📚 Part of the GLPI-103 Regulatory Dossier — Reader's Guide. This article shows the live document; edits to the source appear here automatically.

🧪

Mock / simulation document

This is a mock / simulation document, made for a portfolio and for learning. The drug (GLPI-103), the sponsor, the people, and the data are all fictional. It is not a real regulatory submission and has no clinical, legal, or regulatory standing. What is real is the shape of the thing — the document structure, the standards it follows, and the analysis methods; the content inside is illustrative.

📄

About this document — a plain-language guide

What it is. The Study Data Reviewer's Guide (SDRG) — the human-readable guide to the SDTM (collected) datasets.

Why it exists. SDTM datasets are standardized but still need explanation: what was collected, any conformance deviations, and how to navigate them. The SDRG orients a reviewer to the raw standardized data.

How it is produced here. It is generated from the standardized study datasets (the SDTM and ADaM data and their define.xml 'data dictionary'), so the guide always describes the exact datasets a regulator would receive.

Format & governing standard. SDTMIG (v3.x) · Define-XML 2.1 · MedDRA (AE coding)

Study Data Reviewer's Guide (SDRG) — GLPI103-301

🧭

A guide to the collected data

Where the ADRG covers the analysis data, the SDRG covers the collected ('SDTM') data — what was gathered, how it maps to the standard, and any quirks a reviewer should know before opening the datasets.

Field	Value
Document ID	DEF-301-SDRG
Version	1.0
Status	Final (portfolio)
Study No.	GLPI103-301
CDISC STUDYID	GLPI103
Standards	SDTMIG (v3.x) · Define-XML 2.1 · MedDRA (AE coding)
Confidentiality	Confidential — portfolio use

Companion to define.xml. Describes the submitted SDTM datasets, conformance, and any reviewer-relevant notes. Generated by tools/build_define.py from the actual datasets.

Change History

Version	Date	Author	Summary
1.0	2026-06-29	Data Management / Biostatistics	Initial version

1. Introduction

This SDRG supports review of the tabulation (SDTM) datasets for Study GLPI103-301, a Phase 3, randomized, double-blind, double-dummy, active-controlled trial of GLPI-103 (IV and Oral) versus oral semaglutide in T2DM (PROT-301).

2. Protocol Summary

1:1:1 randomization (N=900 actual), stratified by baseline HbA1c (<8.5 vs ≥8.5%), BMI (<30 vs ≥30), and Region; 52-week treatment; primary endpoint CFB HbA1c at Week 52 (PROT-301, REF-002).

3. Standards & Dictionaries

🌐

Why Korean values become English codes

The simulated patients are Korean, but CDISC submissions require standardized English terms and a specific transport format. Source values (sex, race, region, and so on) are mapped to the standard English controlled terminology so the datasets are submission-legal.

Item	Value
SDTM IG	3.x
Define-XML	2.1 (`define.xml`)
Controlled Terminology	CDISC CT (SEX, RACE, AE severity NY, SC EDLEVEL/MARISTAT)
MedDRA	AE `AEDECOD`/`AEBODSYS` coded (PT/SOC)

4. Datasets Submitted

Dataset	Class	Records	Structure
DM	Special Purpose	900	One record per subject
SC	Findings	1,980	One record per characteristic per subject (EDLEVEL, MARISTAT, EMPLST)
VS	Findings	23,144	One record per vital sign per visit
LB	Findings	28,930	One record per lab test per visit
AE	Events	815	One record per adverse event

5. Domain-Level Notes

DM — USUBJID = STUDYID-SITEID-SUBJID. SITEID (S001–S017) maps to Korean provinces; COUNTRY=KOR, RACE=ASIAN. Actual = planned arm (no treatment switching modeled).
SC — Subject characteristics derived from enrollment demographics (Nemotron-Personas-Korea): EDLEVEL (Education Level), MARISTAT (Marital Status), EMPLST (Employment Status). Values mapped to English CDISC-aligned controlled terms (source distribution is Korea-grounded).
VS — HEIGHT collected at baseline only; WEIGHT/BMI/SYSBP/DIABP per visit.
LB — HBA1C (primary analyte), FPG, ALT, AST, EGFR. Standardized numeric results in LBSTRESN.
AE — Treatment-emergent events; AESEV (MILD/MODERATE/SEVERE), AEREL (RELATED/NOT RELATED), AESER (Y for 10 serious events, incl. 3 with fatal outcome). GI disorders are the most frequent SOC.

6. Conformance

define.xml is Define-XML 2.1 and well-formed (7 datasets incl. ADaM).

Automated subset validation (tools/validate_cdisc.py → DEF-301_validation_report.md) implements a representative subset of CDISC/Pinnacle 21 rules — required variables, controlled terminology (SEX, AESEV, AEREL, AESER, SC test codes), referential integrity (USUBJID across domains ↔ DM ↔ ADSL), and define↔data consistency. Result: 0 Errors, 0 Warnings.

Official Pinnacle 21 Community validation is the formal conformance step. It could not be run in the build environment (the download is registration-gated and the tool is a GUI Java application; programmatic access returns HTTP 403). Manual steps for the reviewer/sponsor:

Register and download Pinnacle 21 Community from pinnacle21.com (bundles its own JRE).
New Validation → Source = data/sdtm/ (and data/adam/), Define = regulatory/m5-clinical/define/define.xml, Standard = SDTM-IG 3.x / ADaM-IG 1.x.
Export the report and append it here with issue dispositions.

7. Known Data Characteristics (simulated dataset)

10 serious adverse events and 2 deaths recorded (AESER="Y" for serious events; 3 fatal outcomes).
53 major protocol deviations modeled → PPS (847) is a strict subset of FAS (900); see ADRG.
Screen-failure rate (45.0%) exceeds the planning assumption (≈30%) due to realistic source demographics.