PAIRED BINARY OUTCOME AUDIT
StatMedAnalysis | 1 November 2026

Purpose: Reconstruct complete pairs and directional disagreements before an exact McNemar test.

Instructions: Save a working copy. Complete each prompt using aggregate summaries or design decisions. Do not include patient identifiers, actual allocation sequences, or confidential records. Write NOT REPORTED, NOT APPLICABLE (with reason), or UNRESOLVED instead of silently leaving a field blank. Resolve flagged issues before using outputs in a report. This is an original planning aid, not a validated instrument or approval certificate.

1. Binary outcome definitions and order of the two conditions or occasions:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

2. Pairing key and how completeness was checked in the approved environment:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

3. Yes/yes, yes/no, no/yes and no/no aggregate counts:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

4. Incomplete pairs, exclusions and missingness assumptions:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

5. Direction b (yes to no) and direction c (no to yes):
Response: __________________________________________________________
Evidence/action: ___________________________________________________

6. Number of discordant pairs b+c and conventional zero-discordance handling:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

7. Null hypothesis and independent-pair assumption:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

8. Exact two-sided binomial, mid-p or asymptotic method; justify and state which:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

9. Effect estimate and appropriate interval to accompany the p-value:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

10. Multiplicity and prespecification status:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

11. Sensitivity questions, responsible analyst and next action:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

12. Software/version, source and independent recomputation/date:
Response: __________________________________________________________
Evidence/action: ___________________________________________________

Related guide and tool: https://www.statmedanalysis.com/resources
Keep the completed copy in your approved research workspace.