Evaluation of data quality at the Hungarian National Cancer Registry, 2000-2019

Abstract

Background: The Hungarian National Cancer Registry (HNCR) was legally established as a population-based cancer registry in 1999, and its operation started in 2000 supporting the planning and development of the Hungarian oncology network as well as informing national cancer control policies. Ensuring comparable, accurate, and complete data on malignant and in situ neoplasms is critical in determining the applicability of the database. The aim of this study was to perform a comprehensive evaluation of the data quality at the HNCR.

Methods: Based on qualitative and semiquantitative methods from current international guidelines, we assess the comparability, completeness, validity, and timeliness of the collected data over the diagnostic period 2000-2019, with a focus on the year 2018.

Results: Coding practices and the classification system used at the HNCR are based on the International Classification of Diseases (ICD-10), which differs from the internationally recommended ICD-O. The annual trends in incidence did not indicate major fluctuations, that may have resulted from data collection discrepancies, while comparisons of the mortality-to-incidence ratio (M:I) compared with 1 minus 5-year observed survival indicated some systematic differences requiring further exploration. The age-standardized (European standard) incidence rate per 100 000 measured by the HNCR in 2018 was very high: 647.9 for men and 501.6 for women, 11.6% and 14.6% higher than the International Agency for Research on Cancer (IARC) estimates respectively. Behind the overall differences between the two data sources, we identified that the vast majority were due to ill-defined ICD codes: malignant neoplasm of other and ill-defined sites (C76), and malignant neoplasm without specification of site (C80). Otherwise, there were no major discrepancies by localization. The proportion of morphologically verified cancer cases was 57.8% overall, that of death certificates was 2.3%, and that of unknown primary tumors was 1.4%.

Conclusion: Further implementations and interventions are required to ensure that the operations, coding practices, and the classification system used at the national registry are in accordance with international standards, and to increase the completeness and validity of the collected cancer data. In particular, the low morphologically verified proportion questions the overall accuracy of the stated diagnoses within the database. Nevertheless, our examination implies that the data of the HNCR are reasonably comparable, and without doubt fulfill the requirements to support national oncology services and cancer planning. However, most importantly, a review of registry personnel and resource requirements to run the national population-based cancer registry should be an essential part of Hungary's national cancer strategy.

Keywords: Cancer registry; Comparability; Completeness; Data quality; Hungary; Timeliness; Validity.