How to read this source
Start with headings and summary blocks. Use this page to verify citations, claim wording, and chapter-level evidence support before interpreting conclusions.
Internal Research Source
Statistical Analysis Plan Template
Reusable SAP template for endpoint definitions, sample-size logic, subgroup analysis, and reporting standards.
templateEvidence status: active sourceUpdated 2026-02-15
Used by chapters
Share
Section index
- 1) Scope and Objectives
- 2) Analysis Populations
- 3) Endpoint Definitions
- 4) Sample Size Calculation
- 4.1 Proportion endpoint planning
- 4.2 Continuous-error endpoint planning
- 5) Model and Decision Logic
- 6) Missing Data and QC Handling
- 7) Subgroup and Fairness Analyses
- 8) Robustness and Stress Analyses
- 9) Multiplicity and Error Control
- 10) Interim Analysis and Stopping Rules
- 11) Output Tables and Figures
- 12) Acceptance Criteria
--- doc_id: ASTRACBC-OCULAR-SAP-TEMPLATE title: Statistical Analysis Plan Template - Ocular Smartphone-Only SaMD version: v0.1-draft status: template owner: Biostatistics audience: Clinical Validation, Regulatory, Data Science device_name: AstraCBC updated_at: 2026-02-15 ---
Statistical Analysis Plan (SAP) Template
1) Scope and Objectives
- Study ID:
- Claims covered:
- Populations covered:
2) Analysis Populations
- Intent-to-diagnose (ITD)
- Per-protocol (PP)
- Safety population
Define inclusion/exclusion from each population explicitly.
3) Endpoint Definitions
| Endpoint | Type | Metric | Decision threshold | CI method | |---|---|---|---|---| | | | | | |
4) Sample Size Calculation
4.1 Proportion endpoint planning
Use:
`n ~= z^2 * p(1-p) / w^2`
Worked example fields:
- confidence level (z):
- expected proportion (p):
- desired half-width (w):
- resulting `n`:
4.2 Continuous-error endpoint planning
Use:
`n ~= (z*sigma / w)^2`
Worked example fields:
- sigma assumption:
- desired half-width:
- resulting `n`:
5) Model and Decision Logic
- Score calibration method:
- Threshold locking process:
- Abstain policy:
- Tie-break/arbitration policy:
6) Missing Data and QC Handling
- Missingness taxonomy:
- Reacquire handling:
- Abstained sample accounting:
- Sensitivity analyses:
7) Subgroup and Fairness Analyses
Required dimensions:
- device family
- OS major version
- camera capability tier
- age/sex/skin tone proxy groups
Gate template:
- max allowed subgroup delta:
- minimum subgroup sample:
8) Robustness and Stress Analyses
Include perturbation tests:
- illumination
- glare
- motion/blur
- compression
- ambient-subtraction mismatch
9) Multiplicity and Error Control
- Family-wise control plan:
- Secondary endpoint handling:
10) Interim Analysis and Stopping Rules
- Interim looks:
- Alpha spending:
- Safety stop criteria:
11) Output Tables and Figures
Minimum outputs:
- Confusion matrix by subgroup
- Calibration curve
- ROC/PR curves
- Abstain rate distributions
- Error histograms
- Site/device holdout results
12) Acceptance Criteria
- Primary endpoints met with predefined CIs
- Subgroup gates pass
- Abstain rates within limits
- No unresolved data integrity deviations