Warfarin Prescribing Practices Nearing End of Life

Indications for anticoagulant therapy—most commonly venous thromboembolism (VTE) and atrial fibrillation (AF)—increase with age and the presence of comorbidities. However, clinicians often face complex decisions when considering anticoagulation in patients with limited life expectancy. These individuals may have goals that prioritize comfort and reduced healthcare contact, such as avoiding hospitalization or frequent monitoring, rather than long-term prevention of thrombotic events, especially when the absolute risk reduction is probably modest (e.g., 2–3% or less/year). They also tend to have higher bleeding risks, which should be balanced against potential therapeutic benefits.[ 1 ] [ 2 ] Hence, the importance of patient-centric anticoagulation care protocols, especially relevant to groups with limited life expectance, such as cancer.[ 3 ] Furthermore, such patients have been under-represented in anticoagulant trials, leaving uncertainty about therapeutic benefits important to these patients.[ 4 ] Vitamin K antagonists (VKAs), while having efficacy to prevent thromboembolism, are associated with a generally higher bleeding risk compared with direct oral anticoagulants (DOACs) and require regular laboratory monitoring of the international normalized ratio (INR) due to a narrow therapeutic range. This INR monitoring and associated VKA dose adjustments can be burdensome, especially for patients with functional or cognitive impairment that may be common in those with life-limiting illnesses. In this issue of Thrombosis and Haemostasis , Kempers et al[ 5 ] analyzed data from 18,145 Dutch patients prescribed VKAs between 2013 and 2019 who also had a life-limiting condition. The authors examined trends in anticoagulant continuation versus discontinuation and associated hospitalizations due to bleeding or thrombosis. Data were drawn from five anticoagulation clinics, the Netherlands Cancer Registry, and hospital databases. Life-limiting illnesses were defined as conditions with a median life expectancy of 2 to 4 years based on national statistics, including some patients with liver disease, dementia, heart or lung disease, complicated diabetes mellitus, hip fracture in patients over 70 years, and cancer. This study is part of a European collaboration, called the SERENITY project, which aims to create a shared decision tool to aid anticoagulation decisions at the end of life.[ 6 ] Most patients (78%) in the study were anticoagulated for AF, and the commonest life-limiting illness included was heart disease (60%). Notably, anticoagulation was continued until death in 69% of patients. Rates of hospitalization for thrombosis were low (<1%/100 patient-years for VTE, <2% for stroke), as were bleeding-related hospitalizations (<3%/100 patient-years), regardless of whether VKAs were continued. Although causality cannot be inferred—since clinicians likely adjusted treatment based on individual risk—these low event rates suggest that hospitalization for these outcomes is overall low, emphasizing the importance of patient preference in the decision-making process. The study has several limitations inherent to researching this patient population. First, individuals with life-limiting conditions are heterogeneous in terms of thrombotic and bleeding risk, comorbidities, and willingness to undergo medical interventions, and uniform recommendations for this group are unlikely. Second, predicting prognosis in chronic illnesses is notoriously difficult, which contributes to the relatively wide variability in survival across patients in this study and somewhat limits its applicability. Finally, randomized trials in this group are challenging to conduct, meaning clinicians often rely on retrospective observational data such as this which is subject to confounding and selection bias.[ 7 ] This research was partially conducted during the pre–DOAC era. Current guidelines now recommend DOACs over VKAs as first-line therapy for both VTE and AF.[ 8 ] [ 9 ] While DOACs are appealing for patients with limited life expectancy due to their lower bleeding risk and lack of need for regular laboratory monitoring, this also means they may be continued more often in the absence of clear benefit.[ 10 ] Another study conducted within the SERENITY collaboration, which included Welsh patients with cancer and an even more limited life expectancy receiving DOACs, found that rates of anticoagulation continuation until death remained similarly high.[ 11 ] Ultimately, decisions about anticoagulant therapy in patients with limited life expectancy require a patient-specific approach that weighs bleeding risk, thrombotic risk, life expectancy, and the potential impact of thrombotic events on quality of life. Most importantly, patient values and preferences should play an important role, given the absence of high-quality data to guide decisions about the putative benefits and risks of anticoagulation and, as shown by this study, the low adverse event rates irrespective of whether anticoagulation is continued or de-prescribed. For instance, preventing a disabling ischemic stroke in AF may carry more importance for some patients than preventing VTE. Additionally, patients with advanced cancer may prefer avoidance of therapies such as anticoagulants that do not directly improve symptom control or overall quality of life. Kempers et al[ 5 ] found that most patients remained anticoagulated until death. Even in cases where anticoagulation was de-prescribed, this was only done at a median duration of 14 days before death. These findings suggest that opportunities for timely de-prescribing are frequently missed. To summarize, this study is an important step forward to re-evaluate anticoagulant management decisions in patients with limited life expectancy who have not received the requisite study. Additional studies are needed, particularly in DOAC-treated patients, as well as in other populations and ethnic groups, the latter being an important consideration given the reported racial differences in bleeding risks.[ 12 ] [ 13 ] Publication History Received: 07 May 2026 Accepted: 07 May 2026 Article published online: 22 May 2026 © 2026. Thieme. All rights reserved. Georg Thieme Verlag KG Oswald-Hesse-Straße 50, 70469 Stuttgart, Germany