Professor John Mathews, Executive Director of the Menzies Foundation, writes below that important benefits have come from research based on data linkage, including this recent study, which found an increased cancer risk for children and adolescents exposed to radiation from CT scans.
But there are significant barriers to the wider use of data linkage – an issue of concern also identified by a recent Productivity Commission paper, “Deep and Persistent Disadvantage in Australia”, which suggested that better use could be made of administrative data sets held by governments.
Making the case for better use of data linkage
John Mathews writes:
To understand most of the health problems that we face, it is necessary to study the pooled experiences of large numbers of patients with the same problem.
For example, to understand whether deep venous thrombosis (DVT or blood clots in the leg) was caused by long haul air travel, researchers accessed de-identified government records of those people arriving into Perth airport from overseas who were subsequently admitted to hospital in WA with DVT.
This research showed that the risk of DVT was increased in the two weeks after the flight, but was back to baseline by the third week after arrival.
Similarly, to identify which prescribed medicines can cause birth defects if taken during early pregnancy, researchers in WA examined de-identified PBS records and birth records to determine which medications were more likely to be followed by a birth defect.
Each of these studies depended on the linkage of information from different data sources (eg Perth airport arrivals linked with WA hospital admissions, or PBS medications linked with birth records). During linkage, relevant information for each person is taken from the several different data sources, and combined in a ‘linked research file’ from which personal identifiers have been removed to protect privacy.
Data linkage has great potential to improve health, hitherto largely untapped. Along with the more routine research analysis of single data-sets, the analysis of linked data-sets will help to identify previously unsuspected risks to health, improve health outcomes for many people, and improve the efficiency and effectiveness of the health system as a whole.
Australia spends at least $100 billion each year on health, including significant sums on the electronic recording of health services, PBS medications, health outcomes, and other relevant data.
Unfortunately, most data-sets are not easily accessible for public interest research. The Australian community has a right to expect that all the data collected on its behalf are being well used for public-benefit purposes, and are not being unnecessarily locked up by governments.
State-based record linkage was pioneered by WA, with many important projects since the late 1960s, carried out under stringent ethical guidelines to protect privacy. In 1987, all states and territories agreed to partner with the Commonwealth in establishing the Australian Institute of Health and Welfare to oversight and report on national health statistics, while the Population Health Research Network (PHRN), established in 2009, has facilitated data linkage with all states and territories.
Nevertheless, in our federal system, there are still obvious impediments to data sharing and data linkage. For national projects requiring access to unit data it is still necessary for researchers to obtain approvals from multiple data custodians and ethics committees acting for Commonwealth agencies, all states and territories, and multiple institutional ethics committees.
For example, to examine whether childhood exposure to low-dose radiation from CT scans (medical X-rays) would increase the risk of cancer in later life, more than 20 separate approvals were necessary, and it took almost five years for the Commonwealth to agree to release de-identified national records of Medicare-funded CT scans.
The results of that study, recently published, show an average increase of 16% in subsequent cancer risk for every CT exposure before the age of 20 years.
Exploring the barriers
Despite the efforts of PHRN, many researchers planning data-linkage projects are still experiencing long delays with approvals from government data custodians, or report that they have had to abandon projects altogether.
How has this impasse to be explained? Although members of the public, data custodians and governments need re-assurance that privacy will be maintained, the legal requirements and ethical guidelines for privacy protection are well developed, and there is no instance of any privacy breach resulting from data-linkage in Australia to justify the delays with research approvals.
What other factors might be at work? Perhaps some in government believe that if information is available for research use by persons external to government, the results may reflect adversely on policy in one or other jurisdiction, identify new health risks with budgetary implications, or otherwise weaken their control of the policy agenda.
However, the (bipartisan) public interest position should be to welcome research that might inform policy and practice. This implies that the default position should be for transparency ie for release of de-identified data for research purposes unless there are strong countervailing reasons such as national security, or unless the research proposal is judged to be of insufficient quality or relevance by research-funding agencies such as NHMRC.
NHMRC is already working behind the scenes with researchers and with other government agencies to facilitate improved arrangements for data sharing and data-linkage in Australia. However, because NHMRC is itself an agency of government, it is not easily able to explore the potential conflicts of interest that are blocking the sharing and linkage of data, and thus impairing the productivity of health researchers and of the health system as a whole.
To promote a wider public discussion, and to canvas principles that could guide our nation forward, the Menzies Foundation is hosting a Data Linkage Workshop in Melbourne on August 16, to be attended by community and media representatives, leading researchers, senior public servants and members of parliament.
The Foundation would welcome suggestions from Croakey readers about how to engage with other interested members of the public.
We would also like to hear your suggestions about how to consolidate community and bipartisan political support for the idea that the sharing and linkage of de-identified data for public benefit research, as a routine activity, will greatly improve our health and social outcomes, and improve the overall productivity of the health system.
• Professor John Mathews is Executive Director of the Menzies Foundation and a Professorial Fellow at the University of Melbourne.
PostScript from Croakey
As mentioned, the potential for wider use of data linkage to improve understanding and policy around disadvantage was also highlighted recently by the Productivity Commission. Its paper, “Deep and Persistent Disadvantage in Australia”, says:
“Government agencies, at all three levels of government, hold very large administrative datasets which may assist in unlocking a deeper understanding of factors influencing disadvantage, the government programs that are accessed by those experiencing disadvantage, and how those programs assist (or hinder) those who are the most vulnerable.
Administrative data can cover long periods of time and also largely address the problem of existing survey bias where deeply and persistently disadvantaged groups are poorly represented in surveys.
However, in comparison with other countries, the accessibility of administrative data to researchers in Australia is limited.
There are a number of departmental datasets that could potentially shed light on the characteristics and pathways of people with multiple disadvantages.”
• Declaration: The Menzies Foundation will be covering Melissa Sweet’s expenses to attend the Data Linkage Workshop.