Harnessing the potential of European routine data
During our work tracking down datasets from across the globe, we have identified datasets from nearly every country in Europe. This highlights the breadth and depth of ongoing research. With such a broad mix of countries, it is unsurprising the types of studies found are equally varied and exciting. It has been particularly interesting to find not only cohort studies, but that routinely collected data such as medical records, have also been integral in advancement of population health research.
Routinely collected data are an essential resource for organisations such as hospitals, schools, and the government where the primary objective is not research.These bodies use data for registration, transaction and record keeping facilitating the delivery of services. For example, patient records are created to enable a clinician to record diagnoses, prescriptions, referrals, and tests requested.
Linking any pre-existing and routinely collected information can provide invaluable opportunities for researchers to examine health outcomes and disease patterns over time in huge and diverse populations.
How can we benefit from using routinely collected data for research purposes?
These data are well suited to longitudinal analysis and can help create exceptionally large datasets, with huge volume of information on multiple participants.
Using existing data can be less expensive in terms of financial resources and time.
These datasets can potentially cover large portions of the population, which can be particularly useful for populations that are typically less accessible.
This may also help to reduce the burden on respondents and are less affected by declining response rates, attrition, and loss to follow up, compared to typical longitudinal surveys.
Why are there so many registers across Europe?
Europe is home to countries leading the way in using routine data for health research. Sweden, along with other Nordic countries, are particularly strong in this field as each of these countries have welfare state models with universal and tax-funded health care systems, population-based nationwide registries and a personal identity number assigned to each citizen from birth. This offers researchers the unique opportunity to not only link questionnaire data to administrative registers about demographics, employment, and health, but also allow for long-term follow up.
In Denmark, for example, physicians, pharmacists, and hospitals all communicate using a centralised database since 2010. Each patient can be tracked longitudinally using their own personal identification number. In countries with the most developed data linkage, they all use a unique number to identify individuals and link data together. By linking a person’s education and healthcare records, we can examine the relationship between health and educational attainment.
The UK (United Kingdom) also benefits from a centralised medical database, the NHS (National Health Service), and progress is being made to map onto the work from our Nordic neighbours in terms of data linkage. NHS data can be used to create deidentified, anonymous datasets for research on health and disease. The holders of these data apply a strict statistical disclosure control in accordance with the NHS Digital protocol to ensure confidentiality is maintained.
The future of data linkage
Administrative health systems data, such as UK Hospital Episode Statistics, can provide a view of patient interaction with healthcare services, to understand treatment success and health trajectories. UK administrative data is currently a largely untapped, but information-rich, resource. This wealth of data, the majority of which was not originally created for research but as a by-product of government services, has the potential to provide powerful insights into our society and in turn point to areas where change is needed.
There have also been huge improvements across the globe. It was very difficult until recently for researchers in countries like Australia to access government-linked data. However, in recent years, the Australian government has encouraged open access to administrative data. This is helping to allow for a more clearly defined process through which researchers can access de-identified data.
What are the challenges of using routinely collected data?
Limited coverage of information which can only be used to construct proxy indicators in social science research.
These data can be affected by changes in government legislation, potentially making data less omparable over time.
Statistical and methodological difficulties when working with these data as they have not been collected for research. These data often require extensive cleaning and correctly linking participants can be challenging and time consuming.
Multiple barriers to accessing data.
Although these data can access populations that are potentially less represented in research, certain groups may still be missing, for example, those unable to access government services such as the homeless.
Harnessing the power of data across Europe
Longitudinal cohort data collected for research can offer detailed measures of complex constructs, such as depression, anxiety, and psychosis across time. The opposing benefits and disadvantages across both routine data and purposely collected longitudinal cohort data demonstrates the need for both types across the globe. What routine data offers in breadth; cohort data offers in depth. Routine data can be a fantastic way of harnessing huge volumes of existing data on a scale previously unseen in mental health research.
Researchers need options for data to find the most appropriate source for their research, rather than a one size fits all approach. We must continue to move forward towards accessing, sharing, and linking routine data in a safe and confidential manner whilst continuing to invest in fantastic longitudinal cohort data, offering detailed rich measures over time.
As we continue our quest to find the most promising longitudinal datasets for mental health research, we are asking for your help. If you know of a dataset not identified, please let us know, particularly of any outside academia!
Follow @L_Arseneault, @wellcometrust, @MQmentalhealth, and @ODIHQ on Twitter. Email us at landscaping-wellcome@kcl.ac.uk with any studies or information that may be useful for us to consider
