End-to-end CDISC Workflow (Mock Study)

CDISC
SAS
SDTM
ADaM
TLF
From raw sources to SDTM, ADaM, and TLFs using SAS (SAS OnDemand).

This portfolio project shows an end-to-end CDISC pipeline implemented in SAS OnDemand: curation of raw-like sources → SDTM domains → ADaM analysis datasets → TLFs (tables/listings/figures).
Repo: https://github.com/Ousmanerabi/clinical-trials-programming-portfolio

At a glance

  • Objective: produce regulatory-ready data flow (SDTM → ADaM → TLF) for a mock clinical study.
  • Stack: SAS (OnDemand), CDISC SDTM & ADaM principles.
  • What you’ll see here: a concise walkthrough, code excerpts, and links to the full programs and outputs.
Note

Why a mock study?
To demonstrate real-world clinical programming skills without sharing proprietary datasets, while following CDISC conventions and good programming practices.

Data sources & scope

  • Inputs: synthetic/raw-like CSV/XLSX files and SAS datasets stored in the repo (for demonstration).
  • Study artifacts: trial design shells and example CRF mapping logic where needed.
  • Outputs: SDTM domains (e.g., DM, AE, EX, VS), then ADaM datasets (e.g., ADSL, ADAE), and selected TLFs.

Repository layout (key folders)

clinical-trials-programming-portfolio/ 01_RAW/ # raw-like inputs (csv/xlsx/sas7bdat)

02_SDTM/
programs/ # SAS programs (domain-specific)
out/ # SDTM outputs (sas7bdat/preview png)
03_ADaM/

programs/ # SAS programs (ADSL, ADAE, etc.)
out/ # ADaM outputs
04_TLF/
programs/ # TLF SAS code (tables, listings, figures)
out/ # rendered artefacts (rtf/pdf/png)
docs/ # helper docs (specs, trial design)

SDTM (Study Data Tabulation Model)

Goal: create standardized, submission-ready domains that mirror CRF content and study design.

Highlights - Patient key: USUBJID built from STUDYID + subject identifier.
- Controlled terminology & value-level mapping (e.g., SEX, ARM, AESER).
- Date handling (partial dates), label/length alignment.

Example — DM (Demography) derivation (excerpt)

/* 02_SDTM/programs/sdtm_dm.sas */ 

data sdtm.dm(label="Demographics"); 

set raw.demography; 

length STUDYID $12 USUBJID $40 DOMAIN $2 ARM $20 ARMCD $8 SEX $1 AGEU $5; 

DOMAIN = "DM"; 

USUBJID = catx("-", STUDYID, put(SUBJID, z4.)); 

AGEU = "YEARS"; 

/* … additional recodes / controlled terms … */

 keep STUDYID DOMAIN USUBJID SUBJID SEX AGE AGEU ARM ARMCD; run;

proc sort data=sdtm.dm; by USUBJID; 

run;

Quality gates (SDTM): uniqueness of USUBJID, required variables by domain, controlled codes, and basic cross-domain checks (e.g., DM vs AE subject coverage).