(S-007) Understanding the Utilization of SDTM and ADaM Datasets in Pharmacometric Programming

Erin Dombrowsky – Director, Quantitative Pharmacology and Data Analytics, Bristol Myers Squibb; Rebecca Humphrey – Principal Data Programmer, Clinical Pharmacology and Pharmacometrics, Simulations Plus, Inc.

Objectives: Pharmacometric (PMx) programmers are skilled in handling all data-related activities to support modeling and simulation efforts, including merging disparate datasets to create a single, analysis-ready dataset. Creating PMx datasets can be challenging due to their intricate structure and need for precise data manipulation and modelers may not have programmer support. The source for PMx datasets is a combination of SDTM (Study Data Tabulation Model) and ADaM (Analysis Data Model) datasets. The large number of these datasets, along with their complexity, can be overwhelming or confusing for a modeler who does not work with them regularly. Understanding their structure and purpose is essential for high-quality results.

Methods: The Clinical Data Interchange Standards Consortium (CDISC) develops data standards to streamline the drug development process, ensuring consistent, high-quality data across studies. SDTM and ADaM are two standards essential for population pharmacokinetic (popPK) and exposure-response (E-R) analyses.

SDTM datasets are organized into four observation classes: events, findings, interventions, and special purpose. Events include medical history (MH), findings cover laboratory results (LB) and pharmacokinetic concentrations (PC), interventions consist of exposure (EC/EX) and concomitant medications (CM), and special purpose datasets include demographics (DM).

ADaM datasets transform SDTM data into formats ready for statistical analysis and reporting, ensuring traceability from data collection to analysis. ADSL provides a single record per subject and includes key variables such as demographics, baseline characteristics, key dates, and population flags. ADAE contains derived variables such as treatment-emergent flags and grouped adverse event categories, essential for E-R safety analysis.

Both SDTM and ADaM have Implementation Guides (IG) that provide guidance on the structure, format, and organization of datasets, including typical variables, their labels, and controlled terminology.

Results: SDTM and ADaM datasets provide a standardized structure for the data necessary for PMx analysis. They enhance dataset quality, enable automation, streamline subsequent analyses, and improve communication. Collaboration with biostatistics colleagues ensure the proper variables are used.

The PMx programming community actively presents, publishes, and shares tools for creating analysis datasets using SDTM and ADaM as source information. The CDISC Basic Data Structure for ADaM popPK IG is a standardized template for model-based popPK, serving as a valuable reference for modelers.

SDTM and ADaM are usually in SAS data formats (e.g., .sas7bdat or .xpt), but modelers can use the haven R package to import data while preserving variable labels, facilitating the use of SDTM and ADaM datasets in R.

In addition to variable names and labels, the use of --TEST/--TESTCD and PARAM/PARAMCD in SDTM and ADaM datasets, respectively, can be key to locating necessary data for PMx programming.

Conclusions: While understanding SDTM and ADaM datasets may initially seem complex, recognizing their standardized organization and consistency simplifies the process.

Citations: [1] Yan, Y., Sukumar, P., & Thanneer, N. “Using R to Create Population Pharmacokinetic Data Set.” Proceedings of PharmaSUG 2023. San Francisco, California.
[2] Chen L, Dombrowsky E, Boyle B, Tang C, Thanneer N. PmWebSpec: An Application to Create and Manage CDISC-Compliant Pharmacometric Analysis Dataset Specifications. AAPS J 26, 39 (2024). https://doi.org/10.1208/s12248-024-00910-0
[3] Clinical Data Interchange Standards Consortium (CDISC). (2021, 29 November). Study Data Tabulation Model Implementation Guide: Human Clinical Trials Version 3.4 (Final). Retrieved from https://www.cdisc.org/standards/foundational/sdtmig/sdtmig-v3-4
[4] Clinical Data Interchange Standards Consortium (CDISC). (2021, 29 November). Analysis Data Model (ADaM) Implementation Guide Version 1.3. Retrieved from https://www.cdisc.org/standards/foundational/adam/adam-implementation-guide-v1-3
[5] Clinical Data Interchange Standards Consortium (CDISC). (2023, October 6) Basic Data Structure for ADaM popPK Implementation Guide v1.0. Retrieved from: https://www.cdisc.org/standards/foundational/adam/basic-data-structure-adam-poppk-implementation-guide-v1-0

Keywords: pharmacometric programming, CDISC standards, clinical data