Statistical Inspection Readiness Checklist

I. Overall Readiness

☐ Statistical lead / data owner clearly identified
☐ Full understanding of study design and primary/key secondary endpoints
☐ Clear awareness of which analyses support regulatory decision-making
☐ Clear distinction between primary, secondary, and exploratory analyses

---

II. Protocol and SAP Alignment

☐ All analyses are pre-specified in the SAP
☐ Final SAP approved before database lock
☐ SAP version control is clear (dates, approvals, change history)
☐ Any deviations from protocol or SAP are documented and justified
☐ Post hoc analyses are clearly labeled as exploratory in the CSR

Typical inspector question:

Why was this analysis not pre-specified?

---

III. Data Traceability

☐ Every key result can be traced back to source data
☐ Clear linkage between ADaM and SDTM datasets
☐ Data derivation rules are fully documented
☐ Programming logic is consistent with data specifications
☐ No “black-box” variables or undocumented calculations

Result → Program → ADaM → SDTM → CRF → Source Data

---

IV. Programming and Reproducibility

☐ All statistical programs are under version control
☐ Programs can be rerun to reproduce identical results
☐ No manual editing of datasets or outputs
☐ QC evidence available (double programming, independent review, validation)
☐ Outputs are fully consistent with the CSR

---

V. Missing Data and Deviations

☐ Missing data handling methods pre-specified in SAP
☐ Methods align with ICH E9 / E9(R1) principles
☐ Sensitivity analyses support the primary conclusions
☐ Protocol deviations are clearly defined and categorized
☐ Deviation handling is not outcome-driven

---

VI. Statistical Soundness

☐ Statistical methods are appropriate for the study design
☐ Assumptions are reasonable and explainable
☐ Multiplicity control strategy clearly defined
☐ Analysis populations (ITT, PP, SAF, etc.) clearly specified
☐ No unnecessary complexity without scientific justification

---

VII. Documentation and Evidence

☐ Protocol (all versions)
☐ SAP (final and historical versions)
☐ TFL shells
☐ SDTM / ADaM specifications
☐ Program inventory with version history
☐ Audit trails and change logs

---

VIII. Inspection Interview Readiness

☐ Able to explain the rationale for each key analysis
☐ Able to clearly describe data sources and derivations
☐ Distinguishes facts from interpretations
☐ Answers based on documentation, not memory
☐ Avoids ad hoc analyses or on-the-spot commitments

---

IX. Common Red Flags

❌ SAP finalized after database lock
❌ Unexplained discrepancies between SAP and results
❌ Programs not reproducible
❌ Undocumented data derivations
❌ Modifying data or programs during inspection

---

X. Rapid Self-Check Questions

• Can I explain how the primary endpoint was derived in 5 minutes?
• Can I trace any reported value back to its source data?
• Are sensitivity analyses available to support conclusions?
• Is there any result that only one person knows how to reproduce?

---

Summary

Inspection readiness for statisticians means that every result is reproducible, traceable, and scientifically justified—at any time.

Inspection Target	Possible Timing	Primary Purpose
Clinical trial sites	During trial conduct / Pre-Approval Inspection (PAI)	Data integrity and GCP compliance
Sponsor	IND / NDA stages	Oversight systems and accountability
CRO	Any stage	Vendor oversight and delegated responsibilities
Statistics / data systems	NDA / PAI	Reproducibility and data integrity
Manufacturing facilities	NDA review / Post-approval	GMP compliance

Inspection vs. Audit — Comparison Table

Dimension	Inspection	Audit
Conducted by	Regulatory authorities (e.g., FDA, NMPA, EMA)	Sponsor, Quality Assurance, or third-party auditors
Legal nature	Regulatory and enforcement activity	Quality management activity
Mandatory	Yes — cooperation is required	No — conducted by agreement or internal policy
Primary purpose	Determine regulatory compliance and impact on regulatory decisions	Identify risks and drive continuous improvement
Timing	Risk-based, during review, or triggered by signals	Planned, periodic, or risk-based
Scope	Focused on critical data, systems, and processes	Systematic review of processes and controls
Approach	Evidence-based, traceability-driven, result-oriented	Process-based, improvement-oriented
Notification	May be announced or unannounced	Typically pre-scheduled
Findings	Observations (e.g., FDA Form 483), inspection reports	Audit findings and internal observations
Consequences	May affect IND/NDA approval, licensing, or lead to enforcement actions	Leads to CAPA and internal corrective actions
Negotiability of outcome	No	Yes
Refusal allowed	No	Yes

---

Key takeaway

An audit is performed to prevent inspection failures; an inspection is the ultimate regulatory test of audit effectiveness.