Best Evidence Topic Report: Should Anticoagulant Be Initiated in Patients with Sepsis-Induced New-onset Atrial Fibrillation?
Date First Published:
November 29, 2025
Last Updated:
November 29, 2025
Report by:
Dr Khunassanan Nunthakunatip, Foundation Year 2 Doctor (Manchester University NHS Foundation Trust)
Search checked by:
Dr Sofia Borhan, Foundation Year 2 Doctor
Three-Part Question:
patients who presents to ED with sepsis-induced new-onset AF therapeutic anticoagulation a lower stroke risk and/or a higher bleeding risk
Clinical Scenario:
An 88-year-old man from a care home arrives at the Emergency Department with abnormal vital signs indicating septic shock. Assessment reveals pneumonia and new-onset atrial fibrillation (AF). The patient’s relatives are concerned about stroke risk from AF and ask if anticoagulation should be started immediately. The clinical team explains this is a complex decision in sepsis-induced new-onset AF due to uncertain evidence, and will review current literature to guide shared decision-making.
Search Strategy:
Ovid MEDLINE (1946 to May 22, 2025) and Embase (1974 to May 22, 2025) were searched using a combination of MeSH terms and free-text keywords: (atrial fibrillation$.ti,ab. OR AF.ti,ab. OR *Atrial Fibrillation/) AND (sepsis.ti,ab. OR *Sepsis/) AND (anticoag$.ti,ab. OR *Anticoagulants/).
Search Details:
Sepsis was defined by the ICD-10 code A 41.9 during the search. The search was limited to English-language articles. Duplicate results were removed via Ovid’s deduplication tool. A supplementary search was performed using Google Scholar’s ‘Cited by’ function and the reference lists were screened to identify papers that might have been missed during our search process.
Inclusion criteria:
- Hospitalised patients
- Sepsis-induced new-onset AF
- Anticoagulation initiated during admission or within 30 days of discharge
- Study reported stroke risk
Exclusion criteria:
- Pre-existing AF or non-sepsis-induced AF
- Non-peer-reviewed studies or non-empirical articles (e.g. review, commentary, opinion piece or case study)
- Conference abstracts
Inclusion criteria:
- Hospitalised patients
- Sepsis-induced new-onset AF
- Anticoagulation initiated during admission or within 30 days of discharge
- Study reported stroke risk
Exclusion criteria:
- Pre-existing AF or non-sepsis-induced AF
- Non-peer-reviewed studies or non-empirical articles (e.g. review, commentary, opinion piece or case study)
- Conference abstracts
Outcome:
The database search yielded 254 results. After removing fifteen non-English articles and thirty-eight duplicates, 201 papers were screened. Our supplementary searches did not identify any further relevant papers. Two reviewers independently assessed titles, abstracts, and full texts using predefined inclusion and exclusion criteria.
Conference abstracts (n = 147), case reports (n = 5), editorials (n = 2), letters (n = 1) and review articles (n = 5) were excluded. Twenty-four papers were excluded based on their title and/or abstract. Full text review of the remaining seventeen papers was carried out and thirteen were excluded.
The final four papers were included in our analysis (Table 1).
Conference abstracts (n = 147), case reports (n = 5), editorials (n = 2), letters (n = 1) and review articles (n = 5) were excluded. Twenty-four papers were excluded based on their title and/or abstract. Full text review of the remaining seventeen papers was carried out and thirteen were excluded.
The final four papers were included in our analysis (Table 1).
Relevant Paper(s):
| Study Title | Patient Group | Study type (level of evidence) | Outcomes | Key results | Study Weaknesses |
|---|---|---|---|---|---|
| Bleeding, stroke, and mortality risk of patients with septic shock receiving anticoagulation for atrial fibrillation Sahitya Allam, MD,1 Jonathan Na, MD,1 Joanne Moon, MD,1 Yash Desai, MD,1 Christopher Messner, MD,1 Robert Potenza, MD,1 Mark Sonbol, MD,1 Faisal Abushullaih, MD,1 Abdullah Aljudaibi, MD,1 Maria Abraham, MD,1 Kevin Chen, MD,1 Ethan Kotloff, MD,1 Simin Hossain, MD,1 Elnaz Esmati, MD,1 Thomas Kutner, MD,1 Gregory Norcross, MD,1 James Childress, MD,1 Paul Han, MD,1 Ian Welch, MD,1 Michael Sokolow, MBA,2 Vincent See, MD, MS,3 Libin Wang, MD, PhD3 2024 United States | N* (post-PSM) = 112 . Male = 61.7% . Mean Age = 64.9. Adults over the age of 18 who were admitted to medical intensive care units with a diagnosis of sepsis and atrial fibrillation between 2016 and 2020 at the University of Maryland Medical System in Baltimore. | Single-center, retrospective observational cohort study (Level 3) |
Ischemic stroke risk (post- PSM) = No significant difference. | Ischemic stroke risk (post- PSM) = No significant difference. | Single-centre design conducted in the USA limits generalisability. The very low number of stroke events (only 3 cases in total) indicates that the study was underpowered to detect a difference, increasing the risk of a type II error. Observational design means the study remain prone to selection bias, despite use of propensity score matching. Only in-hospital outcomes assessed; long-term stroke and bleeding risks not captured. |
| (0.89% AC vs 1.79% no AC; P = 0.56). | (0.89% AC vs 1.79% no AC; P = 0.56). | ||||
| Total bleeding risk (BARC** 1-5b; post-PSM) = No significant difference. (RR 1.60; 95% CI 0.72–3.54). | Total bleeding risk (BARC** 1-5b; post-PSM) = No significant difference. (RR 1.60; 95% CI 0.72–3.54). | ||||
| Mortality rate (post-PSM) = Significant reduction in mortality in AC group. (46.4% in AC vs 66.1% in no AC; P = 0.04). | Mortality rate (post-PSM) = Significant reduction in mortality in AC group. (46.4% in AC vs 66.1% in no AC; P = 0.04). | ||||
| Practice Patterns and Outcomes Associated With Anticoagulation Use Following Sepsis Hospitalizations With New-Onset Atrial Fibrillation Allan J. Walkey , MD, MSc; Laura C. Myers , MD, MPH; Khanh K. Thai , MS; Patricia Kipnis , PhD; Manisha Desai , PhD; Alan S. Go , MD; Yun Lu , MPH; Heather Clancy, MPH; Ycar Devis, MPH; Romain Neugebauer, PhD; Vincent X. Liu, MD, MS 2023 United States | N* = 3,992. Male = 53%. Mean age = 78. Adults aged 40 years and older who were hospitalised with sepsis, developed new-onset atrial fibrillation, and survived to hospital discharge between 2011 and 2018 across 21 hospitals within the Kaiser Permanente Northern California healthcare system. | Multi-centre, retrospective observational cohort study (Level 2) |
Ischemic stroke/TIA risk (1-year adjusted risk) = Significant higher risk of stroke/TIA in AC group (5.69% in AC vs 2.32% no AC; risk difference 3.37%, 95% CI 0.36–6.38). | Ischemic stroke/TIA risk (1-year adjusted risk) = Significant higher risk of stroke/TIA in AC group (5.69% in AC vs 2.32% no AC; risk difference 3.37%, 95% CI 0.36–6.38). | Study population limited to an insured US cohort (Kaiser Permanente) which limits generalisability. Stroke outcome measured using administrative coding, which may lack clinical specificity and sensitivity. The observed increase in stroke risk among anticoagulated patients may reflect selection bias (i.e. patients at higher risk were more likely to be anticoagulated). Adherence to oral anticoagulation was not measured, and dose/duration were not verified. |
| Major bleeding risk (1-year adjusted risk) | Major bleeding risk (1-year adjusted risk) | ||||
| Defined by hospitalisation with diagnosis of intracranial haemorrhage and gastrointestinal haemorrhage identified using validated algorithm. = No significant difference (6.51% in AC vs 7.10% no AC; risk difference −0.59%, 95% CI −3.09 to 1.91). | Defined by hospitalisation with diagnosis of intracranial haemorrhage and gastrointestinal haemorrhage identified using validated algorithm. = No significant difference (6.51% in AC vs 7.10% no AC; risk difference −0.59%, 95% CI −3.09 to 1.91). | ||||
| Mortality rate (unadjusted) = 10.4% vs 29.1% (Absolute difference 18.7%) | Mortality rate (unadjusted) = 10.4% vs 29.1% (Absolute difference 18.7%) | ||||
| Anticoagulant Use and Risk of Ischemic Stroke and Bleeding in Patients With Secondary Atrial Fibrillation Associated With Acute Coronary Syndromes, Acute Pulmonary Disease, or Sepsis Michael J. Quon, MD,a Hassan Behlouli, PHD,b Louise Pilote, MD, MPH, PHD 2018 Canada | N* = 102 . Male = 47% . Mean age = 77.1. Adults aged 65 years and older with sepsis and newly diagnosed atrial fibrillation, recruited between 2009 and 2012 from 33 intensive care units across 24 hospitals participating in the U.S. Critical Illness and Injury Trials Group–Critical Illness Outcomes Study (USCIITG-CIOS) network. | Multi-center, retrospective, observational cohort study (Level 2) | Stroke risk = No significant difference | Stroke risk = No significant difference | The very low number of stroke events (only 6 cases in total) indicates that the study was underpowered to detect a difference, increasing the risk of a type II error. Clinical data were derived from administrative coding, which may risk misclassification. Observational design means results remain prone to selection bias. |
| (7.1% AC vs 5.5% no AC; OR 1.98, 95% CI 0.29–13.47). | (7.1% AC vs 5.5% no AC; OR 1.98, 95% CI 0.29–13.47). | ||||
| Bleeding risk | Bleeding risk | ||||
| Included intracerebral bleeding, gastrointestinal bleeding, intraocular bleeding, hematuria, hemoptysis, epistaxis, and unspecified location of bleeding. = No significant difference | Included intracerebral bleeding, gastrointestinal bleeding, intraocular bleeding, hematuria, hemoptysis, epistaxis, and unspecified location of bleeding. = No significant difference | ||||
| (17.9% AC vs 20.5% no AC; OR 0.96, 95% CI 0.29–3.21). | (17.9% AC vs 20.5% no AC; OR 0.96, 95% CI 0.29–3.21). | ||||
| Mortality = Not reported | Mortality = Not reported | ||||
| Practice Patterns and Outcomes Associated With Use of Anticoagulation Among Patients With Atrial Fibrillation During Sepsis Allan J. Walkey, MD, MSc, Emily K. Quinn, MS, Michael R. Winter, MPH, David D. McManus, MD, MSc, Emelia J. Benjamin, MD, ScM 2016 United States | N* = 5,585. Male = 49.2%. Mean age = 74.9. Adults aged 18 years and older who were hospitalised with sepsis and developed new-onset atrial fibrillation, recruited between 2011 and 2018 from 27 hospitals within the Kaiser Permanente Northern California (KPNC) healthcare system. | Multi-centered, observational, retrospective cohort study (Level 3) |
Stroke risk = No significant difference. | Stroke risk = No significant difference. | Only in-hospital outcomes assessed; long-term stroke and bleeding risks not captured. Oral anticoagulants excluded from primary analysis, limiting applicability to common real-world practice. Clinical data were derived from administrative coding, which may risk misclassification. Lack of reported p-values for key outcomes makes interpretation of the significance of these findings difficult. |
| Relative risk = (RR 0.85, 95% CI 0.57–1.27; P = 0.31) | Relative risk = (RR 0.85, 95% CI 0.57–1.27; P = 0.31) | ||||
| Bleeding risk | Bleeding risk | ||||
| Defined by hospitalisation with diagnosis of intracranial haemorrhage and gastrointestinal haemorrhage identified using validated algorithm. = No significant difference (RR 0.97, 95% CI 0.83–1.14). | Defined by hospitalisation with diagnosis of intracranial haemorrhage and gastrointestinal haemorrhage identified using validated algorithm. = No significant difference (RR 0.97, 95% CI 0.83–1.14). | ||||
| Mortality risk = Not reported | Mortality risk = Not reported |
Author Commentary:
The four observational studies included in this review specifically addressed whether anticoagulation in patients with sepsis-induced new-onset AF reduces the risk of stroke. None demonstrated a statistically significant reduction in stroke risk with anticoagulation. Due to the observational nature of these studies, clinicians may have been more likely to prescribe anticoagulation to patients perceived as suitable candidates. These were typically individuals with fewer contraindications and potentially better baseline status, introducing a selection bias that could favor detection of treatment benefit. Despite this favorable selection bias, no protective effect against stroke was observed. In fact, one study found that anticoagulation was paradoxically associated with a higher incidence of stroke. The authors attributed this to confounding by indication, as higher-risk patients were more likely to be anticoagulated.2
Bleeding risk was not consistently increased with anticoagulation after adjustment for confounders, but some studies did observe a higher rate of bleeding events, particularly in unadjusted analyses.3 The only study to report a significant reduction in in-hospital mortality among anticoagulated patients (Allam et al., 2024) likely reflects selection bias rather than therapeutic effect, as anticoagulation was more often prescribed to patients with better prognostic indicator.3 Given the lack of difference in stroke risk between groups, reduced mortality is unlikely to be explained by a reduction in thromboembolic events.
Methodological limitations temper these conclusions. All studies employed retrospective designs with heterogeneous sepsis definitions, variable anticoagulation regimens, and differing follow-up intervals. Competing risks—particularly early mortality after sepsis—may further obscure true treatment effects.
The consistent absence of stroke reduction, along with unclear bleeding profiles, challenges the assumption that anticoagulation strategies for primary AF apply to sepsis-induced AF. Until data from large, pragmatic, multicentre randomised controlled trials become available, routine anticoagulation for stroke prevention in this setting remains unsupported.
Bleeding risk was not consistently increased with anticoagulation after adjustment for confounders, but some studies did observe a higher rate of bleeding events, particularly in unadjusted analyses.3 The only study to report a significant reduction in in-hospital mortality among anticoagulated patients (Allam et al., 2024) likely reflects selection bias rather than therapeutic effect, as anticoagulation was more often prescribed to patients with better prognostic indicator.3 Given the lack of difference in stroke risk between groups, reduced mortality is unlikely to be explained by a reduction in thromboembolic events.
Methodological limitations temper these conclusions. All studies employed retrospective designs with heterogeneous sepsis definitions, variable anticoagulation regimens, and differing follow-up intervals. Competing risks—particularly early mortality after sepsis—may further obscure true treatment effects.
The consistent absence of stroke reduction, along with unclear bleeding profiles, challenges the assumption that anticoagulation strategies for primary AF apply to sepsis-induced AF. Until data from large, pragmatic, multicentre randomised controlled trials become available, routine anticoagulation for stroke prevention in this setting remains unsupported.
Bottom Line:
Current evidence available does not support routine anticoagulation for stroke prevention in patients with sepsis-induced new-onset AF.
Level of Evidence:
Level 2: Studies considered were neither 1 or 3
References:
- Sahitya Allam, MD,1 Jonathan Na, MD,1 Joanne Moon, MD,1 Yash Desai, MD,1 Christopher Messner, MD,1 Robert Potenza, MD,1 Mark Sonbol, MD,1 Faisal Abushullaih, MD,1 Abdullah Aljudaibi, MD,1 Maria Abraham, MD,1 Kevin Chen, MD,1 Ethan Kotloff, MD,1 Simin Hossain, MD,1 Elnaz Esmati, MD,1 Thomas Kutner, MD,1 Gregory Norcross, MD,1 James Childress, MD,1 Paul Han, MD,1 Ian Welch, MD,1 Michael Sokolow, MBA,2 Vincent See, MD, MS,3 Libin Wang, MD, PhD3. Bleeding, stroke, and mortality risk of patients with septic shock receiving anticoagulation for atrial fibrillation
- Allan J. Walkey , MD, MSc; Laura C. Myers , MD, MPH; Khanh K. Thai , MS; Patricia Kipnis , PhD; Manisha Desai , PhD; Alan S. Go , MD; Yun Lu , MPH; Heather Clancy, MPH; Ycar Devis, MPH; Romain Neugebauer, PhD; Vincent X. Liu, MD, MS. Practice Patterns and Outcomes Associated With Anticoagulation Use Following Sepsis Hospitalizations With New-Onset Atrial Fibrillation
- Michael J. Quon, MD,a Hassan Behlouli, PHD,b Louise Pilote, MD, MPH, PHD. Anticoagulant Use and Risk of Ischemic Stroke and Bleeding in Patients With Secondary Atrial Fibrillation Associated With Acute Coronary Syndromes, Acute Pulmonary Disease, or Sepsis
- Allan J. Walkey, MD, MSc, Emily K. Quinn, MS, Michael R. Winter, MPH, David D. McManus, MD, MSc, Emelia J. Benjamin, MD, ScM. Practice Patterns and Outcomes Associated With Use of Anticoagulation Among Patients With Atrial Fibrillation During Sepsis