To the Editor: Optimising the management of cancer patients requires objective decisions about "best treatment" strategies, based on high quality data collected systematically from all treated patients (or at least a representative sample of them). Relating treatment and stage at diagnosis to individual outcome can allow monitoring of whether treatment is consistent with best practice, and can provide a systematic foundation for evidence-based care. Clinical cancer data collection also allows treatment services to be evaluated, as institutions can monitor throughput and endpoints. However, institution-based data collections may not be representative of all cancer patients, and aggregation of data from several institutions is needed to obtain a comprehensive picture.

Population-based cancer registries, which operate in all Australian States and Territories, include data on the site and morphology of cancers. Notification of cases to the registries is mandatory for hospitals and pathology providers, and survival of patients is assessed by linkage to mortality data. The registers do not routinely record stage or treatment data. Until recently, there have been no nationally agreed data items or standard data definitions to facilitate the collation of clinical cancer data across institutions.

In 1999, the National Cancer Control Initiative (NCCI) commissioned a nationwide consultation process to seek expert advice on developing a core clinical cancer data set. Representatives from the State and Territory population-based cancer registries, the Australian Institute of Health and Welfare and many large cancer treatment centres were consulted.1 A workshop was held in Melbourne in July 2000 to identify key items for inclusion in the data set, and a group was established to work on data definitions. These are now available on the NCCI's website (<http://www.ncci.org.au/projects/data/dat01.htm>). The data set is designed to be compatible with, and expand on, data currently collected by State cancer registries. Definitions are consistent with the New South Wales clinical cancer data set,2 and we acknowledge the input from this source. Items would be collected by treatment centres. Some institutions would need to standardise information already collected for ongoing patient management, while others would need to establish and maintain new collections. The Faculty of Radiation Oncology of the Royal Australian and New Zealand College of Radiologists has recommended incorporation of the NCCI data set into its proposed quality assurance program. Collation of data across institutions requires careful attention to patient identification issues in order to protect privacy and avoid duplication of data from multiple sources.

Use of the data set by clinicians and health planners and evaluators at a national level is the ultimate aim. This would require funding and commitment, and attention to issues of privacy, confidentiality, and data ownership. At present, adoption of the data set on a voluntary basis by treatment centres is the best way forward.