Related Experiment Videos

Discharge status validation of the Chang Gung Research database in Taiwan

Yu-Tung Huang1, Ying-Jen Chen2, Shang-Hung Chang3

  • 1Center for Big Data Analytics and Statistics, Department of Medical Research and Development, Chang Gung Memorial Hospital at Linkou, Taoyuan, Taiwan.

Biomedical Journal
|December 31, 2021
PubMed

Insights

This study validated the Chang Gung Research Database (CGRD) for mortality data accuracy. Findings show high accuracy for electronic medical records after 2010, recommending their use for future research.

Area of Science:

  • Medical Informatics
  • Health Data Science
  • Epidemiology

Background:

  • The Chang Gung Research Database (CGRD) is a large, multi-institutional electronic medical records database in Taiwan.
  • Validation of CGRD's data accuracy, particularly for discharge status and mortality, is crucial but has been limited.
  • This study addresses the need to validate the accuracy of CGRD for mortality outcome research.

Purpose of the Study:

  • To validate the accuracy of discharge status, with a specific focus on mortality, within the Chang Gung Research Database (CGRD).
  • To assess the reliability of CGRD data for epidemiological studies relying on mortality outcomes.

Main Methods:

  • An observational study was conducted, linking CGRD data with the Taiwan Disease Registry (TDR).
  • Validation metrics included accuracy, positive predictive value (PPV), and underestimated mortality rate (UEM).
  • Analyses considered variations by year, sex, age, and primary cause for admission (PCA).

Main Results:

  • Overall accuracy for mortality coding in CGRD discharge status exceeded 97% within one week post-discharge.
  • Data accuracy improved annually, surpassing 98% after 2010.
  • Underestimated mortality rates (UEM) were below 10% after 2010, though accuracy varied by age and primary cause for admission, with elderly patients showing lower accuracy and higher UEM.

Conclusions:

  • The Chang Gung Research Database (CGRD) demonstrates high accuracy for mortality data, especially for inpatient records after 2010.
  • Prioritizing CGRD data from 2010 onwards is recommended for mortality outcome follow-up studies.
  • Careful consideration of age and primary cause for admission is advised when interpreting CGRD mortality data.
Abstract

Related Concept Videos