Clinical Trial Protocol Skill

Step 4: Clinical Protocol Operations and Statistics

Purpose

This subskill generates the operational, assessment, and statistical sections (8-11) of a clinical study protocol. It adds efficacy and safety assessments, adverse event reporting, statistical analysis plan with sample size calculation, regulatory/operational considerations, and references to complete the protocol.

Prerequisites

Required Files:

• waypoints/intervention_metadata.json (must contain protocol_template field from Step 2)

• waypoints/01_clinical_research_summary.json

• waypoints/02_protocol_foundation.md (sections 1-5 from Step 2)

• waypoints/03_protocol_intervention.md (sections 6-7 from Step 3)

• waypoints/02_protocol_metadata.json (must show step_3_status: "completed")

• scripts/sample_size_calculator.py (for sample size calculation)

What This Subskill Does

1. Verifies Step 3 completion

2. Performs sample size calculation (asks user for parameters, runs calculator script)

3. Reads protocol template (from metadata) and shared guidance

4. Synthesizes operational and statistical content for Sections 8-11

5. Creates new file waypoints/04_protocol_operations.md with Sections 8-11

6. Concatenates all protocol sections into waypoints/02_protocol_draft.md

7. Updates protocol and intervention metadata

Execution Flow

Step 1: Verify Phase 3 Completion

Read waypoints/02_protocol_metadata.json and verify:

• step_3_status is "completed"

• waypoints/02_protocol_foundation.md exists (Sections 1-5)

• waypoints/03_protocol_intervention.md exists (Sections 6-7)

If Step 3 not completed:

Error: Step 3 must be completed before Phase 4.
Phase 3 status: [current status]

Please complete Step 3 first to generate intervention details (Sections 6-7).

Exit.

If Phase 3 completed: Proceed to Step 2.

Step 2: Calculate Sample Size

BEFORE generating Section 9, perform sample size calculation.

2.1 Extract Endpoint Information

Read from waypoints/02_protocol_draft.md Section 3 (Objectives and Endpoints):

• Primary endpoint description

• Classify endpoint type:

  • Continuous: Mean change, score change (e.g., "Mean change from baseline in systolic blood pressure")

  • Binary: Proportion, success rate, response rate (e.g., "Proportion of participants achieving treatment success")

  • Time-to-event: Survival, time to progression (Note: Not commonly used for primary endpoint in Phase 2/3, but possible)

2.2 Determine Study Design

From Section 4 (Study Design):

• RCT (randomized controlled trial) vs Single-arm

• Superiority vs Non-inferiority

2.3 Prompt User for Statistical Parameters

For CONTINUOUS endpoints:

Ask user conversationally:

📊 Sample Size Calculation - Continuous Endpoint

Primary Endpoint: [From Section 3, e.g., "Mean reduction in systolic blood pressure at 6 months"]

To calculate sample size, I need the following information from you:

1. Expected effect size (mean difference between groups):
   Based on similar trials from Phase 1: [e.g., "5 mmHg reduction expected"]

   Please confirm or provide your expected effect size (as a number): ___

2. Expected standard deviation:
   Based on similar trials from Phase 1: [e.g., "Standard deviation typically 15 mmHg"]

   Please confirm or provide expected standard deviation: ___

3. Study design (default: superiority):
   - Superiority (showing intervention is better than control)
   - Non-inferiority (showing intervention is not worse than control)

   Your choice (superiority/noninferiority, default: superiority): ___

Optional parameters (press Enter to use defaults):
- Alpha (Type I error rate): 0.05 (5%)
- Power: 0.80 (80%)
- Expected dropout rate: 0.15 (15%)

Would you like to customize these optional parameters? (Yes/No, default: No)

If user wants to customize optional parameters, ask for each.

For BINARY endpoints:

Ask user conversationally:

📊 Sample Size Calculation - Binary Endpoint

Primary Endpoint: [From Section 3, e.g., "Treatment success rate at 6 months"]

To calculate sample size, I need the following information from you:

1. Expected proportion in CONTROL group:
   Based on similar trials from Phase 1: [e.g., "60% success rate in control"]

   Please provide control group proportion (0 to 1, e.g., 0.60 for 60%): ___

2. Expected proportion in TREATMENT group:
   Based on similar trials from Phase 1: [e.g., "75% success rate in treatment"]

   Please provide treatment group proportion (0 to 1, e.g., 0.75 for 75%): ___

3. Study design (default: superiority):
   - Superiority (showing intervention is better than control)
   - Non-inferiority (showing intervention is not worse than control)

   Your choice (superiority/noninferiority, default: superiority): ___

Optional parameters (press Enter to use defaults):
- Alpha (Type I error rate): 0.05 (5%)
- Power: 0.80 (80%)
- Expected dropout rate: 0.15 (15%)

Would you like to customize these optional parameters? (Yes/No, default: No)

For TIME-TO-EVENT endpoints (if applicable):

Note: Sample size calculator may not support time-to-event. If primary endpoint is time-to-event, use pattern-based approach from similar trials in Phase 1, and add disclaimer.

2.4 Run Sample Size Calculator Script

For non-inferiority designs, do not run the calculator, whether the design comes from Section 4 (step 2.2) or from the user's answer at 2.3. The calculator has no option for a non-inferiority margin, so it cannot size these trials. Instead, follow the "If script fails" steps in 2.5, and show the "If sample size calculation fails" warning under Error Handling with [error message] replaced by "The calculator cannot size non-inferiority trials because it takes no non-inferiority margin." If Section 4 and the user's answer at 2.3 name different designs (keeping the superiority default counts as an answer), also tell the user that they differ, so they can correct whichever one is wrong.

BEFORE running the calculator, display this message to the user:

🔬 Calculating Sample Size

Running statistical power calculation with your parameters...

This will:
• Calculate the required sample size per arm
• Adjust for expected dropout rate
• Generate detailed assumptions for the statistical section
• Save results to waypoints/02_sample_size_calculation.json

Please wait while the calculation completes.

For continuous endpoints:

python scripts/sample_size_calculator.py \
  --type continuous \
  --effect-size [user provided value] \
  --std-dev [user provided value] \
  --alpha [user provided or 0.05] \
  --power [user provided or 0.80] \
  --dropout [user provided or 0.15] \
  --design superiority \
  --output waypoints/02_sample_size_calculation.json

For binary endpoints:

python scripts/sample_size_calculator.py \
  --type binary \
  --p1 [control proportion] \
  --p2 [treatment proportion] \
  --alpha [user provided or 0.05] \
  --power [user provided or 0.80] \
  --dropout [user provided or 0.15] \
  --design superiority \
  --output waypoints/02_sample_size_calculation.json

AFTER running the calculator successfully, display this message to the user:

✅ Sample Size Calculation Complete

Results:
• Sample Size Per Arm: [n_per_arm from calculation]
• Total Sample Size (Unadjusted): [n_total from calculation]
• Adjusted for Dropout ([dropout]%): [adjusted_n_total from calculation]
• Final Enrollment Target: [adjusted_n_total] participants

These results will be incorporated into Section 9.2 (Sample Size Determination) of the protocol.

2.5 Read Calculation Results

If script succeeds:

• Read waypoints/02_sample_size_calculation.json

• Extract:

  • Endpoint type

  • Sample size per arm (n_per_arm)

  • Total sample size (n_total)

  • All input assumptions (effect_size, std_dev or p1/p2, alpha, power, dropout)

  • Adjusted sample size accounting for dropout

If script fails:

• Use pattern-based approach: Look at similar trials from Step 1 research, use their enrollment numbers

• Do not use a waypoints/02_sample_size_calculation.json left from an earlier run. Wherever later instructions in this file use the calculation results or that file, use the enrollment target from similar trials instead; wherever they report the calculation as completed or the file as created, say the calculation was not performed (in Step 6, set sample_size_calculated to false).

• Add prominent disclaimer: "Sample size calculation could not be performed. Enrollment target based on similar trials: [NCT numbers]. Requires biostatistician review and validation."

Step 3: Read Protocol Template and Shared Guidance

1. Read waypoints/intervention_metadata.json to get the protocol_template and template_source field values

2. Read the protocol template file:

   • The path is stored in protocol_template field (could be assets/[name].md OR a user-provided path)

   • Read the file at that path (focus on Sections 8-11)

   • Note: The template_source field indicates whether this is "directory" or "user_provided" (for logging/debugging only)

3. Read waypoints/02_protocol_foundation.md and waypoints/03_protocol_intervention.md to understand existing context (especially study design, endpoints, population, and Schedule of Activities)

Internalize:

• Template structure for Sections 8-11

• Content depth targets from shared guidance

• Sample size calculation results (from Step 2)

• Quality principles (comprehensive over brief, specific over generic)

Step 4: Synthesize Operations and Statistics Content

Generate Sections 8-11 following template structure, adapted for this specific intervention and indication.

Use data from:

• Sample size calculation results (waypoints/02_sample_size_calculation.json)

• Intervention metadata

• Phase 1 clinical research summary

• Existing protocol draft (Sections 1-8)

Follow guidance from:

• Selected protocol template from Step 3 (structure, regulatory requirements)

Section Generation Instructions

Section 8: STUDY ASSESSMENTS AND PROCEDURES (800 lines)

8.1 Efficacy Assessments (100 lines)

Primary Endpoint Measurement (40 lines):

• Endpoint name: [From Section 3]

• Detailed description of measurement procedure

• Instrument or scale used: Include name, range, clinically meaningful difference

  • If validated scale: Cite validation reference, describe domains

  • If clinical measurement: Describe procedure in detail

• Timing of assessments: Reference Schedule of Activities (Section 2.3)

• Who performs assessment: Trained investigator, blinded assessor, central reader, participant self-report

• Standardization procedures: Training requirements, quality control procedures, calibration (for devices)

• Data collection: Source documents, eCRF entry

Secondary Endpoint Measurements (60 lines):

For EACH secondary endpoint (2-5 endpoints):

• Endpoint name: [From Section 3]

• Measurement procedure: Similar detail as primary

• Validated instrument specification: Name, version, scoring

• Timing: Per SoA

• Assessor: Who performs

• Quality control

Common validated instruments (examples - use appropriate for indication):

• Quality of life: EQ-5D-5L, EORTC QLQ-C30, SF-36

• Symptom scales: Disease-specific scales

• Functional assessments: Karnofsky, ECOG, Barthel Index

• Cognitive assessments: MMSE, MoCA

• Imaging criteria: RECIST v1.1 for solid tumors, Lugano for lymphoma

8.2 Safety and Other Assessments (200 lines)

Laboratory Tests (60 lines):

• Timing: Reference Schedule of Activities

• Hematology specifications:

  • Complete blood count (CBC) with differential

  • Parameters: WBC, RBC, hemoglobin, hematocrit, platelets, ANC, lymphocytes, monocytes, eosinophils, basophils

  • Method: Automated hematology analyzer

  • Fasting: Not required

• Chemistry panel specifications:

  • Comprehensive metabolic panel

  • Parameters: Sodium, potassium, chloride, CO2, glucose, BUN, creatinine, calcium, albumin, total protein, alkaline phosphatase, ALT, AST, total bilirubin

  • Additional: Creatinine clearance (calculated by Cockcroft-Gault equation)

  • Fasting: Specify if required for glucose

• Coagulation (if relevant): PT/INR, aPTT

• Urinalysis (if relevant): Dipstick or microscopic

• Disease-specific labs: Thyroid function, cardiac biomarkers, tumor markers, etc.

• Sample handling: Collection, processing, storage, shipping to central lab

• Reference ranges: Local lab normal ranges

Imaging (40 lines):

• Modality: CT, MRI, PET, X-ray, ultrasound

• Anatomical coverage: Chest/abdomen/pelvis, brain, etc.

• Protocol specifications: Contrast vs non-contrast, slice thickness, image acquisition parameters

• Timing: Per Schedule of Activities

• Quality control: Central imaging review (if applicable)

• Assessment criteria: RECIST v1.1, modified RECIST, WHO criteria

• Image storage: PACS, DICOM format

Physical Examinations (30 lines):

• Complete physical examination: Baseline and final visit - all body systems

• Directed physical examination: Interim visits - focus on intervention-related findings

• Components: General appearance, vital signs, HEENT, cardiovascular, pulmonary, abdominal, neurological, musculoskeletal, skin

• Disease-specific exams: Tumor measurements, neurological exam, etc.

Vital Signs (20 lines):

• Parameters: Blood pressure, heart rate, respiratory rate, temperature, oxygen saturation

• Measurement procedures: Position (seated, supine), rest period before measurement, equipment specifications

• Timing: Per SoA

• Clinically significant changes: Define thresholds requiring action

Electrocardiogram (ECG) (if required) (30 lines):

• Type: 12-lead ECG

• Timing: Per SoA

• Interpretation: Investigator interpretation, central reading (if applicable)

• QTc calculation: Specify formula (Fridericia, Bazett)

• QTc prolongation criteria: Define threshold for concern (e.g., QTc >480 ms or increase >60 ms from baseline)

Cardiac Assessments (if required) (30 lines):

• Echocardiogram or MUGA scan

• Left ventricular ejection fraction (LVEF)

• Timing: Per SoA

• LVEF decline criteria: Define threshold for dose modification or discontinuation

Disease-Specific Assessments (30 lines):

• Tumor measurements: Bidimensional or RECIST

• Symptom scores: Disease-specific validated scales

• Functional assessments: ADL, IADL

• Other disease-specific measures

Quality Control for All Assessments (20 lines):

• Training of assessors: Required training, certification

• Standard operating procedures: SOPs for each assessment

• Calibration: Equipment calibration schedules

• Data quality checks: Range checks, consistency checks

• Central review: Specify assessments requiring central review

8.3 Adverse Events and Serious Adverse Events (300 lines)

CRITICAL: Generate ALL subsections 8.3.1-8.3.9 with complete regulatory definitions.

8.3.1 Definition of Adverse Events (AE) (20 lines):

• ICH-GCP definition: Any untoward medical occurrence in a participant administered a pharmaceutical product or medical device, whether or not related to treatment. Includes unfavorable signs, symptoms, or disease temporally associated with intervention use.

• Examples in this study context: [List 5-10 anticipated AEs based on intervention]

• Distinguish AEs from expected disease symptoms: [Specify how to differentiate]

• Pre-existing conditions: Worsening = AE

8.3.2 Definition of Serious Adverse Events (SAE) (20 lines):

• Regulatory definition per 21 CFR 312.32 (drugs) or 21 CFR 812.3(s) (devices)

• SAE criteria (any one of):

  1. Death

  2. Life-threatening

  3. Hospitalization (new or prolonged)

  4. Persistent/significant disability

  5. Congenital anomaly/birth defect

  6. Important medical event requiring intervention

• Exclusions: Planned protocol hospitalizations, elective procedures, social admissions

• Examples: [List 5-10 potential SAEs for this intervention]

8.3.3 Classification of an Adverse Event (100 lines):

8.3.3.1 Severity of Event (30 lines):

• CTCAE v5.0 grading system (Common Terminology Criteria for Adverse Events)

• Grade 1 (Mild): Asymptomatic/mild; no intervention; [2-3 examples for intervention]

• Grade 2 (Moderate): Minimal intervention indicated; limiting instrumental ADL; [2-3 examples]

• Grade 3 (Severe): Medically significant; hospitalization/disabling; limiting self-care ADL; [2-3 examples]

• Grade 4 (Life-threatening): Urgent intervention indicated; [2-3 examples]

• Grade 5 (Death): Death related to AE

Important: Severity (Grade) ≠ Seriousness (SAE). Example: Grade 1 reaction requiring hospitalization = SAE.

CTCAE reference: https://ctep.cancer.gov/protocoldevelopment/electronic_applications/ctc.htm

8.3.3.2 Relationship to Study Intervention (30 lines):

5-point attribution scale:

• Not Related: Clearly explained by other causes; [1 example]

• Unlikely: Doubtfully related; alternatives more likely; [1 example]

• Possible: Reasonable temporal relationship but other factors possible; [1 example]

• Probable: Likely related; alternatives less likely; [1 example]

• Definite: Clear temporal relationship, known pattern, no other explanation; [1 example]

Assessment: Investigator assesses each AE; document rationale; when uncertain, use "Possible" or "Probable"

Reporting: Related = Possible/Probable/Definite; Unrelated = Not Related/Unlikely; Unexpected related SAEs require expedited FDA reporting

8.3.3.3 Expectedness (20-40 lines):

• Expected: Listed in Investigator's Brochure, device labeling/IFU, Protocol Section 3.3, or FDA-approved labeling

• Unexpected: Not listed OR greater severity/frequency/different body system than listed

SUSAR: Suspected Unexpected Serious Adverse Reaction = Unexpected + Related + Serious (requires expedited FDA/IRB reporting)

Reference documents for this study: [List specific IB, device labeling, Protocol Section 3.3]

8.3.4 Time Period and Frequency for Event Assessment and Follow-Up (50 lines):

AE Collection Start:

• AE collection begins: At time of informed consent OR at first dose of study intervention (specify)

• Pre-treatment AEs: AEs occurring after consent but before first dose are captured as baseline conditions (medical history)

AE Collection End:

• For drugs: AE collection ends 30 days after last dose of study intervention

• For devices: AE collection may extend longer (specify: 90 days, 6 months, or through final visit) depending on device type and potential for late effects

• For long-acting agents or biologics: May extend AE collection period (specify)

• SAEs: Continue to collect SAEs beyond 30-day period if related to study intervention

Frequency of Assessment:

• At each study visit per Schedule of Activities (Section 2.3)

• Continuous: Participants instructed to report new symptoms immediately between visits (provide contact information)

• Elicited by investigator: "Have you had any health problems since your last visit?"

• Volunteered by participant: Any symptoms volunteered by participant

Follow-Up:

• AEs followed until: Resolution, stabilization, or determination that event is unlikely to resolve (chronic condition)

• Ongoing AEs at study completion: Continue follow-up until resolution or for 30 days after last dose (minimum)

• SAEs: Follow until resolution or stabilization

Late-Onset Events:

• Events occurring after 30-day follow-up: If participant contacts site with event potentially related to intervention, report as AE

8.3.5 Adverse Event Reporting (50 lines):

Participant Responsibility:

• Participants instructed to report all new symptoms, health problems, or changes in health status immediately

• Contact information: 24/7 contact number provided on wallet card

Site Responsibility:

• Document ALL AEs in participant's medical record (source documentation)

• Record AEs in electronic Case Report Form (eCRF)

• Timeframe: Within 24 hours of becoming aware of AE

AE Documentation (for each AE):

• AE term: Use participant's words, then translate to medical terminology

• Start date: Date AE first occurred or was first observed

• Stop date: Date AE resolved or stabilized

• Severity: CTCAE Grade (1-5)

• Seriousness: SAE criteria met (Yes/No)

• Relationship: Attribution (Not Related, Unlikely, Possible, Probable, Definite)

• Action taken: Intervention held, dose reduced, discontinued, none, etc.

• Outcome: Resolved, resolved with sequelae, ongoing, fatal

• Treatment: Medications or interventions given for AE

MedDRA Coding:

• All AEs coded using Medical Dictionary for Regulatory Activities (MedDRA) version [specify version, e.g., 25.1]

• Lowest Level Term (LLT) and Preferred Term (PT) assigned

• System Organ Class (SOC) assigned for analysis

Routine Submission to Sponsor:

• Non-serious AEs: Reported to sponsor per scheduled data transmission (typically weekly or monthly)

• Sponsor monitors AE data for trends and signals

8.3.6 Serious Adverse Event Reporting (80 lines):

IMMEDIATE Reporting Requirements:

To Sponsor:

• Timeframe: Within 24 hours of site becoming aware that AE meets SAE criteria

• Method: Phone call or email to sponsor safety contact (provide contact information)

• Initial report: May be preliminary if full information not yet available

• Follow-up report: Complete SAE report form within 24-48 hours

• Ongoing follow-up: Updates as new information becomes available until resolution

SAE Report Form Contents:

• Participant information: Study ID, initials, age, sex

• SAE details: Event term, start/stop dates, outcome

• Seriousness criteria: Which SAE criterion met (death, life-threatening, hospitalization, etc.)

• Severity: CTCAE grade

• Relationship to intervention: Attribution

• Expectedness: Expected vs unexpected

• Action taken with intervention: Held, reduced, discontinued, none

• Concomitant medications and procedures

• Investigator narrative: Description of event, clinical course, outcome

To IRB:

• Timeframe: Per IRB requirements (typically within 5-10 calendar days for unexpected related SAEs)

• Method: IRB submission portal or email

• Events reportable to IRB: Unexpected + Related SAEs, deaths, protocol deviations involving increased risk

• Annual report: Summary of all SAEs in annual continuing review

To FDA (Sponsor Responsibility):

• Sponsor submits IND Safety Reports (for drugs, 21 CFR 312.32) or IDE reports (for devices, 21 CFR 812.150)

• Fatal or life-threatening unexpected related SAEs: 7 calendar days (initial report), 15 days (follow-up)

• Other unexpected related SAEs: 15 calendar days

• Annual report: Summary of SAEs in IND/IDE annual report

Expedited Reporting (SUSARs):

• Suspected Unexpected Serious Adverse Reactions (SUSARs): Serious + Unexpected + Related

• Reported by sponsor to FDA within regulatory timeframes

• Investigators at all sites notified of SUSARs

Follow-Up Reports:

• Submit follow-up SAE reports as new information becomes available

• Final report when SAE resolves or stabilizes

8.3.7 Reporting Events to Participants (30 lines):

When to Inform Participants:

• New safety information emerges that may affect risk/benefit or willingness to participate

• SUSAR occurs at this site or other sites

• Protocol amended to address new risks

Process:

• Investigator notifies participants of new safety information

• Provide written summary of new information

• Allow participants to ask questions

• Document notification in study records

Consent Form Updates:

• If new risks identified, informed consent form amended

• Amendment submitted to IRB for approval

• Re-consent process required for currently enrolled participants

• New participants sign updated consent form

Participant Right to Withdraw:

• Participants always have right to withdraw from study

• New safety information may prompt some participants to withdraw

8.3.8 Events of Special Interest (30 lines):

Intervention-Specific Events Requiring Special Monitoring:

For drugs/biologics:

• Infusion reactions: Hypersensitivity, cytokine release syndrome

• Immune-related AEs (for immunotherapy): Pneumonitis, colitis, hepatitis, endocrinopathies, dermatitis, neuropathies

• Organ-specific toxicities: Cardiotoxicity, nephrotoxicity, hepatotoxicity, neurotoxicity

• Hypersensitivity reactions: Anaphylaxis, severe allergic reactions

• Tumor lysis syndrome (for oncology interventions)

For devices:

• Device malfunctions: Any failure of device to operate as intended

• Device migration or displacement

• Device fracture or breakage

• Infections related to device: Surgical site infections, device-related infections

• Need for device removal or revision

• Thrombosis or embolism related to device

Reporting and Management:

• All events of special interest reported to sponsor within 24 hours

• Follow protocol-specific management algorithms (reference Section 7.2 for dose modifications)

• For devices: Device malfunctions reported to sponsor and manufacturer immediately; may require device return for analysis

• Document detailed description, management, and outcome

8.3.9 Reporting of Pregnancy (40 lines):

All Pregnancies Reported Immediately:

• Pregnancy in female participant

• Pregnancy in partner of male participant

Detection:

• Pregnancy testing per Schedule of Activities: Serum β-hCG at screening, urine pregnancy tests at subsequent visits for women of childbearing potential (WOCBP)

• Participant self-report: Any missed menstrual period or suspected pregnancy reported immediately

Reporting:

• Report pregnancy to sponsor within 24 hours of awareness

• Pregnancy report form: Estimated date of conception, estimated due date, pregnancy test result

Study Intervention:

• Intervention discontinued immediately upon pregnancy confirmation

• No further doses administered

Follow-Up:

• Follow pregnancy to outcome: Live birth, miscarriage, elective termination, ectopic pregnancy, stillbirth

• Outcome report: Include pregnancy outcome, complications, birth defects (if any)

• Infant follow-up: Obtain information on infant health at birth (if participant consents)

• Timeframe: Follow through delivery and obtain outcome information

Partner Pregnancies:

• Male participants must report if their partner becomes pregnant during study

• Report to sponsor within 24 hours

• Follow to outcome

Pregnancy Registry (if applicable):

• If pregnancy registry exists for intervention, provide information to participant

• Encourage participation in registry (voluntary)

8.4 Unanticipated Problems (40 lines)

Definition of Unanticipated Problems Involving Risks (per OHRP/FDA):

An unanticipated problem involving risks to participants or others is any incident, experience, or outcome that meets ALL of the following criteria:

1. Unexpected: Not described in terms of nature, severity, or frequency in protocol, consent form, investigator brochure, or device labeling

2. Related or possibly related to participation in research

3. Suggests that research places participants or others at greater risk of harm than previously known or recognized

Examples:

• Cluster of similar AEs suggesting increased risk

• Single occurrence of serious injury that was not anticipated

• Protocol deviation that increases risk to participants

• Breach of confidentiality

• Loss of study drug/device

• Complaint from participant about research conduct raising safety concerns

• Error in informed consent process affecting participant rights/welfare

Not Unanticipated Problems:

• Individual SAEs that are expected and consistent with known risks (these are just SAEs, not UPs)

• Protocol deviations that do not increase risk or affect participant rights

Reporting:

• To IRB: Report to IRB within timeframe specified by IRB (typically 5-10 days)

• To Sponsor: Report to sponsor within 24 hours

• To FDA: Sponsor determines if FDA reporting required

Documentation:

• Description of unanticipated problem

• Analysis of how it meets UP criteria

• Actions taken to prevent recurrence

• Impact on research and participants

• Protocol amendments or corrective actions implemented

Section 9: STATISTICAL CONSIDERATIONS (600 lines)

Use sample size calculation results from Step 2 (waypoints/02_sample_size_calculation.json) throughout this section.

9.1 Statistical Hypotheses (30 lines):

For Primary Endpoint:

• Null hypothesis (H0): [State null hypothesis, typically "There is no difference between [intervention] and [control] in [primary endpoint]"]

• Alternative hypothesis (Ha or H1): [State alternative, typically "There is a difference between [intervention] and [control] in [primary endpoint]" for superiority, or "[Intervention] is non-inferior to [control]" for non-inferiority]

• One-sided vs two-sided testing: State whether testing is one-sided or two-sided (typically two-sided for superiority, one-sided for non-inferiority)

• Alpha level: 0.05 (two-sided) or 0.025 (one-sided)

If Multiple Primary Endpoints:

• State hypotheses for each

• Specify multiplicity adjustment (see Section 10.4.2)

For Key Secondary Endpoints (if formal testing planned):

• State hypotheses

• Specify if testing is conditional on primary endpoint significance

9.2 Sample Size Determination (150 lines):

Use exact values from waypoints/02_sample_size_calculation.json.

Paragraph 1: Primary Endpoint for Sample Size (15-20 lines):

• The sample size calculation is based on the primary efficacy endpoint: [Endpoint name from Section 4]

• Endpoint type: [Continuous/Binary/Time-to-event]

Paragraph 2: Detailed Assumptions and Calculations (80 lines):

For Continuous Endpoints:

• Primary endpoint: [Specify, e.g., "Mean change from baseline in systolic blood pressure at 6 months"]

• Expected mean difference between groups: [X] units [From calculation: effect_size]

  • Based on: [Cite 2-3 similar trials from Step 1 research that informed this assumption]

• Common standard deviation: [Y] units [From calculation: std_dev]

  • Based on: [Cite similar trials]

• Statistical test: Two-sample t-test or ANCOVA adjusting for baseline

• Two-sided alpha: [From calculation: alpha, typically 0.05]

• Power: [From calculation: power, typically 0.80 or 80%]

• Allocation ratio: 1:1 [or specify different ratio]

• Sample size formula: [Specify formula or software]

• Calculated sample size per arm: [n_per_arm from calculation]

• Total sample size (both arms): [n_total from calculation]

For Binary Endpoints:

• Primary endpoint: [Specify, e.g., "Proportion of participants achieving treatment success at 6 months"]

• Expected proportion in control group (P1): [From calculation: p1]

  • Based on: [Cite 2-3 similar trials]

• Expected proportion in treatment group (P2): [From calculation: p2]

  • Based on: [Cite similar trials]

• Absolute difference (P2 - P1): [Calculate and state]

• Statistical test: Chi-square test or Fisher's exact test

• Two-sided alpha: [From calculation: alpha, typically 0.05]

• Power: [From calculation: power, typically 0.80 or 80%]

• Allocation ratio: 1:1 [or specify different ratio]

• Sample size formula: [Specify formula or software]

• Calculated sample size per arm: [n_per_arm from calculation]

• Total sample size (both arms): [n_total from calculation]

For Time-to-Event Endpoints (if applicable):

• Primary endpoint: [Specify]

• Median time in control group: [M1] months

• Median time in treatment group: [M2] months

• Hazard ratio (HR): [Calculate from medians]

• Statistical test: Log-rank test

• Accrual period: [X] months

• Follow-up period: [Y] months

• Number of events required: [Calculated number]

• Calculated sample size per arm: [n_per_arm]

• Total enrollment target: [n_total]

Paragraph 3: Adjustment for Dropout (30-40 lines):

• Expected dropout rate: [From calculation: dropout, typically 15%]

  • Based on: [Cite similar trials showing dropout rates]

• Rationale: Account for participants lost to follow-up, withdrawals, and missing data

• Adjusted sample size per arm: [n_per_arm] / (1 - [dropout]) = [adjusted_n_per_arm]

• Total enrollment target: [adjusted_n_total]

Formula shown:

Adjusted N = N / (1 - dropout rate)
Adjusted N per arm = [n_per_arm] / (1 - [dropout]) = [adjusted_n_per_arm]
Total enrollment: [adjusted_n_total]

Paragraph 4: Sensitivity Analyses (40 lines):

• To assess robustness of sample size, calculations performed under alternative assumptions:

Alternative Scenario 1: Lower effect size (20% reduction):

• Effect size: [X * 0.8] [or P2-P1 reduced by 20%]

• Calculated N per arm: [Calculate alternative N]

• Interpretation: [State whether study still adequately powered]

Alternative Scenario 2: Higher effect size (20% increase):

• Effect size: [X * 1.2] [or P2-P1 increased by 20%]

• Calculated N per arm: [Calculate alternative N]

• Interpretation: [State increased power]

Alternative Scenario 3: Higher variance (20% increase in SD):

• Standard deviation: [Y * 1.2]

• Calculated N per arm: [Calculate alternative N]

• Interpretation: [State impact on power]

Conclusion: The proposed sample size is robust to moderate deviations from assumptions. If actual effect size or variance differs substantially from assumptions, study may be underpowered or overpowered.

Paragraph 5: Power for Secondary Endpoints (20 lines):

• Sample size is powered for the primary efficacy endpoint only

• Secondary endpoints:

  • Expected to have [greater/similar/lower] effect sizes

  • Study may have [sufficient/insufficient] power to detect clinically meaningful differences in secondary endpoints

  • If power is limited, secondary endpoint analyses may be considered exploratory and hypothesis-generating

• Acknowledge that study is not formally powered for multiple comparisons across all secondary endpoints

Paragraph 6: Justification and Biostatistician Review (40 lines):

• Sample size assumptions informed by similar trials:

  • [NCT number and citation 1]

  • [NCT number and citation 2]

  • [NCT number and citation 3]

• Sample size calculation performed using [software, e.g., "nQuery version X", "PASS version X", "R statistical software", "SAS PROC POWER"]

• IMPORTANT DISCLAIMER: "Sample size calculations are preliminary and subject to biostatistician review and validation prior to study initiation. Final sample size may be adjusted based on updated assumptions or regulatory feedback."

9.3 Populations for Analyses (50 lines):

Define each analysis population clearly:

Intent-to-Treat (ITT) Population:

• Definition: All randomized participants, analyzed according to treatment assignment as randomized, regardless of treatment actually received

• Inclusions: All participants who undergo randomization

• Exclusions: None

• Use: Primary population for efficacy analyses (conservative approach, preserves randomization)

Modified Intent-to-Treat (mITT) Population (if applicable):

• Definition: All randomized participants who received at least one dose of study intervention

• Inclusions: Randomized participants with ≥1 dose

• Exclusions: Randomized but never dosed (e.g., consent withdrawal before dosing)

• Use: May be primary efficacy population if specified; supportive efficacy analysis

• Rationale: [If using mITT as primary, justify - e.g., "Participants never dosed have no opportunity to benefit"]

Per-Protocol (PP) Population:

• Definition: All participants who completed study per protocol without major protocol deviations

• Inclusions: Participants who: Received ≥[X]% of planned intervention doses (e.g., ≥80%), Completed all required efficacy assessments, Did not have major protocol deviations affecting efficacy

• Exclusions: Major protocol deviations (specify what constitutes major)

• Use: Supportive efficacy analysis (provides estimate of treatment effect under ideal adherence)

Safety Population:

• Definition: All participants who received at least one dose (or partial dose) of study intervention

• Analyzed according to treatment actually received (as-treated)

• Inclusions: All participants with ≥1 dose

• Exclusions: Randomized but never dosed

• Use: Primary population for all safety analyses

Evaluable Population (if applicable):

• Definition: Participants with adequate data for specific endpoint assessment

• Inclusions: Participants with valid baseline and post-baseline measurements for specific endpoint

• Use: Sensitivity analysis for specific endpoints

Primary Population Specification:

• Primary efficacy analyses: ITT population [or mITT if specified]

• Safety analyses: Safety population

• Supportive efficacy analyses: PP population

9.4 Statistical Analyses (400 lines):

CRITICAL: Generate ALL subsections 9.4.1-9.4.9 with detailed methods.

9.4.1 General Approach (30 lines):

Descriptive Statistics:

• Categorical: n (%)

• Continuous: mean ± SD, median, range/IQR, min, max

Inferential Statistics:

• Alpha: 0.05 (two-sided unless specified)

• 95% confidence intervals for treatment effects

• Missing data per Section 10.4.10

Statistical Software: [R/SAS/Stata version X.X]

Pre-Specified Covariates: Age, sex, baseline primary endpoint value, stratification factors, [other relevant covariates]

Assumptions: Normality (Shapiro-Wilk, Q-Q plots), homogeneity of variance (Levene's), proportional hazards (Schoenfeld residuals). If violated: transformations or nonparametric tests.

Multiplicity: Primary endpoint - no adjustment. Secondary endpoints: [Specify adjustment method or state exploratory]

9.4.2 Analysis of the Primary Efficacy Endpoint(s) (50 lines):

For EACH primary endpoint, generate:

Endpoint: [Name from Section 3] Definition: [Full definition] Timing: [Timepoint] Scale: [Continuous/Binary/Time-to-event, units] Population: Intent-to-Treat (ITT) [or mITT]

Statistical Procedure:

If Continuous:

• ANCOVA: Y = Treatment + Baseline_Value + Age + Sex + [Stratification_Factors] + Error

• Least squares means, treatment effect (difference), 95% CI, two-sided p-value, effect size

If Binary:

• Logistic regression: logit(P[success]) = Intercept + Treatment + [Covariates]

• OR with 95% CI, p-value (Wald/likelihood ratio), risk difference, relative risk

• Alternative: Chi-square or Fisher's exact (if counts <5)

If Time-to-Event:

• Cox regression: h(t) = h0(t) × exp(Treatment + [Covariates])

• HR with 95% CI, p-value, Kaplan-Meier curves, log-rank test, median time-to-event

• Check proportional hazards (Schoenfeld residuals)

Diagnostics: Check assumptions, examine residuals, identify outliers

Presentation: Table (statistics, effect, CI, p-value), Figure (mean change/proportion/KM curves)

Sensitivity Analyses: Per-protocol, complete case, alternative missing data methods, unadjusted

If Multiple Primary Endpoints: [Specify multiplicity adjustment method or hierarchical testing order]

9.4.3 Analysis of the Secondary Endpoint(s) (40 lines):

For EACH secondary endpoint (2-5 endpoints):

Secondary Endpoint [N]: [Name]

• Definition: [Full definition]

• Scale: [Continuous/Binary/Time-to-event]

• Population: ITT

• Method: [ANCOVA/Logistic/Cox regression - similar to 10.4.2]

• Treatment effect: [Mean difference/OR/HR with 95% CI], p-value

• Presentation: Table, figure

[Repeat for each secondary endpoint]

Multiplicity Strategy: [Hierarchical testing (specify order) OR No adjustment (exploratory) OR Bonferroni/Holm adjustment]

9.4.4 Safety Analyses (60 lines):

Analysis Population: Safety population (all participants with ≥1 dose)

Adverse Events (AEs):

• MedDRA Coding: All AEs coded using MedDRA version [X.X] (System Organ Class, Preferred Term)

• Summary tables: Frequency (n) and percentage (%) of participants with:

  • Any AE

  • Treatment-related AEs (Possible, Probable, or Definite relationship)

  • Grade 3-4 AEs (by CTCAE grading)

  • Serious AEs (SAEs)

  • AEs leading to dose reduction

  • AEs leading to treatment discontinuation

  • Fatal AEs (Grade 5)

• AEs by System Organ Class and Preferred Term: Sorted by decreasing frequency

• Denominator: Number of participants in safety population

• Comparison between arms: Descriptive only (no formal statistical testing), but may calculate risk difference if clinically relevant

Serious Adverse Events (SAEs):

• Separate detailed table for SAEs

• Frequency by SOC and PT

• Relationship to study intervention

• Outcomes: Resolved, ongoing, fatal, resolved with sequelae

AEs Leading to Discontinuation:

• Separate table

• By SOC and PT

• Treatment group comparison

Deaths:

• Individual participant listings with narratives

• Cause of death

• Relationship to study intervention

• Timing relative to last dose

Laboratory Data:

• Shift tables: Baseline CTCAE grade → Worst post-baseline CTCAE grade

  • For key lab parameters: Hemoglobin, ANC, platelets, creatinine, ALT, AST, bilirubin

• Potentially Clinically Significant (PCS) values: Define thresholds for each parameter

  • Example: Hemoglobin <8 g/dL, ANC <1.0 × 10^9/L, ALT >3× ULN

• Frequency of PCS values by treatment group

• Summary statistics: Mean, SD, change from baseline for continuous lab values

Vital Signs:

• Summary statistics at each time point: Mean, SD

• Change from baseline: Mean change, 95% CI

• Outliers: Frequency of values outside normal ranges (e.g., SBP >180 mmHg, HR >120 bpm)

ECG/Cardiac Assessments (if applicable):

• QTc interval: Mean, SD, change from baseline, frequency of QTc >480 ms or increase >60 ms

• LVEF: Mean, SD, change from baseline, frequency of LVEF decline >10% or LVEF <50%

Device-Related Complications (for devices):

• Device malfunctions: Frequency and type

• Device-related infections

• Need for device removal or revision

• Comparison to predicate device rates (if available)

9.4.5 Baseline Descriptive Statistics (30 lines):

Table 1: Baseline Characteristics by Treatment Arm

Purpose: Demonstrate comparability of treatment groups at baseline

Population: ITT population

Variables included:

• Demographics:

  • Age: Mean ± SD, median, range

  • Sex: n (%) Male, Female

  • Race: n (%) for each category (White, Black/African American, Asian, Other)

  • Ethnicity: n (%) Hispanic/Latino, Not Hispanic/Latino

• Disease Characteristics:

  • Disease duration: Mean ± SD, median

  • Disease stage/severity: n (%) for each category

  • Baseline primary endpoint value: Mean ± SD, median

• Prior Treatments:

  • Prior [specific treatment 1]: n (%)

  • Prior [specific treatment 2]: n (%)

  • Number of prior treatments: Mean ± SD, median

• Baseline Efficacy Measures:

  • Baseline values for primary and key secondary endpoints: Mean ± SD for continuous, n (%) for categorical

• Baseline Laboratory Values:

  • Key lab parameters: Mean ± SD

Presentation:

• Three columns: Intervention Arm, Control Arm, Total

• Continuous variables: Mean ± SD, median (range)

• Categorical variables: n (%)

• NO statistical testing between arms: Baseline comparisons are descriptive only; no p-values presented

• Rationale: Randomization ensures balance; hypothesis testing at baseline is inappropriate

9.4.6 Planned Interim Analyses (40 lines):

If Interim Analyses Planned:

Purpose: [Efficacy/futility/safety monitoring] Timing: [X% enrollment/X months/X events] Number: [1, 2, or more] DSMB: Conducts analyses; investigators/sponsor blinded; recommends continue/modify/terminate

Methods:

Efficacy: Group sequential design, O'Brien-Fleming or Lan-DeMets alpha spending, overall alpha = 0.05 maintained Futility: Conditional power approach; stop if < [threshold, e.g., 20%] Safety: DSMB reviews unblinded data; no formal boundary

Interim Analysis Plan:

Impact: CIs and p-values adjusted for multiple looks; estimates adjusted for early stopping bias if applicable

Reporting: If early stop, report reason, results, analyses conducted

If NO Interim Analyses:

• No formal efficacy/futility analyses; study continues to full enrollment unless safety concerns

• Safety Monitoring: Investigators (continuous); DSMB/SMC reviews aggregate data [frequency]; DSMB may recommend termination; operates under charter (Section 10.6)

9.4.7 Sub-Group Analyses (50 lines):

Purpose: Explore consistency of treatment effect across participant subgroups (hypothesis-generating, not confirmatory)

Pre-Specified Subgroups:

1. Age: <65 years vs ≥65 years

2. Sex: Male vs Female

3. Race/Ethnicity: [Specify categories based on enrollment, e.g., White vs Non-White]

4. Baseline disease severity: [Define categories, e.g., Mild vs Moderate/Severe]

5. Disease-specific subgroups:

   • [e.g., Biomarker status: Positive vs Negative]

   • [e.g., Disease stage: Early vs Advanced]

   • [e.g., Prior treatment: Yes vs No]

6. [Additional pre-specified subgroups relevant to intervention and indication]

Analysis Method:

Interaction Tests:

• Test for subgroup × treatment interaction in regression model

• Model: Outcome = Treatment + Subgroup + Treatment×Subgroup + [Covariates]

• P-value for interaction term indicates whether treatment effect differs between subgroups

• Significance level: P < 0.10 considered suggestive of interaction (more lenient than 0.05 due to low power)

Forest Plots:

• Treatment effect estimate (mean difference, OR, or HR) with 95% CI for each subgroup

• Overall treatment effect for comparison

• Visual assessment of consistency

Interpretation:

• Subgroup analyses are exploratory and hypothesis-generating

• Not adjusted for multiplicity (multiple subgroups tested)

• Interpret with caution: Risk of false positives

• Consistency assessment: Determine if treatment effect is generally consistent across subgroups

• If interaction p < 0.10: Describe difference and note exploratory nature

Reporting:

• Subgroup analyses reported in exploratory analyses section

• Clearly labeled as exploratory

• Limitations acknowledged

9.4.8 Tabulation of Individual Participant Data (20 lines):

All individual participant data will be listed in appendices (not summarized in tables).

Listings Include:

1. Demographics: Individual participant demographics (ID, age, sex, race, ethnicity, study arm)

2. Adverse Event Listings:

   • All AEs: Every AE for every participant (ID, AE term, start/stop date, severity, relationship, outcome)

   • SAEs: Separate listing for serious AEs

   • AEs Leading to Discontinuation: Separate listing

   • Deaths: Separate listing with narratives

3. Concomitant Medications: All medications taken during study (ID, medication name, dose, route, start/stop date, indication)

4. Laboratory Data: All lab values at all time points (ID, parameter, value, unit, date, normal range, CTCAE grade)

5. Efficacy Endpoints: Individual endpoint values at all time points (ID, endpoint, value, date)

6. Protocol Deviations: All deviations (ID, deviation description, date, impact)

Sorting: Listings sorted by treatment group, then subject ID, then visit or date

Purpose:

• Support summary tables and analyses

• Available for regulatory inspection

• Enable identification of individual participant patterns

9.4.9 Exploratory Analyses (40 lines):

Purpose: Generate hypotheses for future studies; not powered or adjusted for multiplicity

Exploratory Analyses Include:

1. Exploratory Endpoints (from Section 4):

   • [List specific exploratory endpoints]

   • Descriptive statistics and treatment comparisons

   • No formal hypothesis testing

2. Correlation Analyses:

   • Correlation between biomarkers and efficacy endpoints

   • Example: "Assess correlation between [biomarker X] levels and primary endpoint response"

   • Methods: Pearson or Spearman correlation, scatter plots

3. Post-Hoc Subgroup Analyses:

   • Subgroups not pre-specified in Section 10.4.7

   • Generated based on emerging data patterns

   • Clearly labeled as post-hoc

4. Responder Analyses:

   • Define responder criteria: [e.g., "Participants achieving ≥50% improvement in primary endpoint"]

   • Proportion of responders in each treatment arm

   • Characteristics of responders vs non-responders

5. Dose-Response Analyses (if multiple dose levels):

   • Assess relationship between dose and efficacy

   • Assess relationship between dose and safety

6. Longitudinal Analyses:

   • Trajectory of endpoints over time

   • Mixed models for repeated measures

   • Growth curve analyses

Methods:

• Descriptive statistics predominant

• Exploratory modeling

• Graphical displays

Multiplicity:

• No adjustment for multiplicity

• Purely hypothesis-generating

Interpretation:

• Results interpreted with caution

• Acknowledge exploratory nature and risk of false positives

• Use to inform future research questions

9.4.10 Missing Data Handling (30 lines):

Primary Approach (choose based on endpoint/design):

MMRM (Mixed Model for Repeated Measures): For continuous endpoints with multiple timepoints; includes all data without imputation; MAR assumption Multiple Imputation: MICE or similar; 20-50 imputations; combined via Rubin's rules; MAR assumption LOCF: Generally NOT recommended unless field standard; assumes no change after dropout

Rationale: [Justify method based on expected patterns and literature]

Sensitivity Analyses:

1. Complete Case: Observed outcomes only (MCAR assumption)

2. Worst-Case: Intervention = worst, Control = best (conservative)

3. Best-Case: Intervention = best, Control = worst (optimistic)

4. Tipping Point: Identify assumptions that would change conclusions

5. Pattern-Mixture: Model outcomes by missingness pattern (if potentially informative)

Missing Covariates: Complete case (if few) OR multiple imputation

Dropouts: Compare time to discontinuation (KM, log-rank), tabulate reasons, assess if differential dropout suggests informative missingness

Reporting: Describe patterns, report n (%) missing, reasons by arm, sensitivity results, impact on conclusions

Section 10: SUPPORTING DOCUMENTATION AND OPERATIONAL CONSIDERATIONS (400 lines)

10.1 Informed Consent Process (50 lines):

Documents: ICF for adults; assent for minors (if <18 years); optional consents (future specimen/data use, genetic research)

Procedures:

• Who: PI or qualified trained designee

• When: Before study-specific procedures; adequate time to review

• Language: Participant's native language; 8th grade reading level; avoid jargon

• Process: Verbal explanation, written ICF, time to review (≥24 hrs if possible), questions answered, emphasize voluntary participation and withdrawal rights

Key Elements (21 CFR 50.25): Research purpose, duration, procedures, experimental nature, risks/discomforts, benefits, alternatives, confidentiality, compensation (if applicable), voluntary participation, withdrawal rights, contact information

Documentation: Signed/dated by participant (or LAR) and person obtaining consent; original in study records, copy to participant, documented in medical record

Re-Consent: If protocol amended with new risks, re-consent required; amended ICF submitted to IRB; currently enrolled and new participants sign updated form

Compliance: 21 CFR Part 50, ICH-GCP E6(R2), IRB requirements, HIPAA Authorization (US studies)

10.2 Study Discontinuation and Closure (50 lines):

Circumstances for Early Study Termination:

1. Unacceptable safety profile: SAE rate or severity exceeds acceptable risk

2. Futility: Interim analysis demonstrates low probability of achieving study objectives

3. External data: New data from other studies renders study obsolete or unethical

4. Regulatory authority decision: FDA or other regulatory body requests study termination

5. Sponsor decision: Business, financial, or strategic reasons

6. Inadequate enrollment: Unable to enroll sufficient participants within reasonable timeframe

Procedures Upon Early Termination:

Notification:

• Notify FDA (for IND/IDE studies): Within required timeframe per regulations

• Notify IRB: Immediately, with reason for termination

• Notify participants: Letter sent to all enrolled participants explaining termination

• Notify investigators: All sites notified

• Update ClinicalTrials.gov: Update study status to "Terminated" with reason

Safety Follow-Up:

• Enrolled participants: Continue safety follow-up per protocol or abbreviated schedule

• Ongoing AE monitoring: AEs followed to resolution or stabilization

• Final safety assessment: Encourage participants to complete final safety visit

Data Collection:

• Collect all data up to termination point

• Complete data entry for available data

• Perform final database lock

Analysis and Reporting:

• Analyze all available data

• Final study report prepared

• Report submitted to FDA and IRB

• Results posted on ClinicalTrials.gov (per 42 CFR Part 11 if applicable)

Study Closure (Planned Completion):

• Last participant completes last visit

• All data collected and entered

• All queries resolved

• Database locked

• Statistical analysis performed

• Final study report prepared

• Close-out visits at all sites

• Study records archived

10.3 Confidentiality and Privacy (40 lines):

HIPAA Compliance (if US study):

• Participants sign HIPAA Authorization for use and disclosure of Protected Health Information (PHI)

• Authorization covers: Use of PHI for research, disclosure to sponsor, disclosure to regulatory authorities, disclosure for safety reporting

• Authorization specifies: Types of PHI to be used, who may use/disclose PHI, purpose, expiration (or statement that no expiration), right to revoke

Data Protection:

• Deidentified data: Participants assigned unique subject identification numbers (e.g., "001-001")

• No personal identifiers in study database: No names, addresses, dates of birth (only age), medical record numbers

• Secure data storage: Electronic data stored on password-protected, encrypted servers

• Physical records: Kept in locked filing cabinets in secure area

• Limited access: Only authorized study personnel have access to identifiable data

• Site master log: Links subject ID to identifiable information; kept separately in locked location

Data Transfer:

• Data transmitted to sponsor via secure encrypted transmission

• No PHI included in routine data transmissions

• Identifiable data (e.g., for SAE reporting) sent via secure methods

GDPR Compliance (if EU participants or sites):

• Data processing per General Data Protection Regulation (GDPR) requirements

• Data Processing Agreement between sponsor and sites

• Participant rights: Access to data, rectification, erasure (right to be forgotten), data portability, restriction of processing

• Legal basis for processing: Participant consent, legal obligations (regulatory requirements)

• Data retention: Specify retention period (per ICH-GCP: minimum 2 years after last approval or longer per regulations)

• Data Protection Officer: Appointed if required

Confidentiality Breach:

• Report to sponsor, IRB, and participants (if identifiable)

• Mitigation measures implemented

• Report to relevant authorities (OCR for HIPAA, supervisory authority for GDPR)

10.4 Future Use of Stored Specimens and Data (30 lines):

Optional Consent for Future Research:

• Separate section in ICF or separate consent form

• Participants may consent to study but decline future use

• Voluntary: No impact on study participation if declined

Future Research Uses:

• Biospecimens (blood, tissue, other samples) may be stored for future research

• Types of future research: Unspecified at this time; may include genetic/genomic research, biomarker discovery, other exploratory research

• Identifiability: Specimens coded or deidentified

• Sharing: Specimens may be shared with other researchers (academic, commercial)

• Commercial use: Potential for commercial use; no financial benefit to participant

• Results: Future research results not returned to participants (unless clinically actionable and IRB approved)

Storage:

• Duration: Specify (e.g., "Specimens stored for up to 15 years" or "indefinitely")

• Location: Biorepository or sponsor facility

• Destruction: Participant may request specimen destruction at any time

Re-Contact:

• Participant may be re-contacted for additional studies if consented

• Contact information kept separately from specimens

Withdrawal:

• Participant may withdraw consent for future use at any time

• Specimens destroyed upon request

• Data already generated from specimens may be retained

10.5 Key Roles and Study Governance (40 lines):

Sponsor: Overall oversight; regulatory submissions (IND/IDE, safety/annual reports, amendments); funding; monitoring/QA; data management (eCRF, database); safety reporting; study supplies; site support; closeout (final report, results posting, archiving)

Principal Investigator (PI): Conduct study per protocol; regulatory compliance (21 CFR 50, 56, 312/812, ICH-GCP); IRB submissions/approvals; informed consent; participant safety/AE management; SAE reporting (sponsor within 24 hrs, IRB per timeline); data accuracy; staff supervision/training; study supplies accountability; available for FDA inspections; protocol deviation reporting

Medical Monitor (if applicable): Safety oversight; SAE review/causality; dosing decisions; stopping rule evaluation (if no DSMB); protocol questions; emergency unblinding authority; safety report assistance

Biostatistician: Write SAP; review/validate sample size; generate randomization schedule; perform analyses; interim analyses (if planned); prepare statistical sections; support regulatory submissions

Steering Committee (if applicable): Protocol design/amendments; study oversight; publication/authorship decisions; members: PI, sponsor, KOLs, biostatistician

DSMB: See Section 10.6

10.6 Safety Oversight (50 lines):

For DEVICES (21 CFR 812):

IDE Classification:

• Significant Risk (SR): Implants, life-support, or substantial diagnostic/therapeutic importance → Full IDE (FDA+IRB approval, investigational labeling, annual FDA reports, safety reports)

• Non-Significant Risk (NSR): Doesn't meet SR criteria → Abbreviated IDE (IRB approval only, FDA may audit)

Approval Process:

• SR: IDE to FDA (30-day review) → IRB approval → Study begins

• NSR: IRB approval with NSR determination → Study begins

Device Classification: Class III (usually SR, requires PMA); Class II (SR or NSR); Class I (usually exempt unless SR)

Predicate Safety (510(k) pathway): Leverage predicate device safety data, justify similar profile

For DRUGS (21 CFR 312):

IND Requirements:

• Phase [2 or 3]; IND submission to FDA (30-day review, may proceed if no hold)

• IRB approval, informed consent (21 CFR 50)

• Safety reporting: 15-day (unexpected related SAEs), 7-day (fatal/life-threatening)

• Annual reports: Safety, progress, amendments, IB updates

For ALL STUDIES:

DSMB/SMC:

When Required: High-risk (mortality/morbidity endpoints), serious toxicities, vulnerable populations, blinded trials with interim analyses, large multi-site, NIH-funded (risk-based)

When Not Required: Low-risk, short duration, small studies with internal monitoring

DSMB Structure: 3-7 independent experts (clinical, biostatistician, ethicist); operates under formal charter

Meetings: [Quarterly/semi-annually/annually]; closed sessions (unblinded data), open sessions (blinded data)

Responsibilities: Review unblinded safety/efficacy, assess benefit-risk, evaluate stopping rules, recommend continue/modify/terminate

If No DSMB: Internal monitoring by medical monitor/sponsor; periodic safety review; IRB reviews at continuing review

Stopping Rules:

• Safety: Excess SAEs ([threshold-fold increase]), specific Grade 4 toxicities (>[X]%), deaths exceed threshold

• Futility: Conditional power <20% (Section 10.4.6)

• Efficacy: Crosses pre-specified boundary (Section 10.4.6)

10.7 Clinical Monitoring (40 lines):

Monitoring Plan:

• Monitoring approach: On-site monitoring and/or central monitoring

• Risk-based approach: Monitoring intensity based on risk level per FDA guidance "Oversight of Clinical Investigations — A Risk-Based Approach to Monitoring"

Monitoring Frequency:

• Site initiation visit: Before first participant enrolled

• Periodic monitoring visits: Frequency based on risk and enrollment rate

  • High-risk: Monthly or quarterly

  • Moderate-risk: Quarterly or semi-annually

  • Low-risk: Semi-annually or annually

• Close-out visit: After last participant completes study

Monitoring Activities:

• Regulatory document review: IRB approval, ICFs, CV of PI, training documentation, delegation log

• Informed consent verification: Review process, verify signed consents for sample of participants

• Source data verification (SDV): Compare eCRF data to source documents for accuracy

  • 100% SDV for key data points (primary endpoint, SAEs, eligibility, informed consent)

  • Targeted SDV for other data points (sample of participants)

• Inclusion/exclusion criteria verification: Confirm eligibility for sample of participants

• AE reporting verification: Confirm SAEs reported to sponsor and IRB within required timeframes

• Study intervention accountability: Verify drug/device accountability records

• Protocol compliance: Identify protocol deviations

• Data quality: Review data completeness and quality

Monitoring Reports:

• Monitor prepares monitoring visit report after each visit

• Report submitted to sponsor

• Site receives findings and responds to any issues

• Follow-up on corrective actions

Central Monitoring:

• Remote data review: Real-time review of eCRF data for quality, consistency, and patterns

• Statistical methods: Detect outliers, unusual patterns, potential fraud

• Risk-based: Focus on key data points

10.8 Quality Assurance and Quality Control (40 lines):

Quality Assurance (QA) Procedures:

Standard Operating Procedures (SOPs):

• Sponsor SOPs: Cover all aspects of study conduct (monitoring, data management, safety reporting, etc.)

• Site SOPs: Cover site-specific procedures (informed consent, AE reporting, lab processing, etc.)

• Training on SOPs required

Training:

• Investigator meeting: All investigators and coordinators attend site initiation meeting

• Protocol training: Detailed training on protocol procedures, eligibility, endpoints, safety

• eCRF training: Training on electronic data capture system

• GCP training: ICH-GCP training required for all study staff

• Device/intervention training: Training on intervention administration (especially for devices)

Audits:

• Sponsor audits: Periodic audits of sites by sponsor QA department

• Regulatory inspections: FDA inspections (for cause or routine)

• Audit preparedness: Sites maintain inspection-ready study files

Quality Control (QC) Procedures:

Data Quality:

• Edit checks: Automated range checks, consistency checks, logic checks in eCRF

• Query management: System generates queries for out-of-range or inconsistent data

• Query resolution: Site responds to queries, provides corrections or explanations

• Data review: Regular data review meetings

Protocol Compliance:

• Protocol deviation tracking: All deviations documented and tracked

• Corrective and preventive actions (CAPA): Implement CAPA for recurring deviations

Quality Metrics:

• Define quality metrics: Query rates, deviation rates, enrollment rates, monitoring findings

• Track metrics over time

• Review at study team meetings

Oversight:

• Sponsor quality management

• Regular study team meetings to review quality

10.9 Data Handling and Record Keeping (60 lines):

Data Collection and Management:

Electronic Data Capture (EDC):

• eCRF system: [Specify system name, e.g., "Medidata Rave", "REDCap", or state "TBD"]

• Web-based access: Sites access eCRF via secure web browser

• User access: Role-based access control; unique usernames and passwords

• Training: Sites trained on eCRF before use

CRF Completion:

• Timeliness: Data entered within [X] days of visit (e.g., 3-5 business days)

• Accuracy: Data transcribed from source documents accurately

• Completeness: All required fields completed

• Queries: Respond to data queries within [X] days (e.g., 5 business days)

Query Resolution:

• System generates automated queries for missing or inconsistent data

• Monitor generates manual queries during monitoring visits

• Site responds: Correct data or provide explanation

• Query closure: Monitor or data manager closes query after review

Data Quality Checks:

• Real-time edit checks during data entry

• Batch edit checks run periodically

• Medical review: Medical monitor reviews safety data

• Statistical review: Biostatistician reviews data for patterns and outliers

Data Management:

• Database design: Sponsor data management team designs database

• Database testing: User acceptance testing before go-live

• Database lock: Database locked for analysis after all queries resolved and monitoring complete

• Data cleaning: Iterative process until database clean

Source Documentation:

Definition of Source Documents:

• Original records of clinical findings, observations, or activities

• Examples: Medical records, progress notes, lab reports, imaging reports, ECG printouts, device interrogation reports, study-specific worksheets

Source Data Verification (SDV):

• Monitor compares eCRF data to source documents

• Verify accuracy, completeness, and consistency

• 100% SDV for key data points; targeted SDV for others

Source Document Requirements:

• Contemporaneous: Recorded at the time of the event

• Original: First recording of information

• Accurate and complete

• Legible

• Attributable: Signed and dated by person performing/recording procedure

• Traceable to participant: Subject ID included

Data Security:

Electronic Systems:

• Secure servers: Data stored on secure, password-protected servers

• Encryption: Data encrypted during storage and transmission

• Access controls: Role-based access; unique user IDs

• Audit trails: 21 CFR Part 11 compliant; system logs all data entries, modifications, deletions with user ID and timestamp

• Backups: Regular automated backups; disaster recovery plan

Physical Records:

• Locked storage: Source documents and regulatory documents stored in locked cabinets

• Restricted access: Only authorized site personnel have access

Study Records Retention:

Duration:

• Per ICH-GCP: Minimum 2 years after last approval of marketing application or longer per local regulations

• Per FDA (devices, 21 CFR 812.140): 2 years after investigation terminated or concluded

• Per FDA (drugs, 21 CFR 312.62): 2 years after marketing application approved or 2 years after IND closed

• Sponsor specifies exact retention period (typically 15-25 years)

What to Retain:

• Regulatory documents: Protocol, amendments, ICF versions, IRB approvals, FDA correspondence

• Source documents: Medical records, lab reports, imaging

• CRFs: Electronic or paper CRFs

• Monitoring reports

• Safety reports

• Final study report

Archiving:

• After study closure, records archived per sponsor procedures

• Sites notified when records may be destroyed

10.10 Protocol Deviations (30 lines):

Definition:

• Protocol deviation: Any departure from protocol procedures, inclusion/exclusion criteria, or good clinical practice

Classification:

Major Deviation:

• Affects participant safety, rights, or welfare

• Affects data integrity or study results

• Examples: Enrollment of ineligible participant, administration of wrong dose, failure to obtain informed consent, failure to report SAE

Minor Deviation:

• Does not affect safety, rights, or data integrity

• Examples: Visit conducted outside window (if clinically acceptable), minor documentation error

Reporting:

• To sponsor: All deviations reported to sponsor

  • Major deviations: Immediately (within 24 hours)

  • Minor deviations: Routine reporting (with data transmission)

• To IRB: Major deviations reported to IRB per IRB requirements (typically within 5-10 days)

• Documentation: Deviation log maintained at site

Management:

• Root cause analysis: Investigate cause of deviation

• Corrective action: Implement corrective action to prevent recurrence

• Preventive action: Identify and address potential deviations before they occur

• Monitor trends: Track deviation rates; identify systemic issues

Impact on Analysis:

• Per-protocol population: Major deviations may exclude participant from PP population (see Section 10.3)

10.11 Publication and Data Sharing Policy (40 lines):

Authorship:

• Per International Committee of Medical Journal Editors (ICMJE) criteria:

  1. Substantial contributions to conception/design, data acquisition/analysis/interpretation

  2. Drafting or revising manuscript critically for intellectual content

  3. Final approval of version to be published

  4. Agreement to be accountable for all aspects

• Investigators and key contributors eligible for authorship

• Sponsor representatives may be co-authors if meet ICMJE criteria

Publication Timeline:

• Primary manuscript: Target within 12 months of study completion

• Presentations: Results presented at scientific meetings (oral or poster)

• Secondary manuscripts: Exploratory analyses, subgroup analyses

Review and Approval:

• Sponsor review: Sponsor reviews manuscript to ensure accuracy and protect confidential information

• Review period: [X] days (e.g., 30-60 days)

• Sponsor may request delays for patent filing or confidentiality

Authorship Disputes:

• Resolved by steering committee or PI

Data Sharing:

• Individual Participant Data (IPD): Deidentified IPD may be available upon reasonable request after publication

• Who can access: Qualified researchers

• What data: Deidentified data at participant level

• When: After publication of primary manuscript (typically 6-12 months after publication)

• How: Data use agreement required; request submitted to sponsor

• Why: For meta-analyses, secondary analyses (must specify research question)

• Embargo period: [X] months after publication before data shared

ClinicalTrials.gov:

• Results posted per 42 CFR Part 11 (if applicable): Within 1 year of primary completion date

• Includes: Participant flow, baseline characteristics, primary and secondary outcome measures, adverse events

10.12 Conflict of Interest Policy (30 lines):

Financial Disclosure:

• Per 21 CFR Part 54 (Financial Disclosure by Clinical Investigators)

• All investigators and sub-investigators disclose financial interests

Disclosable Financial Interests:

1. Compensation to investigator where any payment is tied to study outcome

2. Proprietary interest in the product (patent, trademark, copyright, licensing)

3. Equity interest in sponsor (exceeding $50,000)

4. Significant payments of other sorts (exceeding $25,000 annually for services not directly related to study)

Disclosure Process:

• Investigators complete financial disclosure form annually and at study completion

• Submit to sponsor

• Sponsor maintains disclosure records

• Sponsor submits to FDA with marketing application

Conflict Management:

• Sponsor reviews disclosures

• If potential COI identified: Assess whether COI could affect data integrity

• Mitigation strategies:

  • Increased monitoring

  • Independent data verification

  • Exclude investigator from study

  • Disclose COI to IRB and in publications

• IRB may review and approve or require additional mitigation

10.13 Additional Considerations (20-40 lines):

Study-Specific Operational Details:

• [Include any study-specific operational considerations not covered elsewhere]

• [Example: Special equipment requirements, special laboratory procedures, biospecimen handling, imaging core lab procedures, centralized assessments]

Emergency Procedures (if applicable):

• Emergency contact information for study team (24/7)

• Emergency unblinding procedures (reference Section 7.4)

• Management of medical emergencies related to intervention (reference Section 7.2 or 9.3)

Special Equipment or Facility Requirements:

• [Specify if study requires special equipment, facilities, certifications]

• [Example: "Sites must have capability for 24-hour monitoring for first infusion"]

10.14 Abbreviations (30 lines):

Generate comprehensive abbreviation list in alphabetical order.

Include all abbreviations used in protocol.

Format: Abbreviation | Full Term

Example abbreviations (generate complete list specific to protocol):

• AE: Adverse Event

• ALT: Alanine Aminotransferase

• ANCOVA: Analysis of Covariance

• ANC: Absolute Neutrophil Count

• AST: Aspartate Aminotransferase

• BP: Blood Pressure

• CFR: Code of Federal Regulations

• CI: Confidence Interval

• CrCl: Creatinine Clearance

• CRF: Case Report Form

• CT: Computed Tomography

• CTCAE: Common Terminology Criteria for Adverse Events

• DSMB: Data Safety Monitoring Board

• ECG: Electrocardiogram

• eCRF: Electronic Case Report Form

• EDC: Electronic Data Capture

• FDA: Food and Drug Administration

• GCP: Good Clinical Practice

• HIPAA: Health Insurance Portability and Accountability Act

• HR: Hazard Ratio

• ICF: Informed Consent Form

• ICH: International Council for Harmonisation

• IDE: Investigational Device Exemption

• IFU: Instructions for Use

• IND: Investigational New Drug

• IRB: Institutional Review Board

• ITT: Intent-to-Treat

• IV: Intravenous

• LVEF: Left Ventricular Ejection Fraction

• MedDRA: Medical Dictionary for Regulatory Activities

• MMRM: Mixed Model for Repeated Measures

• MRI: Magnetic Resonance Imaging

• OR: Odds Ratio

• PD: Pharmacodynamics

• PI: Principal Investigator

• PK: Pharmacokinetics

• PMA: Premarket Approval

• PP: Per-Protocol

• PRO: Patient-Reported Outcome

• QTc: Corrected QT Interval

• RCT: Randomized Controlled Trial

• SAE: Serious Adverse Event

• SD: Standard Deviation

• SOC: System Organ Class

• SoA: Schedule of Activities

• SOP: Standard Operating Procedure

• SUSAR: Suspected Unexpected Serious Adverse Reaction

• ULN: Upper Limit of Normal

• WOCBP: Women of Childbearing Potential

[Generate complete list specific to this protocol]

Section 11: REFERENCES (60-150 lines)

Generate 30-60 references organized by category using Step 1 research data.

A. FDA Guidance Documents (5-10 from Phase 1):

FDA. [Title]. [Date]. Available at: [URL] [List guidance documents from Phase 1 relevant to indication/intervention]

B. Similar Clinical Trials (10-20 from Phase 1):

[Authors]. [Title]. [Journal]. [Year];[Vol]:[Pages]. ClinicalTrials.gov Identifier: NCT[number]. [List similar trials that informed protocol design from Phase 1]

C. Device-Specific References (IF device - 5-10):

• FDA. [Predicate Device] 510(k) Summary. K[number]. [Date]. Available at: https://www.accessdata.fda.gov/...

• FDA. De Novo Classification Order for [Device]. DEN[number]. [Date]. Available at: https://www.accessdata.fda.gov/...

• FDA. [Device] Classification. Product Code: [ABC]. Available at: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfPCD/classification.cfm

• FDA. Class II Special Controls Guidance: [Device Type]. [Date]. Available at: https://www.fda.gov/...

D. Background Literature (10-20):

[Standard citation format - AMA or Vancouver style] [Key publications on disease, intervention mechanism, prior studies]

E. Intervention Documentation (2-5):

• [Manufacturer]. [Drug/Device Name] Investigator's Brochure/IFU. Version [X]. [Date]. [Mark TBD if needed]

• [Manufacturer]. [Device Name] Technical Specifications. [Date]. [Mark TBD if applicable]

F. Regulatory References (5-10):

1. FDA. 21 CFR Part 50: Protection of Human Subjects. Available at: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=50

2. FDA. 21 CFR Part 56: Institutional Review Boards. Available at: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/CFRSearch.cfm?CFRPart=56

3. FDA. 21 CFR Part 312 (Drugs) OR 21 CFR Part 812 (Devices). Available at: https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfcfr/...

4. ICH E6(R2): Guideline for Good Clinical Practice. Nov 2016. Available at: https://database.ich.org/sites/default/files/E6_R2_Addendum.pdf

5. WMA Declaration of Helsinki: Ethical Principles for Medical Research. 2013. Available at: https://www.wma.net/policies-post/... [Additional FDA/EMA guidance relevant to intervention/indication]

Step 5: Create Operations File

Create waypoints/04_protocol_operations.md with Sections 8-11 only.

Procedure:

1. Generate Section 8 content

2. Generate Section 9 content (using sample size calculation results)

3. Generate Section 10 content

4. Generate Section 11 content

5. Write all sections to new file waypoints/04_protocol_operations.md

6. Confirm file creation successful

Note: The final concatenation of all sections will be performed in Step 5.

Step 6: Update Protocol Metadata

Update waypoints/02_protocol_metadata.json:

• Set step_4_status to "completed"

• Add sections 9, 10, 11, 12 to sections_completed array

• Update protocol_status to "sections_complete" (final concatenation pending)

• Add waypoint_files object tracking all protocol section files

• Update notes

Example:

{
  "intervention_id": "[from metadata]",
  "intervention_name": "[from metadata]",
  "protocol_version": "1.0 Draft",
  "protocol_date": "[current date]",
  "study_design": "[from Phase 1]",
  "enrollment_target": "[from sample size calculation: adjusted_n_total]",
  "primary_endpoint": "[from Section 4]",
  "duration_months": "[from Phase 1 or generated]",
  "regulatory_pathway": "[IDE or IND]",
  "protocol_status": "sections_complete",
  "waypoint_files": {
    "foundation": "02_protocol_foundation.md",
    "intervention": "03_protocol_intervention.md",
    "operations": "04_protocol_operations.md"
  },
  "step_2_status": "completed",
  "step_3_status": "completed",
  "step_4_status": "completed",
  "sections_completed": [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12],
  "sections_pending": [],
  "sample_size_calculated": true,
  "sample_size_n_per_arm": "[from calculation]",
  "sample_size_n_total": "[from calculation]",
  "notes": [
    "DRAFT for planning purposes",
    "All sections (1-12) generated",
    "Final concatenation pending in Step 5",
    "Sample size calculation completed",
    "Requires biostatistician review",
    "Requires clinical expert review",
    "Requires IRB approval",
    "Requires FDA feedback"
  ]
}

Step 7: Update Intervention Metadata

Update waypoints/intervention_metadata.json:

• Add "04-protocol-operations" to completed_steps array

• Set protocol_status to "sections_complete"

Step 8: Display Summary

Display concise summary with:

• Intervention name, protocol version, date

• Step 4 completion status

• ALL Sections generated: 1-12

• Operations file created with size

• Sample size calculation results

• DRAFT disclaimer

• Next steps: Proceed to Step 5 for final protocol concatenation

Example output:

✅ Step 4: Operations and Statistics - COMPLETED
✅ ALL PROTOCOL SECTIONS GENERATED

Intervention: [Name]
Protocol: Version 1.0 Draft (2025-12-03)

📊 Sample Size Calculation:
  • Primary Endpoint: [Endpoint name]
  • Sample Size Per Arm: [n_per_arm]
  • Total Enrollment Target: [adjusted_n_total]
  • Power: [power]%, Alpha: [alpha]

Protocol Status: All 12 Sections Generated
  ✓ Section 1: Statement of Compliance
  ✓ Section 2: Protocol Summary
  ✓ Section 3: Introduction
  ✓ Section 4: Objectives and Endpoints
  ✓ Section 5: Study Design
  ✓ Section 6: Study Population
  ✓ Section 7: Study Intervention
  ✓ Section 8: Discontinuation
  ✓ Section 9: Study Assessments (with AE/SAE reporting)
  ✓ Section 10: Statistical Considerations (with sample size)
  ✓ Section 10: Supporting Documentation and Operations
  ✓ Section 11: References

Files Created/Updated:
  • waypoints/04_protocol_operations.md (NEW: Sections 8-11, ~[lines] lines)
  • waypoints/02_protocol_foundation.md (Sections 1-5)
  • waypoints/03_protocol_intervention.md (Sections 6-7)
  • waypoints/02_protocol_metadata.json (UPDATED)
  • waypoints/02_sample_size_calculation.json (CREATED)
  • waypoints/intervention_metadata.json (UPDATED)

⚠️  DRAFT STATUS: This protocol is preliminary and requires:
  • Biostatistician review and validation of sample size
  • Clinical expert review of all content
  • IRB approval before implementation
  • FDA Pre-Submission or IND/IDE feedback
  • Legal/compliance review

Next Steps:
  1. Proceed to Step 5 to concatenate all sections into final protocol
  2. Review complete protocol after Step 5
  3. Validate sample size calculation with biostatistician
  4. Submit to FDA for Pre-Submission meeting (if applicable)
  5. Submit to IRB for approval

Output Files

Created:

• waypoints/04_protocol_operations.md (Sections 8-11)

• waypoints/02_sample_size_calculation.json (sample size calculation results)

Updated:

• waypoints/02_protocol_metadata.json (step 4 marked complete, waypoint_files added)

• waypoints/intervention_metadata.json (step 4 marked complete)

Existing (from prior phases):

• waypoints/02_protocol_foundation.md (Sections 1-6 from Step 2)

• waypoints/03_protocol_intervention.md (Sections 7-8 from Step 3)

To be created in Step 5:

• waypoints/protocol_complete.md (Final concatenated protocol with all 12 sections)

Error Handling

If Step 3 not completed:

Error: Step 3 must be completed first.
Current status: step_3_status = [status]

Please complete Step 3 to generate intervention details (Sections 7-8) before proceeding to Phase 4.

If protocol section files missing:

Error: Required protocol section files not found.
Expected:
  - waypoints/02_protocol_foundation.md (Phase 2 output)
  - waypoints/03_protocol_intervention.md (Phase 3 output)

Please ensure Phases 2 and 3 are completed before proceeding to Phase 4.

If sample size calculation fails:

Warning: Sample size calculation could not be performed.
Error: [error message]

Falling back to pattern-based approach using similar trials from Phase 1.
⚠️  IMPORTANT: Sample size requires biostatistician review and validation.

Proceeding with protocol generation using enrollment target from similar trials.

## Quality Checks

Before finalizing, verify:
- [ ] Sample size calculation completed successfully (or fallback with disclaimer)
- [ ] Section 8 generated with ALL 9 AE/SAE subsections (8.3.1-8.3.9)
- [ ] Section 9 generated with ALL 10 statistical subsections (9.4.1-9.4.10)
- [ ] Sample size section uses actual values from calculation JSON file
- [ ] Section 10 includes all operational subsections
- [ ] Section 11 includes 30-60 references organized by category
- [ ] Content appended successfully to existing protocol draft
- [ ] Protocol metadata updated correctly
- [ ] Intervention metadata updated correctly
- [ ] COMPLETE protocol now contains all 11 sections

## Notes

- **Sample size calculation is interactive** - requires user input
- Output stays within token limits by focusing on Sections 8-11 only
- **This phase generates the final sections** - all 11 sections now generated but not yet concatenated
- User must proceed to Step 5 for final protocol concatenation