Harmonizing patient health data into the OMOP common data model: a case study using tuberculosis data from Douala General Hospital, Cameroon

Introduction Douala General Hospital is a tertiary healthcare facility in Cameroon that serves thousands of patients annually. Despite maintaining extensive patient records with potential for public health research, most data remain paper-based, limiting accessibility, reuse, and interoperability. In departments such as pulmonology, patient information is stored in heterogeneous, non-standardized formats, constraining its use for clinical research, care management, and decision-making. To address these challenges, we implemented a complete Extract, Transform, and Load (ETL) pipeline aligned with the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) Version 5.4. Objective This study aimed to harmonise and integrate tuberculosis (TB) patient data into the OMOP CDM to improve data quality, interoperability, and reusability for research and clinical monitoring. Methods A retrospective TB dataset covering 2009–2022 was digitized from paper records and included 2,426 patient records with 82 variables. Data profiling was performed using WhiteRabbit, mapping with Rabbit-in-a-Hat, and vocabulary alignment using Usagi and ATHENA. Transformed data were loaded into a PostgreSQL OMOP CDM database. Data quality was evaluated using Achilles and the Data Quality Dashboard (DQD), while ATLAS supported visualization and analytical exploration. Results Of 2,374 DQD checks executed, 2,354 passed, and 42 failed, yielding a 99% overall pass rate. Many checks that passed were not applicable to this TB-specific dataset, and all applicable checks demonstrated high data quality after iterative ETL refinement. Most issues were minor and resolved during refinement. Demographically, 55.6% of patients were male, with ages ranging from 15 to 95 years, predominantly between 30 and 40 years. Common conditions included extrapulmonary TB, fever, cough, and chest pain. Most patients contributed one condition and one observation record, reflecting the dataset's cross-sectional nature. Conclusion Harmonising TB data into the OMOP CDM was feasible and successful, demonstrating that standardized research datasets can be generated in resource-limited settings. This approach offers a scalable model for broader health data standardisation efforts in Cameroon and similar contexts.