Hoe verbeteren we de kwaliteit van data in onze kwaliteitsregistraties? Dit onderzoek dat uitgevoerd is met behulp van data uit de Dutch Lung Cancer Audit - Surgery (DLCA-S) en dat verschenen is in het Journal of Thoracic Disease legt het uit:
National quality registries: how to improve the quality of data?
Auteurs: Fieke Hoeijmakers, Naomi Beck, Michel W. J. M. Wouters, Hubert A. Prins, Willem H. Steup
Lees hier het volledige artikel; lees hieronder een samenvatting van het artikel:
Background: Data of quality registries are increasingly used by healthcare providers, patients, health insurance companies, and governments for monitoring quality of care, hospital benchmarking and outcome research. To provide all stakeholders with reliable information and outcomes, reliable data are of the utmost importance.
Methods: This article describes methods for quality assurance of data—used by the Dutch Institute for Clinical Auditing (DICA)—regarding: the design of a registry, data collection, data analysis, and external data verification. For the Dutch Lung Cancer Audit for Surgery (DLCA-S) results of data analysis and data verification were assessed with descriptive statistics.
Results: Of all registered patients in the DLCA-S in 2016 (n=2,391), 98.2% was analysable and completeness of data for calculations of transparent outcomes was 90.7%. Data verification for the year 2014 showed a case ascertainment of 99.4%. Of 15 selected hospitals, 14 were verified. All these hospitals received the conclusion ‘sufficient quality’ on case ascertainment, mortality (0% under-registration) and complicated course (3.3% wrongly registered complications). One hospital was not able to deliver patients lists, and therefore not verified.
Conclusions: Quality of data can be promoted in many different ways. A completeness indicator and data verification are useful tools to improve data quality. Both methods were used to demonstrate the reliability of registered data in the DLCA-S. Opportunities for further improvement are standardised reporting and adequate data extraction.