Identifying clusters of healthcare expenditure trajectories in end-stage organ disease: a retrospective cohort study using linked administrative databases in Singapore.

OBJECTIVES: We aimed to identify subgroups of decedents with end-stage organ disease based on their healthcare expenditure trajectories over the final five years of life, and factors associated with incurring high costs in this period. METHODS: We conducted a retrospective cohort study of patients who died between 2017 and 2019, who had either a primary or secondary diagnosis of advanced dementia, severe liver disease, as well as heart, kidney or respiratory failure in the last five years of their lives. Data was extracted from administrative databases of a regional health system in Singapore. We classified patients into subgroups with distinctly different five-year healthcare expenditure trajectories, using longitudinal k-means clustering. Factors associated with membership in each subgroup were then identified through multinomial modelling. RESULTS: Among 7,154 decedents, three trajectories of consistently low cost (LC, n = 5,756), moderately high cost near death (MC, n = 1,283) and escalating cost near death (EC, n = 115) were identified. Patients with MC often had concurrent end-stage organ diseases and chronic conditions, while patients with EC were younger and often had respiratory failure. Across all subgroups, most patients had at least one emergency department attendance in their final three months and either late or no access to palliative care. Accounting for socio-demographic characteristics and comorbidity, patients with respiratory failure were more likely to have a high-cost trajectory. CONCLUSIONS: Our findings highlight the significance of upstream interventions to address needs of multi-morbid older adults, and pertinence of timely palliative care access to mitigate the use of aggressive care at the end-of-life.