Importance: Palliative care for patients with advanced cancer is often introduced late in the disease course. Whether earlier referral influences patterns of healthcare utilisation and associated spending near the end of life is an important health policy question.
Objective: To estimate differences in healthcare utilisation and associated measured acute-sector spending between earlier and later palliative care referral among adults who died from cancer.
Design: Retrospective population-based cohort study using linked administrative health data.
Setting and Participants: Adults aged 18 years or older who died of cancer-related causes in Victoria between 1 January 2018 and 31 January 2023, identified through the Victorian Cancer Registry and linked to routinely collected administrative health records (n = 38,697 palliative care recipients; 17,409 early, 21,288 late).
Exposure: Earlier palliative care, defined as first receipt more than 90 days before death, or within 6 weeks of diagnosis when survival after diagnosis was less than 90 days, compared with later referral.
Main Outcomes and Measures: Healthcare utilisation in the final 90 days of life, including acute hospitalisations, chemotherapy services, and emergency department presentations. Associated measured acute-sector spending was estimated using activity-based pricing. Adjusted differences were estimated using inverse probability of treatment weighting.
Results: Among 38,697 decedents who received palliative care (17,409 [45.0%] earlier referral; 21,288 [55.0%] later referral), earlier referral was associated with lower measured acute-sector healthcare spending in the final 90 days of life (–$3,858 per person; 95% CI, –$5,131 to –$2,592), equivalent to approximately 14% of mean measured spending over this period. Differences were driven primarily by lower hospitalisation and chemotherapy use.
Conclusions and Relevance: Earlier palliative care referral was associated with lower use of acute hospital-based services and lower measured acute-sector spending near the end of life. These findings suggest referral timing is a modifiable system-level lever that may reallocate care.