Linking data from hospital and cancer registry databases: should this be standard practice?

Varying amounts of data related to cancer diagnosis, treatment and/or outcome are routinely collected by many disparate groups. Routinely combining data from these sources could improve data quality and utility for audit and research purposes. The aim of this study is to demonstrate the benefits of linkage between oncology databases.

We examined colorectal cancer (CRC) data recorded by the Victorian Cancer Registry and two hospital clinical databases between January 2000 and December 2005. Where data were in common, the completeness and accuracy of each dataset were examined. Where content differed, the potential value of making this additional data available to the other database was examined.

Of the 831 cases recorded at the hospitals, 822 (98.9%) were also recorded on the cancer registry. Eight of the 913 cases (0.87%) recorded as having CRC by the registry did not have CRC. Errors in recording of tumour site and tumour or nodal stage were frequent in both databases. Metastasis stage was recorded in only 29 of 822 (3.5%) registry cases examined. Discordance for diagnosis date and death date was also frequent, although the difference was typically minor. Adding additional death data from the registry to the clinical database significantly altered stage-specific and overall survival figures.

A multidirectional flow of data between hospital and registry databases provides multiple opportunities to improve data quality and utility. While issues around data ownership and usage need to be considered, the advantages of routine data linkage are readily apparent.