Chapter 16 Quality and Compliance

16.1 The Invisible Backbone of Statistical Credibility

In clinical trials, quality and compliance are rarely highlighted when everything runs smoothly—but they become immediately visible when something goes wrong. For Project Biostatisticians, quality is not an abstract regulatory topic; it is a set of daily operational practices that determine whether statistical work is defensible under audit and inspection.

This chapter focuses on practical actions that help statisticians ensure their work withstands internal audits, sponsor scrutiny, and regulatory inspections.

16.2 Why Quality and Compliance Matter for Statistics

16.2.1 Statistics Is a Regulated Discipline

Clinical trial statistics operates under explicit regulatory expectations. Statistical outputs are not only scientific results—they are regulated deliverables. Regulators and auditors expect that:

  • Analyses follow predefined plans (Protocol and SAP)
  • Deviations are controlled, justified, and documented
  • Results are reproducible from archived artifacts
  • Key decisions are traceable and reviewable

Quality failures are often not about incorrect calculations, but about lack of process control and insufficient documentation.

16.2.2 The Statistician’s Accountability

For Project Biostatisticians, quality and compliance responsibilities include:

  • Following applicable SOPs for statistical work
  • Aligning methods and reporting with international guidelines (ICH E9 and ICH E3)
  • Maintaining complete, traceable statistical documentation
  • Supporting audit and inspection readiness with confidence and clarity

A useful mindset is:

If you cannot explain and reproduce your analysis under inspection, it does not meet regulatory quality standards.

16.3 Core Frameworks: SOPs, ICH E9, and ICH E3

16.3.1 SOP Compliance: The First Line of Defense

Standard Operating Procedures (SOPs) define how statistical activities must be conducted within an organization. For biostatistics, SOPs commonly cover:

  • Protocol and SAP development and approval
  • Statistical programming and validation expectations
  • QC and peer review processes
  • Data handling, access control, and security
  • Document management and archiving rules

The statistician’s practical responsibility is to: - Know which SOPs apply to each deliverable - Follow them consistently - Document any justified deviations (with rationale and approvals)

Unrecorded deviations are generally far riskier than justified and documented ones.

16.3.2 ICH E9: Statistical Principles for Clinical Trials

ICH E9 establishes expectations for statistical credibility, including:

  • Clear objectives (and, in modern practice, estimand clarity)
  • Appropriate selection and justification of statistical methods
  • Proper handling and interpretation of missing data
  • Control of multiplicity for confirmatory claims
  • Transparent interpretation of results and limitations

From a compliance perspective, ICH E9 alignment means that the analysis must address the intended clinical question and should not be reframed post hoc to fit the observed data.

16.3.3 ICH E3: Clinical Study Reports

ICH E3 governs how results are documented in the CSR, including:

  • A clear description of statistical methods
  • Transparent, coherent reporting of results
  • Consistency between methods, results, and conclusions
  • Adequate explanation and documentation of deviations from the plan

Statistical inconsistencies between SAP, TFLs, and CSR narratives are common triggers for audit findings.

16.4 Statistical Documentation and Archiving

16.4.1 Why Archiving Is a Statistical Responsibility

Archiving is sometimes seen as an administrative task, but for statisticians it is a core quality activity. Archived materials must enable an independent reviewer to:

  • Understand what was planned
  • See what was executed
  • Reproduce key results
  • Trace decisions, changes, and approvals

Poor archiving effectively equals undocumented analysis.

16.4.2 What Should Be Archived (Practical List)

A typical inspection-ready statistical archive should include:

  • Protocol and all amendments
  • SAP and SAP amendments (with approval history)
  • TFL shells and output specifications
  • Final analysis datasets (e.g., ADaM) and metadata
  • Statistical programs, execution logs, and run documentation
  • QC records, checklists, and sign-offs
  • Final TFLs and CSR statistical sections
  • Version histories and change logs for key deliverables
  • Key decision records (e.g., analysis clarifications, deviation justifications)

The goal is completeness and traceability, not minimalism.

16.4.3 Practical Archiving Best Practices

Useful best practices include:

  • Consistent version control
    • File naming conventions that include study ID, artifact type, version, and date
    • Clear “Final/Locked” designation for submission-ready items
  • Logical, navigable folder structure
    • A structure that mirrors the statistical lifecycle (SAP → ADaM → TFL → CSR)
  • Reproducibility support
    • Documentation of software environment, seeds (if applicable), and run instructions
    • Read-only protection for final outputs

A simple test is: if someone unfamiliar with the project cannot navigate and reproduce key results, the archive is not inspection-ready.

16.5 Audit and Inspection Support

16.5.1 Audits vs. Inspections (Operational Differences)

Although often discussed together, audits and inspections differ operationally:

  • Audits
    • May be internal, sponsor-led, or vendor-focused
    • Often assess process compliance and readiness
  • Inspections
    • Conducted by regulatory authorities (e.g., FDA, EMA, NMPA)
    • Focus on data integrity, compliance, and defensibility of conclusions

Statisticians may be involved in both, and expectations for traceability and clarity are high.

16.5.2 The Statistician’s Role During Audit/Inspection

During audits or inspections, statisticians are expected to:

  • Explain analysis decisions clearly and consistently
  • Demonstrate traceability from protocol/SAP to datasets and outputs
  • Provide supporting documentation promptly
  • Answer questions calmly, factually, and without speculation

Inspectors are primarily testing transparency and process control.

16.5.3 Common Statistical Audit/Inspection Questions

Typical questions include:

  • How were analysis populations defined and applied?
  • How were missing data handled, and why is the approach appropriate?
  • Were SAP deviations identified, justified, and approved?
  • How was QC conducted and documented?
  • Can the primary endpoint results be reproduced from archived artifacts?

Many “statistical” findings ultimately reflect documentation gaps, unclear traceability, or uncontrolled changes.

16.6 Practical Guidance for Inspection Readiness

16.6.1 Preparation Before an Audit/Inspection

Proactive preparation should include:

  • Ensuring SAPs and amendments are finalized, approved, and archived
  • Verifying that final TFLs match archived analysis datasets
  • Confirming QC documentation is complete and signed
  • Reviewing known deviations and ensuring written justifications exist
  • Checking that version control and file naming are consistent

Inspection readiness should be continuous rather than reactive.

16.6.2 Behavior During Audit/Inspection

During questioning:

  • Answer only what is asked—avoid unnecessary expansion
  • Use documented evidence whenever possible
  • Avoid guessing; if unsure, acknowledge and follow up with documentation
  • Escalate appropriately when a question requires cross-functional input

A calm, consistent approach increases credibility and reduces follow-up burden.

16.6.3 After an Audit/Inspection: CAPA and Continuous Improvement

After an audit or inspection:

  • Support root cause analysis for any findings
  • Contribute to corrective and preventive actions (CAPA)
  • Update processes, checklists, or SOP interpretations if needed
  • Capture lessons learned for future studies

Quality improvement is an ongoing cycle, not a one-time exercise.

16.7 Key Takeaways

  1. Quality and compliance are foundational to statistical credibility.
  2. SOP adherence protects the study, the sponsor, and the statistician.
  3. ICH E9 and ICH E3 set expectations for defensible methods and reporting.
  4. Documentation and archiving are essential for reproducibility and traceability.
  5. Audits and inspections test transparency and process—not just correctness.
  6. Inspection readiness should be continuous, systematic, and evidence-based.