What researchers mean by... data linkage

Have you ever wondered how a financial institution determines whether you qualify for a loan or mortgage? One piece of information the financial institution may look at is your credit report. This is a report that lists your personal income and debts (such as loans, mortgages and credit cards). Based on this report plus other factors, the financial institution makes a decision.

So how does the financial institution get all of these pieces for your credit report?

They link data – or information – from many sources and bring the information together to get a better sense of a person’s whole financial picture. In the example above, the financial institution may get information from your credit card company, your mortgage broker and your bank, among others.

Connecting an individual person’s information from at least two sources together for a specific purpose is called data linkage.

Many organizations collect data (or information) to do their business. For instance, the Workplace Safety and Insurance Board (WSIB) gathers information while managing compensation claims and collecting employer insurance premiums.

This data – sometimes called secondary or administrative data – is useful for research.

But when you have one source of data (such as from the WSIB) and you can link it to another source of data, the linkage can become more fruitful in answering questions. The data linkage can potentially generate new knowledge about work and health issues.

Data linkage is not new to the Institute for Work & Health (IWH). Researchers have used data linkage for specific research projects. One such project linked WSIB claims for motor vehicle collisions with motor vehicle accident reports maintained by the Ontario Ministry of Transportation. The linkage was based on the drivers’ gender, date of birth, and the accident date. The WSIB claims included information on the drivers’ occupation and the type of injury. The Ministry’s records included information on the collision including the location, number and type of vehicles involved, weather and road conditions and speed. This linkage allowed researchers to describe the context of work-related injuries due to motor vehicle collisions.

In another project, staff from Statistics Canada – with the participation of IWH researchers – linked a 15 per cent sample of Canadians who completed the long form of the 1991 Canadian Census to the Canadian Mortality Database for the years 1991 to 2001. The Census provides self-reported information on income, education, occupation, industry, labour force participation and disability. The Canadian Mortality Database contains copies of death registrations documented by provincial vital statistics registrars and includes information on the date and cause of death. This linkage allowed researchers to describe which causes of death were more or less common among specific occupations in Canada and to address questions about the interplay of disability, labour market participation and mortality. Without this data linkage, this important work could not have been possible.

Privacy and confidentiality issues

Ideally, researchers should obtain individual consent from people whose information is being used in research. However, it may not always be feasible to obtain consent from people whose information is recorded in large administrative databases. Privacy protection legislation accommodates the use of administrative data and data linkage for research purposes without individual consent if it would be impractical to obtain consent, if the potential benefits outweigh the potential harms and if the research could not be conducted in any other way.

When the Institute is planning a study using a data linkage, IWH researchers submit the study protocol for ethical review to a Research Ethics Board at an organization such as the University of Toronto. The submission includes the benefits of conducting the research, the risks involved and the safeguards that are in place to protect the data and the confidentiality of the subjects.

In all cases, individual-level data would never be presented in any report or paper – only aggregate-level data summaries would be reported.

Data linkage is a useful and valuable resource for researchers. However, data linkage must be done in a manner that adheres to privacy, confidentiality and ethical rules.

To read the Institute’s privacy policies, visit: www.iwh.on.ca/privacy-policies

Source: At Work, Issue 60, Spring 2010: Institute for Work & Health, Toronto