Validation of mortality related information in the German Pharmacoepidemiological Research Database
Description
- Death is the most serious, patient relevant endpoint in health care related research. For many questions the cause of death is crucial for the case definition, i.e. to identify insurants with the endpoint of interest. For epidemiological studies information about the exact date of death is essential, eg. to define the person-time at risk to develop a certain endpoint of interest. For studies considering death as endpoint or exclusion criterion, the validity of coding of death is indispensable. Therefore, we validated the date of death and status of death internally and externally in this study. The first part of the study, we compared mortality figures obtained from the German Pharmacoepidemiological Research Database (GePaRD) to national statistics on an aggregated level. Within the same study, the internal consistency of mortality related information was also evaluated, comparing two information sources within GePaRD regarding the date of death. We could show that the age-standardized mortality rate was lower compared to data of the FSOG. We attributed the lower mortality rates to the comparatively high socioeconomic status (SES) in the study population, as it is known that a lower SES is associated with an increased morbidity and mortality. This presumption was supported by the fact that the assessed mortality rate agreed well with national statistics in a federal state, in which the study population was more representative regarding the SES. Furthermore, 94% of dates of death documented in hospital agreed with the dates of death according to the health insurance core data.
Based on an approved data protection concept we conducted a data linkage with a local mortality index (Bremer Mortalitätsindex: BreMI) and were able to validate death information for a subsample of GePaRD on the individual level. Our results suggest a high accuracy also of the date of death in GePaRD which was identical to the gold standard information in 97.5% of all cases. However, only 84,0% of the deaths identified in GePaRD could be linked. Further, 18.6% of insurants with an ending last insurance period not classified as ending by death according to GePaRD died according to BreMI. Algorithms defined to reconstruct the cause of death in GePaRD showed good predictive values for cancers with a high case fatality rate, but achieved worse results for cancers with a medium case fatality rate as well as for acute and chronic cardiovascular diseases. However, the estimated negative predictive values were consistently near 100% in all death causes considered..
Funding period
- Begin: June 2010
End: February 2014
Sponsor
- Pharmaceutical industry
Contact
Selected project-related publications
- Ohlmeier C, Langner I, Garbe E, Riedel O. Validating mortality in the German Pharmacoepidemiological Research Database (GePaRD) against a mortality registry. Pharmacoepidemiology and Drug Safety. 2016;25(7):778-784.
https://doi.org/10.1002/pds.4005 - Ohlmeier C, Langner I, Hillebrand K, Schmedt N, Mikolajczyk R, Riedel O, Garbe E. Mortality in the German Pharmacoepidemiological Research Database (GePaRD) compared to national data in Germany: Results from a validation study. BMC Public Health. 2015;15:570.
https://doi.org/10.1186/s12889-015-1943-7 - Ohlmeier C, Schmedt N, Hillebrand K, Langner I, Mikolajczyk R, Garbe E. Validation of mortality related information in the German Pharmacoepidemiological Research Database (GePaRD). 5. Workshop der Arbeitsgemeinschaft Pharmakoepidemiologie der Deutschen Gesellschaft für Epidemiologie (DGEpi) und der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (GMDS), 14.-15. Juni 2012, Bremen.
- Mikolajczyk R, Ohlmeier C, Budde E, Giersiepen K, Zeeb H, Garbe E. Ascertainment of breast cancer mortality in German health insurance data. 7. Jahrestagung der Deutschen Gesellschaft für Epidemiologie (DGEpi), 26.-29. September 2012, Regensburg.
- Ohlmeier C, Niemeyer M, Garbe E, Mikolajczyk R. Identifizierung von Todesursachen in Daten der Gesetzlichen Krankenversicherung am Beispiel des Lungen- und Pankreaskrebs. 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (GMDS), 16.-21. September 2012, Braunschweig.
- Ohlmeier C, Schmedt N, Hillebrand K, Langner I, Mikolajczyk R, Garbe E. Validation of mortality information in the German Pharmacoepidemiological Research Database (GePaRD). 29th International Conference on Pharmacoepidemiology & Therapeutic Risk Management (ICPE), 25-28 August 2013, Montreal, Canada.