Routinely collected data are increasingly used in healthcare research. These data originate from different sources (e.g. reimbursement claims, hospital discharge reports, electronic medical records, primary care databases, population census, death certificates), and can sometimes be linked at the individual level to other routinely collected data or even to research-oriented datasets (e.g. population-based cohorts, panel surveys, disease-specific registries, data collected through wearable devices). The availability, the coverage, and the quality of these sources of information vary across countries.
Until the beginning of the 2000s, end-of-life care researchers have mostly relied on retrospective surveys of physicians or caregivers, chart reviews, and questionnaire-based studies. Some, however, had already started to use vital statistics and claims data to investigate the determinants of place of death and the utilization of medical services at the end of life.
During the following decade (2000–2010), several sources of routinely collected data have been used extensively to further analyze patterns of end-of-life care: national causes of death registers; SEER-Medicare administrative databases; point-of-care data collected through various palliative care collaborations (e.g. Australia).
More recently, a number of studies have been published based on large healthcare registers linked to multiple data sources, with the possibility to conduct longitudinal analyses and to diversify study designs (including prospective cohorts). New technologies are opening unprecedented opportunities to use routinely collected data for end-of-life care research, for instance using wearable devices and relying on real-time patient-reported outcomes. Most healthcare organizations are moving towards fully-digitalized patient files, with ever-increasing amounts of clinical and biological information being recorded in real time. In many countries, healthcare insurances and national health authorities have assembled large datasets based on record-linkage of claims, drug prescriptions, hospital discharge reports and vital statistics, sometimes connected to sociodemographic and patient-reported data. Computation power, data storage and statistical methods are continuously improving, offering new possibilities for analysis of very large datasets. Finally, data with national coverage considerably reduce the uncertainty related to sampling errors (e.g. under-representation of people with lower SEP) and selection bias (e.g. non-participation, attrition during follow-up), and give the opportunity to conduct cross-country comparisons with large populations at a limited cost.
However, use of routinely collected data in end-of-life care research also comes with potential pitfalls. This is particularly evident when the goal of data mining is to predict clinical outcomes near the end of life.
Moreover, important ethical issues arise from the availability and the linkage of such comprehensive data covering – in some instances – the full population in a given countries or region. The absence of explicit informed consent from the study participants, and the need to ensure the protection of personal data are important problems that need to be addressed. The recent introduction of the General Data Protection Regulation (GDPR) within the European Union will for instance have substantial repercussions on the availability of registry-based data for researchers.
Comprehensive overview of the available evidence is warranted to inform researchers about the opportunities and the challenges that come with routinely collected data in end-of-life care research, and to raise awareness about the need to ensure the quality, the reproducibility, the comparability, and the relevance of future studies.
The taks force will
- Systematically review and describe the available evidence;
- Write a position paper about the main challenges and opportunities that come with big data in end-of-life care research;
- Develop and publish consensus-based recommendations for conducting and reporting end-of-life care research using routinely collected data;