What Anticoagulants Should We Use to Treat Atrial Fibrillation?

Editorial accompanying the article: “Direct Oral Anticoagulants Versus Phenprocoumon: Mortality, Thromboembolism and Major Hemorrhage: Systematic Review and Meta-Analysis of Routine Data Studies” by Christiane Engelbertz, Holger Reinecke, and Jeanette Köppe on the following pages

Oral anticoagulation is the most commonly employed form of stroke prevention in patients with atrial fibrillation (AF). The use of oral anticoagulants for this purpose is based on the findings of medium-sized randomized clinical trials comparing the vitamin-K antagonist warfarin with aspirin that were conducted some time ago (1). Various vitamin-K antagonists are used in different European countries, e.g., phenprocoumon in Germany, acenocoumarol in Austria, fluindione in France, and warfarin in the United Kingdom (2).

These vitamin-K antagonists inhibit synthesis of coagulation factors II, VII, IX, and X. Adjustment of the dose to an international normalized ratio (INR) of 2–3 leads to effective and relatively safe anticoagulation. Treatment with vitamin-K antagonists reduces the rate of ischemic stroke in the patients concerned by two thirds, but doubles the risk of major hemorrhage compared with no anticoagulation.

Large pivotal randomized clinical trials published from 2009 onwards reported on the efficacy of four newly approved direct oral factor-X or factor-II inhibitors (apixaban, dabigatran, edoxaban, and rivaroxaban, the so-called direct oral anticoagulants [DOACs]) in fixed doses compared with warfarin. When used in the approved dosages, these substances showed comparable efficacy in stroke prevention, a lower risk of intracranial hemorrhage, and a similar risk of bleeding events overall (at study level [3] and in meta-analyses of individual patient data [4]).

Comparison of phenprocoumon and DOAC treatment

In this issue of Deutsches Ärzteblatt International Engelbertz and co-authors report a meta-analysis of German routine claims data (5). The aim of the study was to compare the efficacy and safety of the vitamin-K antagonist typically used in Germany, phenprocoumon, with DOAC treatment. The authors analyzed routine data from German patients. To minimize the known confounders, they restricted their meta-analysis to comparisons with risk-adjustment procedures such as matching at patient level. Despite great care in their analytical approach, the results differ from those of the randomized trials: the rate of bleeding events during DOAC treatment seemed lower in Engelbertz et al.’s meta-analysis than for phenprocoumon (20% fewer major hemorrhages). Furthermore, there was a slight (8%) increase in the risk of stroke and possible signs of higher mortality. The authors’ findings also differ from those of nationwide analyses of healthcare data in Denmark (where warfarin is usually used) (6) and France (where fluindione is usually used) (7) and from a multinational analysis that included data from German patients with health insurance from the AOK fund (8). These analyses revealed that the efficacy of DOAC treatment for stroke prevention was equivalent to that of various vitamin-K antagonists, with variable bleeding rates for the different DOACs.

Differing study findings

How can these differences be explained? In short, Engelbertz et al.’s painstaking work highlights one of the central limitations of the data from Germany’s innovative and high-performance healthcare system and illustrates inherent weaknesses of non-randomized comparison:

Using unrandomized data sets, the authors compare not only different treatments, but also differing groups of patients. In routine clinical practice, the use of an anticoagulant is preceded by careful clinical consideration. This leads to detectable and undetectable confounders when performing comparisons. Without access to individual patient data, no statistical correction for this is possible. Even with individual data, the “good doctor effect,” leading to certain patients being treated with a certain anticoagulant for good reasons, cannot be completely eliminated by analytic adjustments.

The need for frequent INR tests, performed either by medical professionals or by the patients themselves, may lead to selection for vitamin-K antagonist treatment of patients who understand the necessity of stroke prevention. This increases treatment adherence, and therefore the bleeding risk, and reduces the danger of stroke, thus contributing to the observed differences. Monitoring of INR also ensures adequate dosing.

Observational studies include data not only from patients treated with licensed, effective doses of DOACs, but also from those given lower DOAC doses. Physicians often prescribe lower doses to avoid hemorrhages, and such dosing practices contribute to the observed effects (fewer bleeding events, weaker stroke prevention).

Although only indirect comparisons between phenprocoumon, acenocoumarol, fluindione, and warfarin are available, the efficacy of vitamin-K antagonists depends crucially on the quality of dose adjustment and the experience of those responsible for adjusting the dose. This can lead to selection of patients who have shown good treatment adherence in routine care. A good example of this is the regional variability of the INR documented in the RE-LY study (9).

The consideration of clinical risk factors for stroke ignores some important contributory factors to the risk of stroke and hemorrhage. Both the severity of cardiovascular comorbidities and the treatment chosen to combat them exert a decisive influence on stroke and hemorrhage risk. Furthermore, recent research shows that the AF burden, i.e., the proportion of the patient’s time taken up by AF, modulates the risk of stroke (10) (www.afburden.org). Without access to individual patient data, including details of medication and precise cardiovascular phenotypes, these crucial confounders cannot be taken into account. Only randomization can compensate for AF burden and the severity of comorbidities.

Inadequate options for data linkage

To overcome these methodological limitations, research and quality assurance need access to nationwide German healthcare data. Data linkage in Germany is hugely hampered by isolated, proprietary data systems and conservative interpretation of data protection regulations. The situation is made worse by inconsistent data collection by physicians in offices and hospitals, health insurers, and various authorities. This represents a barrier not only to efficient, evidence-based medical care, but also to the reliable evaluation of therapeutic and diagnostic approaches. Currently, healthcare patterns and the associated health outcomes in Germany cannot be properly described and analyzed. We need a consistent German health data space to provide answers to important questions such as those raised by Engelbertz et al.

Funding
PK receives financial support from the EU projects AF-B-STEP (IHI, grant agreement 101252780) and MAESTRIA (H2020, grant agreement 965286); the German Center for Cardiovascular Research (DZHK, grant numbers DZHK FKZ 81X2800182, 81Z0710116, and 81Z0710110), supported by the German Federal Ministry for Education and Research (Ki 509167694, 535121142, 564997921); the Dutch Heart Foundation (DHF); The Canadian funding program “Accelerating Clinical Trials”; and the Else Kröner-Fresenius Foundation.

Conflict of interest statement
PK has received research funding from the European Union, the British Heart Foundation, the Leducq Foundation, the Medical Research Council (UK), and the German Center for Cardiovascular Research, as well as from various companies engaged in the development of drugs to treat atrial fibrillation. He is credited as inventor in two Hamburg University patents (atrial fibrillation treatment WO 2015140571, markers for atrial fibrillation WO 2016012783) and is spokesman of the executive board of AFNET Germany.

Manuscript submitted on 4 May 2026, revised version accepted on 4 May 2026

Corresponding author
Prof. Dr. Paulus Kirchhof, MD

Cite this as:
Kirchhof P: What anticoagulants should we use to treat atrial fibrillation? Dtsch Arztebl Int 2026; 123: 375–6. DOI: 10.3238/arztebl.m2026.0080

References

1. Hart RG, Benavente O, McBride R, Pearce LA: Antithrombotic therapy to prevent stroke in patients with atrial fibrillation: a meta-analysis. Ann Intern Med 1999;131: 492–501

2. Le Heuzey JY, Ammentorp B, Darius H, et al. : Differences among western European countries in anticoagulation management of atrial fibrillation. Data from the PREFER IN AF registry. Thromb Haemost 2014; 111: 833–41

3. Ruff CT, Giugliano RP, Braunwald E, et al.: Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: a meta-analysis of randomised trials. Lancet 2014; 383: 955–62

4. Carnicelli AP, Hong H, Connolly SJ, et al.: Direct Oral Anticoagulants versus Warfarin in patients with atrial fibrillation: Patient-level network meta-analyses of randomized clinical trials with interaction testing by age and sex. Circulation 2022; 145: 242–255

5. Engelbertz C, et al.: Direct oral anticoagulants versus phenprocoumon: Mortality, thromboembolism and major hemorrhage. A systematic review and meta-analysis of routine data studies. Dtsch Arztebl Int 2026; 123: 377–84. DOI: 10.3238/arztebl.m2026.0064.

6. Staerk L, Fosbol EL, Lip GYH, et al.: Ischaemic and haemorrhagic stroke associated with non-vitamin K antagonist oral anticoagulants and warfarin use in patients with atrial fibrillation: a nationwide cohort study. Eur Heart J 2017; 38: 907–915

7. Lip GYH, Benamouzig R, Martin AC, Pesce G, Gusto G, Quignot N , et al.: Comparative safety and effectiveness of oral anticoagulants in key subgroups of patients with non-valvular atrial fibrillation and at high risk of gastrointestinal bleeding: A cohort study based on the French National Health Data System (SNDS). PLoS One 2025;20: e0317895

8. Souverein PC, van den Ham HA, Huerta C, Merino EM, Montero D, Leon-Munoz LM, et al. Comparing risk of major bleeding between users of different oral anticoagulants in patients with nonvalvular atrial fibrillation. Br J Clin Pharmacol 2021;87: 988–1000

9. Wallentin L, Yusuf S, Ezekowitz MD, Alings M, Flather M, Franzosi MG, et al. Efficacy and safety of dabigatran compared with warfarin at different levels of international normalised ratio control for stroke prevention in atrial fibrillation: an analysis of the RE-LY trial. Lancet 2010; 376: 975–83

10. Becher N, Metzner A, Toennis T, Kirchhof P, Schnabel RB: Atrial fibrillation burden: a new outcome predictor and therapeutic target. Eur Heart J 2024;45: 2824–2838