Assessment of Data Quality in Relapse Tuberculosis Surveillance Data in Botswana (2022)

Abstract Background Reliable tuberculosis (TB) surveillance data are essential for monitoring programme performance and estimating disease burden. Although completeness is commonly used to assess surveillance data quality, duplicate records within parallel electronic reporting systems may distort surveillance estimates. This study assessed completeness of key surveillance variables and the extent of duplication within relapse TB datasets reported in Botswana during 2022. Methods A cross-sectional study was conducted using relapse TB records extracted from three national electronic health information systems in Botswana: OpenMRS, the Integrated Patient Management System hospital module (MPM), and the Botswana Health Posts clinic module (BHP). Data quality was evaluated through completeness and record uniqueness. Completeness was assessed as the proportion of records with non-missing values for key surveillance variables. Duplicate records were identified using deterministic linkage based on national identification numbers, supplemented by name-based matching procedures. Results A total of 228 relapse TB records were identified prior to deduplication. Completeness of key surveillance variables exceeded 98% across all datasets. The OpenMRS dataset contained 74 records, of which 72 were unique after deduplication (2.7% duplication). In contrast, the MPM dataset contained 153 records corresponding to only 17 unique relapse TB cases after deduplication (88.9% duplication). The BHP dataset contained one relapse TB record with no duplicates. Overall, 90 unique relapse TB cases remained following deduplication. Conclusion Despite high completeness of surveillance variables, substantial duplication within the hospital-based reporting system significantly affected the reliability of relapse TB estimates. These findings demonstrate that completeness alone may inadequately reflect surveillance data quality in settings with multiple electronic reporting platforms. Strengthening patient identification systems, interoperability, and routine deduplication procedures may improve the accuracy and reliability of TB surveillance data in Botswana.