RESUMEN
OBJECTIVE: To evaluate the completeness and consistency of data from hospital-based cancer registries (HCRs) in a Brazilian state. METHODS: This retrospective descriptive study was based on secondary data from an HCR in the state of Espírito Santo (ES) between 2010 and 2017. The data were collected between August and November 2020 by the ES State Health Department (SESA/ES). Cancer data were obtained from the HCR of ES using the tumor registration form of the Brazilian Hospital Cancer Registry Integrator and complete databases within the SESA/ES. The incompleteness of the data was classified as excellent (<5%), good (between 5% and 10%), regular (between 10% and 20%), poor (between 20% and 50%), and very poor (>50%), according to the percentage of the absence of information. Descriptive statistical analyses were performed using Statistical Package for the Social Sciences (SPSS® Inc., Chicago, IL, USA) version 20.0. RESULTS: Complete data were observed for the variables of sex, date of the first hospital visit, and histological type of the primary tumor; that is, there were no missing data. Most epidemiological variables, including age, origin, date of first tumor diagnosis, previous diagnosis and treatment, location of the primary tumor, first treatment received at the hospital, date of death of the patient, and probable location of the primary tumor, were classified as having excellent completeness throughout the study period. However, the variables schooling, smoking, alcohol consumption, occupation, family history of cancer, and clinical staging of the tumor were classified as poor. CONCLUSION: Most epidemiological variables from the HCR in the state of ES, Brazil, showed excellent completeness. It is essential to elucidate the sociodemographic and clinical variables of epidemiological importance for a better understanding of the health-disease process.