J Prev Med Public Health.  2010 May;43(3):257-264. 10.3961/jpmph.2010.43.3.257.

Construction and Validation of Hospital-Based Cancer Registry Using Various Health Records to Detect Patients with Newly Diagnosed Cancer: Experience at Asan Medical Center

• Kim HJ
• Cho JH
• Lyu Y
• Lee SH
• Hwang KH
• Lee MS

• KMID: 986135
• DOI: http://doi.org/10.3961/jpmph.2010.43.3.257

Abstract

OBJECTIVES
An accurate estimation of cancer patients is the basis of epidemiological studies and health services. However in Korea, cancer patients visiting out-patient clinics are usually ruled out of such studies and so these studies are suspected of underestimating the cancer patient population. The purpose of this study is to construct a more complete, hospital-based cancer patient registry using multiple sources of medical information. METHODS: We constructed a cancer patient detection algorithm using records from various sources that were obtained from both the in-patients and out-patients seen at Asan Medical Center (AMC) for any reason. The medical data from the potentially incident cancer patients was reviewed four months after first being detected by the algorithm to determine whether these patients actually did or did not have cancer. RESULTS: Besides the traditional practice of reviewing the charts of in-patients upon their discharge, five more sources of information were added for this algorithm, i.e., pathology reports, the national severe disease registry, the reason for treatment, prescriptions of chemotherapeutic agents and radiation therapy reports. The constructed algorithm was observed to have a PPV of 87.04%. Compared to the results of traditional practice, 36.8% of registry failures were avoided using the AMC algorithm. CONCLUSIONS: To minimize loss in the cancer registry, various data sources should be utilized, and the AMC algorithm can be a successful model for this. Further research will be required in order to apply novel and innovative technology to the electronic medical records system in order to generate new signals from data that has not been previously used.

Keyword

Completeness; Hospital-based cancer registries; Medical records linkage; Tumor registration

MeSH Terms

Adult
Female
Hospitals
Humans
Male
*Medical Records
Middle Aged
Neoplasms/*diagnosis
Organizational Case Studies
*Program Development
*Registries
Republic of Korea