Measuring the quality of care using a large clinical database: Lessons learned in oncology

Abstract

Ever since the Cancer Control Act was enacted, ensuring the quality of cancer care has been a central theme in cancer control policies. Medical record review was initially considered to be the best available method for quality measurement, and it was attempted in several hospitals. However, the resulting labor burden was enormous. Seeking a viable alternative, we started using existing electronic and widely standardized datanamely, data from health insurance claims and hospitalbased cancer registries. The registry data are useful for defining the target population, and the claims data let us examine the care that the patients received. If we can link these data, they can become a powerful information source. In collecting linkable data, we found that the greatest challenge was deleting personal identifiers while maintaining the links between paired data belonging to the same patients. To overcome this challenge, we developed specialized software to enable hospitals to replace patient identifiers in files with common anonymous IDs and then safely submit these data to the National Cancer Center. The process was challenging because collected data often did not conform to the data format standards, but we finally succeeded in collecting data from 232 hospitals. The primary purpose of the current project was to calculate the performance of 13 quality indicators and provide the results to participating hospitals. As a secondary outcome, the data will be used to provide necessary information in the discussion of cancer control.