Evaluate the Statistical Utility After Transformation of a CDISC SDTM Database Into OMOP CDM
Author(s)
Camille Bachot, MSc1, David Pau, MSc2, Amelie Lambert, MSc3, Claire Castagné, MSc3.
1Medical Data Platform Specialist, ROCHE, Boulogne Billancourt Cedex, France, 2Roche, Boulogne-Billancourt, France, 3Roche, Boulogne Billancourt, France.
1Medical Data Platform Specialist, ROCHE, Boulogne Billancourt Cedex, France, 2Roche, Boulogne-Billancourt, France, 3Roche, Boulogne Billancourt, France.
OBJECTIVES: Interoperability between databases is becoming an important issue, to facilitate analyses from multiple sources. The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) is increasingly used in Europe, to standardize databases in order to facilitate interoperability. No published work assesses the loss of data following the transformation to OMOP CDM. The aim is to assess the statistical and scientific usefulness of a transformation from CDISC Study Data Tabulation Model (STDM) to OMOP CDM.
METHODS: An French observational study in early breast cancer was conducted (KADOR study) in 2019. The original database was in CDISC SDTM. A first step involved to transform the SDTM database into OMOP CDM. The conversion was done using OHDSI tools and Data Build Tool . A statistical analysis of the final data in OMOP CDM was carried out. All the results were compared with the initial results, and quality indicators were used to assess the loss of information at the various stages:
RESULTS: 7 CDISC domains containing 73 variables: 25 continuous and 48 categorical regarding patient, disease, surgery and treatments characteristics. Descriptive analyses, correlation matrices, modeling and survival analysis were first performed on the raw study data (SDTM), then these same analyses were reproduced on the OMOP database. Usual statistical indicators and maintenance of relationships between variables were used to quantify the differences between the databases in the different models.
CONCLUSIONS: This work assess the statistical usefulness remaining after the switch to OMOP CDM, thanks to a synthesis of the various indicators, and to ensure the reproducibility of classic statistical analyses used in a research project with this standard. The results/indicators observed on the OMOP CDM database will be presented in the poster.
METHODS: An French observational study in early breast cancer was conducted (KADOR study) in 2019. The original database was in CDISC SDTM. A first step involved to transform the SDTM database into OMOP CDM. The conversion was done using OHDSI tools and Data Build Tool . A statistical analysis of the final data in OMOP CDM was carried out. All the results were compared with the initial results, and quality indicators were used to assess the loss of information at the various stages:
- Indicators regarding the syntactic transformation,
- Indicators regarding the number of statistical tables not generated,Indicators regarding the reliability of results obtained by the comparison
RESULTS: 7 CDISC domains containing 73 variables: 25 continuous and 48 categorical regarding patient, disease, surgery and treatments characteristics. Descriptive analyses, correlation matrices, modeling and survival analysis were first performed on the raw study data (SDTM), then these same analyses were reproduced on the OMOP database. Usual statistical indicators and maintenance of relationships between variables were used to quantify the differences between the databases in the different models.
CONCLUSIONS: This work assess the statistical usefulness remaining after the switch to OMOP CDM, thanks to a synthesis of the various indicators, and to ensure the reproducibility of classic statistical analyses used in a research project with this standard. The results/indicators observed on the OMOP CDM database will be presented in the poster.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
EPH91
Topic
Epidemiology & Public Health, Methodological & Statistical Research, Real World Data & Information Systems
Topic Subcategory
Disease Classification & Coding
Disease
Oncology