Background: High-quality population-based cancer surveillance depends on the completeness and reliability of key sociodemographic and clinical variables. Population-Based Cancer Registries (PBCRs) are internationally recognized as the reference source for cancer incidence data; however, limitations in variable completeness may restrict their ability to support detailed epidemiological analyses. In Brazil, national health information systems integrated within the Unified Health System (SUS) generate large volumes of routinely collected data that may provide complementary clinical information relevant to cancer surveillance. Comparing the quality of information captured by different data sources is essential to identify strengths, limitations, and opportunities to improve cancer surveillance. Objectives: To compare the completeness of selected sociodemographic and clinical variables among female breast cancer cases recorded in the Brazilian National Health Database (BNS) and in the Population-Based Cancer Registry of Belo Horizonte. Methods: A retrospective comparative study was conducted including women aged 18 years or older diagnosed with breast cancer between 2002 and 2014. Cases were identified from two data sources: the National Health Database centered on the individual (BNS), constructed through linkage of national mortality, hospital, and outpatient information systems; and the Population-Based Cancer Registry of Belo Horizonte. Annual numbers of incident cases were compared between sources to assess case capture. Data quality was evaluated by estimating the proportion of missing information for age at diagnosis, race/skin color, vital status (death), and stage at diagnosis (tumor extension). Completeness was defined as the presence of valid, non-missing values. Descriptive analyses were performed, and differences between data sources were assessed using chi-square tests, adopting a 5% significance level. Results: A total of 13,708 breast cancer cases were recorded in the Population-Based Cancer Registry of Belo Horizonte, compared with 9,115 cases identified in the National Health Database. The annual number of cases was consistently higher in the registry, indicating greater capture of incident cases by the PBCR. Both data sources showed high completeness for age at diagnosis and vital status. Race/skin color exhibited substantial levels of missing information in both databases, with 26% missing in the PBCR and 27% in the BNS. In contrast, marked differences were observed for stage at diagnosis. The PBCR showed very high incompleteness for tumor extension, with 82% of records missing this information, whereas the BNS demonstrated better completeness for this variable. These findings indicate that clinical information recorded within SUS healthcare systems may capture staging-related details not consistently available in registry records. ?Conclusions: Although the Population-Based Cancer Registry of Belo Horizonte demonstrated superior capture of incident breast cancer cases, important gaps in variable completeness persist, particularly for race/skin color and stage at diagnosis. The high proportion of missing staging information in the registry represents a critical limitation for population-based analyses and evaluation of cancer care. The better completeness of staging information observed in the national health database suggests that health system data may provide valuable complementary clinical information. Assessing the strengths and limitations of different data sources is essential to improve the quality of cancer surveillance and support robust epidemiological monitoring in Brazil.

Number of cases by year of diagnosis according to data source