Table 2 Scenarios

Scenario 1- health data linked by experts

A study is being conducted by university researchers and they want hospital information (which includes the medical history, name, age, ethnicity, and postcode), a cancer register, and a deaths register to be linked with each other. The researchers want to find out if there is a link between lung cancer and living next to busy main roads. The findings will contribute to better town planning. In order for the study to be successful and so that it provides accurate findings to ultimately help with the management of some forms of cancer, it’s very important for everyone on the cancer register (several thousand people) to be included in the study. The researchers will never have any identifying information because they will not do the linkage themselves and all information that identifies people will be removed before they get the linked data.

Scenario 2 - health data linked by experts

Researchers in collaboration with the ambulance service are conducting research into cardiac arrests and resuscitation to see if call-out response times affect survival rates. They will need to have data about approximately 300,000 people on the ambulance databases, hospital admissions, and death registers linked. The researchers will never have any identifying information because they will not do the linkage themselves and all information that identifies people will be removed before they get the linked data.

Scenario 3 - health & criminal records data linked by experts

University researchers and researchers from a mental health organisation want to study violent behaviour in people experiencing mental health issues. They need to link about 50,000 mental health hospital records Australia-wide (including admissions and discharge information) with police incident information, such as calls for domestic violence. The researchers will never have any identifying information because they will not do the linkage themselves and all information that identifies people will be removed before they get the linked data. The Police will not have access to the mental health hospital records.

Now still talking about the same research, the process of linking such a lot of information from a number of States/Territories is very complex, expensive and very time-consuming. So in this research, once linked, the identifying information will be removed but a key will be held separately. The key connects the identifying information of all the people on the databases and the codes they were given. The key will make it possible for researchers from various States/Territories to ask for de-identified information about their State/Territory so that they can do further research without having to have all the information linked from the start.

Scenario 4 – health, Work Cover & employment data linked by researchers

University researchers are conducting research on work-related stress on behalf of Work Cover to discover whether there is a link between increased levels of stress and work insecurity, for example caused by casual employment. They need to link 100,000 work stress claims containing identifiable general and mental health information with employment data including employment history, leave information, seniority level etc. for the same individuals who are employed at the Government organisations involved in the research. In this instance, the linkage will not be done by an independent team of data linkage experts. Instead, the researchers will do the linkage themselves. A report with de-identified findings will be made available to Work Cover. [In discussions with participants, it was made clear that work stress claims would be accessed from WorkCover^a.]