop from Module 3
  6. Why this is worth the technical investment
  7. Section A recap: from ambiguity to a five-part specification
  8. B. The Five Estimand Attributes
  9. Attribute 1: treatment condition
  10. Attribute 2: population
  11. Attribute 3: variable (endpoint)
  12. Attribute 4: intercurrent-event handling
  13. Attribute 5: population-level summary
  14. The five attributes at a glance
  15. SENTINEL-3's primary estimand, fully assembled
  16. A common deficiency: attributes stated, coherence missing
  17. C. Intercurrent Events vs Missing Data
  18. Defining an intercurrent event
  19. Defining missing data
  20. IE vs missing data, side by side
  21. Case 1: an IE with no missing data
  22. Case 2: missing data with no intercurrent event
  23. SENTINEL-3's intercurrent-event catalogue
  24. Why the distinction matters in the SAP
  25. D. The Five Intercurrent-Event Strategies
  26. The five strategies: an overview
  27. Strategy 1: treatment policy
  28. Strategy 2: hypothetical
  29. Strategy 3: composite variable
  30. Strategy 4: while on treatment
  31. Strategy 5: principal stratum
  32. The five strategies: full comparison table
  33. Choosing a strategy: decision logic
  34. One trial, several strategies at once
  35. Pitfall: confusing 'ITT' with 'treatment policy'
  36. SENTINEL-3's IE strategies across the trial timeline
  37. E. Missing-Data Taxonomy: MCAR, MAR, MNAR
  38. Why the mechanism matters, before the definitions
  39. MCAR: missing completely at random
  40. MAR: missing at random
  41. MNAR: missing not at random
  42. The taxonomy as a 2x2 framing
  43. Applying the taxonomy to SENTINEL-3's two missing-data causes
  44. Section E to F: mechanism drives estimator choice
  45. F. Estimand-Then-Estimator: MMRM, MI and Mixed Models
  46. Estimand-then-estimator logic, stated plainly
  47. MMRM in plain English
  48. MMRM: the model, in notation
  49. MMRM's assumption: why it needs MAR
  50. Multiple imputation (MI) in plain English
  51. MI: the steps, in notation
  52. Matching estimand attributes to estimator choice
  53. SENTINEL-3's estimand-to-estimator map
  54. G. Sensitivity vs Supplementary Analyses; Avoiding LOCF/CC
  55. Sensitivity vs supplementary: the core distinction
  56. Why every primary analysis needs sensitivity analyses
  57. Reference-based (jump-to-reference) multiple imputation
  58. Tipping-point analysis
  59. Tipping-point logic, in notation
  60. Why naive LOCF is discouraged
  61. Why complete-case analysis is discouraged
  62. LOCF/complete-case vs MMRM/MI, side by side
  63. SENTINEL-3's full sensitivity-analysis plan
  64. H. Writing It Into the SAP
  65. Where this content lives in the SAP
  66. What SAP Section 9 must contain
  67. Writing IE-strategy text precisely
  68. Documenting software, procedures and seeds
  69. Regulatory expectations: FDA and EMA missing-data guidance
  70. The full pipeline, recapped end to end
  71. Worked example — specify SENTINEL-3's primary estimand (part 1 of 2)
  72. Worked example — choose IE strategies for two new scenarios (part 2 of 2)

  1. A. Why Interims, and the Type I Error Risk
  2. What is an interim analysis?
  3. Three reasons to look early: ethics, futility, efficiency
  4. SENTINEL-3's interim schedule at a glance
  5. The core risk: multiple looks inflate Type I error
  6. Why 0.05 at each look doesn't work — the arithmetic
  7. Peeking without correction: a cautionary pattern
  8. Interim monitoring is a repeating cycle, not a one-off event
  9. Ethics, futility, efficiency — how they show up in SENTINEL-3
  10. Why pre-specification protects the interim from bias
  11. B. Group-Sequential Design & Alpha Spending
  12. Group-sequential design: the statistical fix
  13. Information fraction: the clock that matters
  14. Why SENTINEL-3 is event-driven, not time-driven
  15. SENTINEL-3's information fraction at each look
  16. Alpha-spending functions: spending the error budget over time
  17. Pocock vs O'Brien-Fleming: two spending philosophies
  18. Lan-DeMets: flexible spending without fixing the calendar
  19. SENTINEL-3's interim schedule
  20. Choosing OBF for a cardiovascular outcomes trial
  21. C. Efficacy & Futility Boundaries; Conditional Power
  22. The SENTINEL-3 efficacy boundary, drawn out
  23. Reading the boundary: what crossing it actually means
  24. Futility boundaries: stopping when success is very unlikely
  25. Conditional power: the reasoning behind a futility call
  26. Binding vs non-binding futility boundaries
  27. Why SENTINEL-3 uses a non-binding futility rule
  28. DSMB decision at a look: a 2x2 view
  29. What crossing an efficacy boundary triggers
  30. Interim results stay confidential until the DSMB acts
  31. D. The DSMB/DMC: Charter, Independence, Firewall
  32. What a DSMB/DMC is, and why it is independent
  33. The DSMB charter: the rulebook for oversight
  34. Open sessions vs closed sessions
  35. The unblinded statistician: the firewall in practice
  36. The firewall, end to end
  37. What the DSMB reviews at a closed session
  38. DSMB independence: what can go wrong without it
  39. Documenting DSMB decisions in the SAP and charter
  40. DSMB independence, membership and firewall — quick recap
  41. E. The Multiplicity Problem; FWER vs FDR
  42. Multiplicity: more tests, more chances for a false positive
  43. Where multiplicity comes from in SENTINEL-3
  44. Family-wise error rate (FWER): controlling ANY false positive
  45. False discovery rate (FDR): a different bargain
  46. FWER vs FDR: which one confirmatory trials use, and why
  47. The alpha 'budget': one fixed resource across the whole design
  48. What happens without multiplicity control: an illustration
  49. Multiplicity is not just endpoints — subgroups too
  50. A quick multiplicity self-check
  51. F. Adjustment Methods
  52. The adjustment toolbox: from simple to sophisticated
  53. Bonferroni: the simple, conservative baseline
  54. Holm's step-down procedure: less conservative than Bonferroni
  55. Hochberg's step-up procedure
  56. Fixed-sequence / hierarchical (gatekeeping) testing
  57. Why fixed-sequence testing needs no alpha split
  58. Graphical (Bretz-Maurer) approaches: flexible alpha propagation
  59. Splitting and recycling alpha across a testing graph
  60. Choosing an adjustment method for your endpoint family
  61. Adjustment methods at a glance
  62. Recycling alpha: what strict fixed-sequence testing gives up
  63. G. Hands-On: SENTINEL-3's Alpha Allocation
  64. Assembling SENTINEL-3's full alpha budget
  65. The primary endpoint's alpha across the three looks
  66. Only the primary spends interim alpha — why the secondaries wait
  67. SENTINEL-3's secondary hierarchy: fixed order, one gate at a time
  68. HANDS-ON — Build SENTINEL-3's alpha-allocation and testing sequence
  69. Reading the finished alpha map
  70. Common pitfalls in an alpha-allocation exercise like this one

  1. A. Writing the Methods Section — Descriptive Conventions
  2. What a well-written methods section must contain
  3. Descriptive-statistics conventions: continuous variables
  4. Descriptive-statistics conventions: categorical variables
  5. Rounding and precision conventions
  6. Baseline and disposition summaries
  7. Two-sided testing and confidence-interval conventions
  8. General conventions vs endpoint-specific methods
  9. Section A recap: conventions set the stage
  10. B. The Model for Each Endpoint
  11. Why the model must match the endpoint and estimand
  12. Cox proportional-hazards model: time-to-event MACE
  13. Cox model: what must be pre-specified alongside it
  14. MMRM: continuous LDL-C change from baseline
  15. MMRM: covariance structure and specification details
  16. Logistic regression: binary endpoints
  17. Negative-binomial model: recurrent and total MACE
  18. Andersen-Gill: an alternative recurrent-event model
  19. Decision tree: which model for this endpoint?
  20. SENTINEL-3 model-per-endpoint summary
  21. C. Covariates and Stratification — Pre-Specification and the Over-Fitting Danger
  22. Why covariates and stratification must be pre-specified
  23. Stratification factors in SENTINEL-3
  24. Choosing adjustment covariates: prognostic value, not convenience
  25. The over-covariate-ing danger
  26. How many is too many? A practical guideline
  27. Covariate-by-treatment interaction terms: when to include
  28. Pre-specified minimal set vs post-hoc kitchen-sink model
  29. Section C recap: fewer, better-justified covariates
  30. D. Missing-Data Methods, Software and Reproducibility
  31. Naming the missing-data method per endpoint
  32. SAS procedures per model
  33. Specifying multiple imputation precisely
  34. Random-number seeds and reproducibility
  35. Version-locking the software environment
  36. Where the SAP ends and the programming specification begins
  37. A common deficiency: vague missing-data text
  38. Section D recap: precision that survives a re-run
  39. E. Subgroups and Interaction Tests
  40. Why subgroups are analysed at all
  41. Pre-specified vs exploratory subgroups
  42. The formal interaction test
  43. Forest plots: the standard subgroup display
  44. Pitfalls: multiplicity and low power in subgroup testing
  45. SENTINEL-3's pre-specified subgroup list
  46. Section E recap: consistency, not a treasure hunt
  47. F. Over- vs Under-Specification — Writing Testable, Unambiguous Text
  48. The two failure modes, framed together
  49. The over-specification trap
  50. The under-specification trap
  51. Over- vs under-specification, side by side
  52. Writing testable, unambiguous method text: do's and don'ts
  53. Rewriting vague text into testable text
  54. Section F recap: the testability test
  55. G. SAP Version Control and the Amendment Process
  56. Why version control matters for a living document
  57. Version numbering conventions: v0.1 to v1.0 and beyond
  58. The SAP change log: what every entry must capture
  59. Distribution list and sign-off tracking
  60. Who can trigger an amendment during a live trial
  61. Blinded vs unblinded amendments
  62. Before vs after database lock and unblinding
  63. The amendment workflow, end to end
  64. Documenting the reason for an amendment: worked examples
  65. Section G recap: a controlled, traceable lifecycle
  66. H. Relationship to the CSR; QC and Validation
  67. How the SAP feeds the CSR
  68. SAP-to-CSR traceability, section by section
  69. TLF shells and the SAP-CSR link, revisited
  70. QC of the SAP document itself
  71. QC and validation of TLFs: independent double programming
  72. Why double programming exposes SAP deficiencies, not just coding errors
  73. QC and validation checklist: SAP and TLFs together
  74. Section H recap: the SAP's purpose, fully realised
  75. Hands-on: draft a methods paragraph and a SAP change-log entry

  1. A. The Reviewer's Mindset and the Gap-Finding Checklist
  2. What reviewing a SAP means
  3. The reviewer's mindset
  4. Pillar 1 — Pre-specification
  5. Pillar 2 — Coherence across the chain
  6. Pillar 3 — Unambiguity: the two-statistician test
  7. Introducing the structured gap-finding checklist
  8. Checklist — Domain 1: Objectives, Estimands & Populations
  9. Checklist — Domain 2: Analysis, Multiplicity, Missing Data & Interim
  10. Checklist — Domain 3: Conventions, Outputs & Sign-off
  11. The false pass: checklist ticked, coherence still broken
  12. B. The Ten Most Common SAP Deficiencies — Part 1
  13. Overview — deficiencies 1 through 5
  14. Deficiency 1 — Vague endpoint definition
  15. Deficiency 1 — before and after
  16. Deficiency 2 — Estimand missing or incoherent
  17. Deficiency 2 — before and after
  18. Deficiency 3 — Undefined / ambiguous analysis population
  19. Deficiency 3 — before and after
  20. Deficiency 4 — Uncontrolled multiplicity
  21. Deficiency 4 — before and after
  22. Deficiency 5 — LOCF-only / naive missing-data handling
  23. Deficiency 5 — before and after
  24. C. The Ten Most Common SAP Deficiencies — Part 2
  25. Overview — deficiencies 6 through 10
  26. Deficiency 6 — No sensitivity analysis
  27. Deficiency 7 — Methods over- or under-specified
  28. Deficiency 7 — before and after
  29. Deficiency 8 — No visit windows / derivation rules
  30. Deficiency 8 — before and after
  31. Deficiency 9 — TLF shells missing
  32. Deficiency 10 — SAP signed after unblinding
  33. Quick reference — all ten deficiencies, one page
  34. Why these ten deficiencies keep recurring
  35. D. Worked Review — A SENTINEL-3 SAP Excerpt, Section by Section
  36. How the worked review is structured
  37. Excerpt — Section 3, Objectives & Endpoints
  38. Worked review pass 1 — Section 3 model answer
  39. Excerpt — Section 6, Analysis Populations
  40. Worked review pass 2 — Section 6 model answer
  41. Excerpt — Sections 8-9, Statistical Methods & Missing Data
  42. Worked review pass 3 — Sections 8-9 model answer
  43. Excerpt — Sections 7, 13 & 1, Conventions, Shells & Sign-off
  44. Worked review pass 4 — Sections 7/13/1 model answer
  45. E. Writing Effective, Actionable Review Comments
  46. Anatomy of an effective review comment
  47. Weak versus strong comments, side by side
  48. Common comment-writing pitfalls
  49. Turning a vague comment into an actionable one
  50. Tracking findings — the review log
  51. Scope discipline — what a review comment should never do
  52. F. Severity Triage and Sign-off Readiness
  53. Severity triage: critical, major, minor
  54. Severity as likelihood times impact
  55. Decision tree — is this a critical finding?
  56. Decision tree — does this belong in the SAP?
  57. From finding to resolution: the escalation workflow
  58. Sign-off readiness — the gating criteria
  59. The sign-off decision matrix
  60. Documenting the sign-off decision
  61. The rubber-stamp risk
  62. G. Capstone — The Full Checklist, Applied End to End
  63. Capstone briefing
  64. Capstone excerpt — Sections 2 & 10
  65. Capstone — full checklist review, model answer
  66. Capstone sign-off decision
  67. Practice — a fifth excerpt, self-check
  68. When review happens in the SAP timeline
  69. Reviewing your own team's SAP vs an external SAP
  70. From this module back to the real world

  1. 📘 Bonus: Statistical Analysis Plans eBook (Free with purchase)

Course Benefits

Benefits ebook icon
Free eBook

Get our exclusive eBook with every purchase - a complete companion guide to the course, yours to keep forever

Benefits cpd_points icon
CPD Points

Gain Continuing Professional Development (CPD) Points, accredited by The Faculty of Pharmaceutical Medicine of the Royal College of Physicians of the United Kingdom. These can be used to count towards the distance learning element of any scheme that comes under the umbrella of The Academy of Medical Royal Colleges or any other scheme for which there is mutual recognition.

Benefits certification icon
Certification

Receive a personal certificate to show your subject knowledge on course completion.

Benefits affordable icon
Affordable

You get excellent value through our cost-effective prices. We can also offer you group discounts on larger purchases.

Benefits flexible icon
Flexibility

The course saves you time through the convenience of online availability. This lets you complete the interactive course at your own comfort.

Benefits up_to_date icon
Keep Up to Date

You will stay up to date with ICH E9 and the ICH E9(R1) estimands addendum, ICH E3 (Clinical Study Report structure), ICH E6(R2) and the emerging E6(R3) Good Clinical Practice principles, and current FDA and EMA guidance on multiplicity and missing-data handling, as our training courses are constantly monitored, reviewed and updated.

Benefits industry_experts icon
Learn from Industry Experts

The course content has been developed by trial biostatisticians and medical writers with hands-on SAP authoring, review and approval experience, so that what you learn reflects how SAPs are actually written, queried and signed off in industry practice.


Our Certified Customers

novartis
NHS
takeda
roche
dhl

Learner Rating & Reviews

4.7
★★★★★★
Average Rating
536 global ratings
★★★★★
87.0%
★★★★★
5.0%
★★★★★
3.0%
★★★★★
3.0%
★★★★★
2.0%
RC

Working with Whitehall training for the last two years of partnership has been a very successful experience – I have fast access to all the GCP course...

SM

I have finalised the demo for the ICH-GCP E6 R3 refresher course. Overall, I liked the content and the interface. I also want to thank Whitehall Train...

Recommended Courses

R Programming for Clinical Trials

R Programming for Clinical Trials

Intermediate
£393 hrs3 points
SAS Programming for Clinical Trials

SAS Programming for Clinical Trials

Intermediate
£393 hrs3 points
Sample Size Calculations

Sample Size Calculations

Intermediate
£393 hrs3 points
Statistics for Clinical Research

Statistics for Clinical Research

Intermediate
£393 hrs3 points