Predictors of antithrombotic deprescribing in older patients with advanced cancer and limited life expectancy in hospice care: a retrospective cohort study

Abstract Introduction Older patients with advanced cancer often continue antithrombotic therapy at the end of life, despite limited benefit and an increased risk of harm. In hospice care, continuation of antithrombotic agents may contribute to medication burden and increase bleeding risk.[1,2] Structured deprescribing could optimise care; however, evidence from real-world hospice populations remains limited. This study evaluated a medication review intervention implemented by a palliative care consultation (PCC) team for hospice patients. The intervention involved clinical assessments of antithrombotic use, considering factors such as bleeding risk, prognosis, drug–drug interactions, and alignment with goals of care to inform deprescribing decisions. Aim To identify predictors of antithrombotic deprescribing in older patients with advanced cancer and limited life expectancy in hospice care. Methods A retrospective cohort study was conducted among patients aged ≥65 years who were admitted to an inpatient hospice unit in Jordan between 2022 and 2023. Medical records were reviewed, and data were extracted. Inclusion criteria were diagnosis of advanced cancer, estimated life expectancy of ≤six months, prescription of at least one antithrombotic agent at hospice admission, and documentation of at least one assessment by the PCC team during admission. Patients who had antithrombotic agents discontinued but without a documented rationale were excluded (n = 43). Data were extracted by a healthcare professional involved in direct patient care and were anonymised before analysis. Extracted variables included demographic information and medication use. Multivariable binary logistic regression was performed to identify independent predictors of antithrombotic deprescribing. Results Among 112 older patients (mean age 74.0 ± 9.0 years), 58% were female and 86.6% were estimated to have less than three months of life expectancy. At hospice admission, 50% of patients had three or more comorbidities, 84.8% were prescribed six or more medications, and most patients (83%) were referred from inpatient wards. A total of 146 antithrombotic agents were prescribed [enoxaparin (52.1%), aspirin (24%), fondaparinux (13%), clopidogrel (10.3%), and warfarin (0.7%)] at admission. The mean number of antithrombotic agents decreased significantly from 1.30 (±0.58) at admission to 0.62 (±0.63) after the PCC team-led intervention (P < 0.001), with 56.3% of patients having at least one antithrombotic agent deprescribed. In total, 77 antithrombotic agents were deprescribed [enoxaparin (46.8%), aspirin (31.2%), clopidogrel (13%), and fondaparinux (9.1%)]. Of these, 67.5% were deprescribed due to high bleeding risk, while 32.5% were deprescribed for reasons such as drug–drug interactions or limited life expectancy. Being prescribed two or more antithrombotic agents was the only significant predictor of non-deprescribing (OR = 0.198; 95% CI: 0.060–0.655; P = 0.008). Variables such as life expectancy and site of referral were not statistically significant. The model had a good fit (Hosmer–Lemeshow P = 0.764). Conclusion Medication complexity, particularly the use of multiple antithrombotic agents, may hinder deprescribing decisions—potentially reflecting clinical uncertainty regarding thrombotic risk. These findings highlight the need for structured deprescribing protocols to improve end-of-life care. Strengths of this study include the use of real-world hospice data and multivariable modelling; limitations include its retrospective, single-centre design. Future multicentre and qualitative studies are warranted to support implementation and better understand clinician decision-making.