Using administrative health data for palliative and end-of-life care research in Ireland

Administrative health data is under-used as a research tool for palliative care research in Ireland. The overarching aim of this thesis was to explore the potential of population based cancer registry data linked to hospital episode data and death certificate data to examine palliative and end-of-life care (PEoLC) for cancer patients, in Ireland. Research objectives include using the available linked data to identify vulnerable subgroups in need of palliative care, to identify and compare characteristics of those who receive palliative with those who do not and to explore the relationship between receiving palliative care and place of death. The knowledge gained was extrapolated to other non-cancer health and social care data collections to evaluate the potential and challenges of these data for PEoLC research and so develop guidelines to maximise its potential as a research tool while identifying the limitations of its use. A population based study of lung cancer decedents found those patients who die within 30 days of diagnosis (short-term survivors) were older, (aged 80 years and over), had more comorbid disease and were more likely to present through the emergency department than those who survived longer. These characteristics are available at diagnosis and could be used to guide decisions on early assessment for palliative care for lung cancer patients at admission. (Short-term survivors) are more likely to die in hospital so reporting place of death by survival time may be useful to evaluate interventions to reduce deaths in acute hospitals. A second study compared the characteristics and place of death of cancer patients receiving specialist palliative care in acute hospitals with those who do not. Almost two thirds of cancer patients who attended a cancer centre in 2016 and died in 2016 had an inpatient palliative care encounter. They were younger, less likely to be married, and more likely to be from a deprived area. Having accounted for sociodemographic factors, there was evidence of regional variation in receiving palliative care. Place of death differed by palliative care encounter, 45% of those who were seen by the palliative care team died in hospital, 33% died in a hospice and 18% died at home. Of those who had no record of a palliative care encounter 50% died in hospital, 16% died in hospice and 28% died at home. These studies demonstrate that Irish administrative health data can be used for PEoLC research however a thorough advance knowledge of the datasets to be used is critical. For each data set, information on how the data are organised (data models) and what data are collected (data dictionaries) are required. Knowledge of what is missing within each dataset is as important as knowing what is available. These considerations inform most aspects of study planning, design and implementation and were used to evaluate the potential of other national and social care data collections for PEoLC research. Efforts to identify and control for bias, both known and unknown are a particular concern when using administrative health data for research. This thesis contributes to population based PEoLC research in Ireland and to the international body of work that uses administrative health care and social care data for PEoLC research. It can inform future use of Irish health data for population based research and has identified several avenues for further research. It is hoped the thesis findings will give impetus to ongoing initiatives to improve the research potential of Ireland’s health and social data collections