https://orcid.org/0000-0002-0031-2704
Figures
Abstract
Background
GARDENIA is a global, multicentre, prospective, non-interventional study in high-risk atrial fibrillation (AF) patients, who were untreated with anticoagulants.
Objectives
Provide contemporary evidence on reasons for non-treatment with an anticoagulant, and outcomes in high-risk AF patients.
Methods
GARDENIA enrolled 704 AF patients with a CHA2DS2-VA score ≥2 and an elevated risk of bleeding (older age (≥70 years), reduced renal function, concomitant antiplatelet or NSAID use, or other conditions associated with increased bleeding risk), and had not been recently treated with an oral anticoagulant (OAC). Patients were followed for at least 4 months. The two-year risks of mortality, stroke and major bleeding with or without direct OACs (DOACs) in these patients was estimated using the GARFIELD risk score.
Results
History of major bleeding (18.7%), physician-assessed high-risk of bleeding (17.1%), and patient refusal (17.6%) were the top reasons for not initiating OACs. The 100-person-year event rates (95% CI) for mortality, stroke/systemic embolism (SE)/transient ischemic attack (TIA), and ISTH major bleeding were 13.22 (10.33, 16.92), 2.12 (1.14, 3.94), and 2.21 (1.13, 3.91), respectively. GARFIELD risk analysis indicated a 31% and 40% increase in the risk of mortality and stroke, respectively, and a ~ 30% decrease in risk of major bleeding, compared to the estimated risk, had the patients been anticoagulated. The study was prematurely terminated due to low rate of patient accrual.
Citation: Kakkar A, Bonaca MP, Giugliano RP, Pieper K, Bloomfield D, Hug B, et al. (2026) Contemporary patients with atrial fibrillation are not anticoagulated despite risks of stroke - Insights from GARDENIA. PLoS One 21(7): e0354382. https://doi.org/10.1371/journal.pone.0354382
Editor: Ufuk Demirci, Celal Bayar University: Manisa Celal Bayar Universitesi, TÜRKIYE
Received: April 29, 2026; Accepted: July 7, 2026; Published: July 28, 2026
Copyright: © 2026 Kakkar et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: This is a disease registry study, and public availability of the data would compromise patient privacy. Publication of this data will constitute a breach of the study’s ethics approval. The aggregate data underlying this article will be shared on reasonable request, sent to the GARDENIA study Steering Committee at info@tri-london.ac.uk or by visiting Contact Us – TRI – Thrombosis Research Institute. The data is secured on TRI servers and actively serviced by dedicated IT and statistical team.
Funding: This study was funded by The Thrombosis Research Institute (TRI), London, UK and Anthos Therapeutics - a Novartis Company (Boston, USA). TRI and Anthos Therapeutics were involved in study design and the decision to publish. Study execution and data collection were done by a third-party CRO.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AF, Atrial fibrillation; CrCl, Creatinine clearance; CRNM, Clinically relevant non-major bleeding; CV, Cardiovascular; DOAC, Direct oral anticoagulant; eCRF, Electronic case report form; HRQOL, Health-related quality of life; K-M, Kaplan-Meier; MI, Myocardial infarction; INR, International Normalised Ratio; NSAID, Non-steroidal anti-inflammatory drugs; OAC, Oral anticoagulants; TIA, Transient ischaemic attack; VTE, Venous thromboembolism
Introduction
Atrial fibrillation (AF) is one of the most common heart conditions and a leading cause of mortality and long-term disability, especially in those above the age of 65 [1]. Older patients with AF often live with frailty and other co-morbid conditions that predispose them to an elevated risk of adverse health outcomes, notably stroke [2]. However, despite the high stroke risk, these patients are often under- or untreated with anticoagulants in routine clinical practice [3]. Concerns regarding increased bleeding risk, cognitive impairment, adverse outcomes and polypharmacy are some of the reasons cited for not receiving appropriate anticoagulation [3–5]. Owing to this, observational studies have previously shown that high-risk AF patients often do not receive the net clinical benefit (NCB) from anticoagulation [6], but contemporary evidence is lacking.
Direct oral anticoagulants (DOACs) that selectively target factor IIa (thrombin) or factor Xa avoid many of the limitations of vitamin K antagonists (VKA) and now constitute a majority of anticoagulation prescriptions for AF globally. Nevertheless, the introduction of DOACs has only modestly improved treatment frequency in high-risk patients [7]. A community-based longitudinal study spanning a decade (2011–2021) found that the introduction of DOACs resulted in an increase in the uptake of OACs from 50.2% in 2011 to 59.4% in 2020 [8]. Another study using electronic health record data from the US showed that among patients diagnosed with AF between 2011 and 2021, one in three high-risk patients (36.1%) received no anticoagulation [9]. Gastrointestinal (GI) bleeding remains a clinical concern with DOACs [10,11]. The FRAIL-AF study showed that switching from VKA to DOAC was associated with an increased risk of GI and urogenital bleeding [12]. The increase in thrombotic events among high-risk patients may be linked to adverse downstream consequences of interventions used to manage bleeding, such as interrupting or discontinuing anticoagulation, administering reversal agents and/or prohaemostatic drugs, and transfusing blood products [13].
Consequently, patient registries and observational studies indicate some degree of variability in how clinicians prescribe OACs in patients with AF [14]. The variability is especially high in the high-risk patient population [15]. The consequence of not receiving anticoagulation includes a higher risk of stroke and mortality. A population-based cohort study reported that the 1-year stroke risk doubled with age, from 66 years (0.7%; 95% CI, 0.5%−0.9%) to 74 years (1.7%; 95% CI, 1.3%−2.1%). Despite safety concerns, DOACs appear to have a net clinical benefit of strokes versus bleeding, especially in the older, high-risk AF patients who have comorbidities [16].
The purpose of this registry – A Global prospective observationAl study of Real-worlD managemEnt of patieNts with atrIal fibrillAtion at high risk of stroke (GARDENIA) (NCT05421533) – was to collect practice-based clinical data in a contemporary population of patients with AF with an elevated risk of stroke and yet who are untreated with anticoagulants. The objectives of this registry were to assess the reasons for withholding OACs, the clinical settings in which these patients present, and the associated clinical outcomes [17,18].
Methods
Study design and objectives
GARDENIA was a global, multicentre, prospective, observational study that evaluated the baseline characteristics and outcomes of high-risk AF patients who were not on an OAC for stroke prevention. The objectives of the study were to i) evaluate the factors associated with the decision to not treat AF patients with guideline- recommended doses of DOACs for the prevention of stroke and ii) determine the incidence of stroke/systemic embolism (SE)/ transient ischaemic attack (TIA), cardiovascular (CV)- and non-CV related mortality, other major CV events, and health-related quality of life (HRQOL). Enrolment was limited to patients not on OAC at baseline. This included patients whose risk factors indicated a need for OAC, yet who were not on an anticoagulation regimen at the time of enrolment due to various concerns. The study enrolled 704 patients across 12 countries from September 2022 to June 2024, who were followed up for a minimum of 4 months from the time of enrolment.
Study population
Full inclusion and exclusion criteria for the study population are detailed in S1 Table. The study enrolled patients diagnosed with AF or atrial flutter (documented by an ECG or cardiac monitor), who had a CHA2DS2-VA score of 2 or more (excluding female sex as a risk factor) and were not receiving OACs at the time of enrolment. In addition, the patients met one or more of the following conditions: (i) age 70 years or older; (ii) reduced renal function (creatinine clearance (CrCl) <30 mL/min); (iii) chronic use of non-steroidal anti-inflammatory drugs (NSAIDs) or antiplatelet agents; and (iv) any other condition associated with increased risk such as history of major or clinically-relevant non-major (CRNM) bleeding, risk of fall or frailty. Patients who had, or would require, a mechanical heart valve, or who had a left atrial appendage occlusion device at the time of screening, were excluded from the study (S1 Table). Frailty was assessed based on the physician’s input in the electronic case report form (eCRF). The physician or site staff responded either yes or no to the question, ‘Do you assess the patient to be frail?’.
Enrolment protocol
Potential study participants were identified during routine clinical care or through medical chart review. Medical data review to identify potential participants included AF diagnosis, age, renal function, relevant bleeding history, chronic use of NSAID medications (yes/no), use of antiplatelet medication(s) (yes/no), indication for antiplatelet use, and use of an OAC (yes/no). A pre-screening log containing the above non-identifying information was used to document screening efforts, enrolment into the registry, or reasons patients did not enrol (e.g., screen failure, refusal). Potential patients, or their legally authorised representative, were contacted about the study during routine clinical care or by telephone or email. Patients interested in enrolling in the study were scheduled for a screening/baseline visit, at which point informed consent was obtained.
Study close-out
The GARDENIA study was originally intended to enrol 1500 OAC-untreated high-risk AF patients and to have a 24-month follow-up period. The study was prematurely terminated 24 months after commencement due to the challenges of recruiting patients within the planned study milestones. Enrolment was closed, and follow-up for at least 4 months was allowed for all patients to ensure sufficient data collection to characterise the study population.
Assessment of risk using the GARFIELD risk score
The GARFIELD risk score has been previously described [19]. It was derived and validated based on data from 52,080 newly diagnosed AF patients and their 2-year clinical outcomes [20]. This tool allows one to calculate the likelihood of all-cause mortality, of a non-haemorrhagic stroke, and of a major bleed. It provides predicted risk for each day from initial diagnosis to two years and gives trajectories of risk assuming one is using VKA, a DOAC, and no oral anticoagulant. In other words, one has expected risk for each of the three outcomes, assuming each of the three possible treatment decisions, across a two-year period.
Statistical design and analysis
Categorical variables are described by frequencies and percentages, and continuous variables by medians and 25th and 75th percentiles. Renal function was calculated using CrCl from the Cockcroft-Gault estimation [21].
Rates of mortality, CV outcomes, bleeding, and subsequent OAC use were measured as numbers of events and Kaplan-Meier (K-M) rates. Because all patients were to receive a 4-month follow-up, rates are reported at that time. Events were measured from the time of enrolment in the study until the first occurrence of the event of interest, end of follow-up, or 4 months, whichever occurred first.. Several patients were in the study longer than 4 months. Therefore, the K-M event rates were also presented at 8 months. All events occurring after the last time point of interest were censored. The 100-person-year event rates were also calculated for the end-of-observation period. The CHA2DS2-VA score was calculated using eCRF data on Congestive heart failure, Hypertension, Age (2 points for age ≥ 75), history of Diabetes, history of Stroke/SE/TIA (2 points) and Vascular disease (CHA2DS2-VA) [22]. A modified HAS-BLED score, excluding labile INR was calculated using eCRF data on hypertension, age, stroke history, prior bleeding, history of renal and liver disease, and history of alcohol use [23]. Labile INR was not included, as the patients were not on warfarin treatment.
In the subset of patients who were given an OAC during the follow-up period, the time from enrolment to initiation of OAC use is displayed using a cumulative incidence curve. The EQ-5D-5L questionnaire was completed by the patients at baseline and at the 4-month visit.
Estimated event rates to 2 years (all-cause mortality, non-haemorrhagic stroke, major bleed) were calculated for all patients for each of two potential treatment decisions (DOAC, no OAC) by applying the GARFIELD risk scores [19]. The average of each of these event rates across time for each treatment arm illustrated the expected outcomes had this set of patients been followed for two years and had they received the treatment of interest. Overlaid was the K-M mortality curve from the GARDENIA population as a measure of calibration. There were not enough stroke or bleeding events in the study to provide similar curves.
Ethical approval
Independent ethics committee and hospital-based institutional review board approvals were obtained, as necessary, for the registry protocol. Additional approvals were obtained from individual study sites. GARDENIA was conducted in accordance with the principles of the Declaration of Helsinki, local regulatory requirements, and the International Conference on Harmonisation Good Pharmacoepidemiological and Clinical Practice Guidelines. Written informed consent was obtained from all study participants.
Results
Baseline characteristics
The baseline characteristics of the patients enrolled in the study are shown in Table 1. The study screened 765 patients between September 2022 and June 2024; those who did not meet the inclusion criteria or who did not consent were excluded. A total of 704 (92% of those screened) participants across 12 countries were enrolled. The top three countries by enrolment were Argentina (153; 21.7%), Poland (109; 15.5%), and the United Kingdom (99; 14.1%). The median age was 79 (Q1, Q3: 73, 86) years, and 41.2% of the participants enrolled were female; 398 (58.7%) participants were determined to be frail by the treating physician. The median (Q1, Q3) CHA2DS2-VA score was 4 (3, 5). The majority of the patients enrolled (92.5%) had prevalent AF. Of the 704 patients enrolled, end-of-study information was available for 697 patients. There were 542/697 (77%) and 311/697 (44%) patients in the study at the 4- and 8-month time points, respectively (S10 Table).
The medical history and care setting where the patients were enrolled are detailed in Table 1. Prior stroke/SE/TIA and history of bleeding were recorded in 112 (16.2%) and 283 (40.5%) patients, respectively. The majority of participants in the study, 482 (68.5%), were identified by cardiologists at clinician office sites. The EQ-5D-5L patient-reported outcome at baseline indicated that >50% of the enrolled patients had an observable limitation with mobility, activity level, or experienced some form of pain. Slight to extreme anxiety/depression and limitations to self-care were reported by 46.9% and 37% of the patients, respectively.
Reasons for withholding oral anticoagulants
Table 2 shows the reasons OACs were withheld throughout the duration of the study. The table depicts both the primary reason (n = 638) and the primary plus the additional reasons (n = 1170) for withholding OACs. Concerns about bleeding risk were the main reason for not initiating an OAC. A previous history of bleeding needing hospitalisation was stated by 119 (18.7%), while 109 patients indicated an underlying condition that was associated with bleeding. Patient refusal to take an OAC was noted in 112 (17.6%) of those enrolled. The physician’s decision to withhold anticoagulants was the most common reason for not being treated with an OAC (n = 261, 47.3%), followed by a combined decision between the physician and the patient (n = 186, 33.7%) (S2 Table).
A subset of 59 participants who were not on an OAC at the time of enrolment were anticoagulated shortly after study initiation. S1 Fig shows the timing of prescription of OACs to one year. Fifty-seven patients with treatment timing data went on to receive an OAC, and 49 of these were during the initial 4-month period [median time to start of OAC: 42 days (Q1, Q3: 11, 73)]. The breakdown of patients is shown in S3 Table. S4 Table provides the reason given for initiating OAC treatment. Three patients were recorded as having experienced stroke/TIA and 7 patients were no longer deemed to be at high bleeding risk.
Fig 1 shows the key factors associated with OAC use. Type of AF, history of hypercholesterolemia, and diabetes were associated with an increase in OAC usage. Frailty was associated with a decrease in OAC use. The complete list of baseline characteristics of patients who started OAC after enrolment in the study are provided in S5 Table.
AF treatment strategy
S6 Table presents the other AF management strategies by baseline OAC status. Prior cardioversion procedures were reported in 76 (12.1%) and 9 (15.3%) participants, and a history of ablation was recorded in 47 (7.4%) and 3 (5.1%) participants who did not start on OACs, and who started on OACs, respectively. None of the patients enrolled had a prior left atrial appendage procedure, as this was an exclusion criterion for the study.
S7 Table shows the baseline medications at the time of enrolment. Notably, more than 85% of the patients enrolled were taking some form of CV medications, and 56.5% were on beta blockers.
Clinical outcomes
The event rates at 4-months, 8-months and for 100-person years for the selected clinical outcomes for 697 patients with follow-up data are shown in Table 3. A total of 44 and 53 deaths were recorded during the 4-month period and the 8-month period, respectively. The 4-month, 8-month and 100-person year mortality rates (95% CI) were 6.48 (4.86, 8.62), 8.53 (6.54, 11.08) and 13.22 (10.33, 16.92), respectively. The CV deaths primarily occurred in the first four months.
The causes of death at 4 months and 8 months are shown in S8 Table. It is worth noting that 21/44 deaths at 4 months were due to non-CV reasons. Among the CV causes of death, six out of the 12 deaths were due to congestive heart failure. At both the 4-month and 8-month timepoints, 11 and 15 deaths were due to unknown reasons, respectively.
The rates of the composite stroke/TIA/SE at 4-months, 8-months, and per 100-person years based on the full observation period were 0.89 (0.40, 1.98), 1.35 (0.67,2.73) and 2.12 (0.03, 1.49), respectively. The one SE event at 8-months timepoint was recorded after the patient had a primary event of stroke, hence did not contribute to the composite stroke/TIA/SE per 100-person event rate. The rates of heart failure at the three time periods were 3.81 (2.59, 5.59), 5.38 (3.81, 7.57) and 7.78 (5.61, 10.79) (Table 3).
The K-M event rates for ISTH major bleeding at 4-months and per 100-person years based on full observation period are shown in Table 4. It is notable that all the major and CRNM bleeding events occurred within the first four months,. One death due to intracranial/spinal haemorrhage was recorded at the 4-month period (S8 Table). The bleeding rates for the 697 patients while not on an OAC are shown in S9 Table. Of the 59 patients who started an OAC, only one had a bleed event subsequent to OAC initiation. The patient had a major upper gastrointestinal bleed event that was captured at the 4-month timepoint.
The per-100-person-year rates of clinical outcomes based on the primary reason for not receiving anticoagulants are listed in S10 Table. Due to the low number of events, the rates presented must be viewed with caution.
The patient disposition at 4-, 8-month and end of study are shown in S11 Table. The reason for withdrawal from the study are shown in S12 Table.
Assessment of risk using the GARFIELD risk tool
The actual mortality rate closely aligned with the estimated mortality rate assuming no OAC treatment using the GARFIELD risk tool (Fig 2a). Both mortality and non-haemorrhagic stroke were estimated to be significantly reduced over the two-year period with the use of a DOAC instead of no OAC in this cohort (Figs 2a and 2b). Also expected, major bleeding rates with DOAC were estimated to be higher than with no OAC use (Fig 2c). Observed rates for ischaemic stroke and bleeding are not presented as there were only 4 strokes and 10 major bleeds at the 4-month timepoint.
The expected 2-year risk of non-haemorrhagic stroke/SE with no OAC (4.8%), decreased by over 40% to 2.8% risk with a DOAC in these same patients (Fig 2b). The estimated difference in major bleeding risk was 3.8% in the DOAC groups vs 3.0% in the no-OAC group (Fig 2c). The covariates used in the risk calculator, and their comparative baseline characteristics between the GARDENIA and GARFIELD-AF populations, are shown in S13 Table.
S14 Table shows the estimated 2-year rates of non-haemorrhagic stroke/SE and major bleeding in patients who remained untreated with an OAC through the duration of the study, and those who were initiated on OAC treatment. The rates were calculated based on the GARFIELD risk rates, assuming the patients were not given an OAC for stroke prevention. The 2-year rates for patients with an OAC treatment were slightly lower than those who were OAC untreated, but the confidence intervals were wide and overlapped between the two groups.
Discussion
The GARDENIA study enrolled AF patients who are at high-risk of stroke, and yet are not on an OAC regimen. Hypertension was the most common comorbidity (86.3%), while approximately 40% had prior history of bleeding, and 58.7% were deemed frail by the treating physician. The top reasons for not receiving an OAC were an underlying condition that was associated with bleeding risk, or prior history of bleeding. The decision to not initiate an OAC was either taken by the treating physician or was a joint decision between the physician and the patient. The 100-person-year event rates for all-cause mortality, composite of stroke/TIA/SE, and ISTH major bleeding were 13.22, 2.12 and 2.21, respectively. These rates were consistent with the GARFIELD mortality risk for patients not on a DOAC. With respect to mortality, based on the baseline characteristics of the GARDENIA patients, the average one-year predicted mortality rate when untreated with an OAC would be 13.2% (12.2%, 14.8%) using the GARFIELD mortality risk score. Taken together, this contemporary study highlights the sizable population of OAC-untreated, high-risk AF population, who are not receiving the net clinical benefit from being anticoagulated.
It is well established that high-risk AF patients are consistently inappropriately anticoagulated, due to safety concerns and perceived bleeding risk [24,25]. A 2020 analysis of Medicare data indicated that 67.1% of AF patients did not receive any OAC within the first year of AF diagnosis, with older age, dementia, frailty, anaemia, and a history of pelvic or hip fracture among predictors of non-initiation of OACs [26]. Similar results were found from meta-analyses reports from New Zealand, South Korea, Sweden and the United Kingdom [5,27]. The findings from this study align with the established evidence regarding not receiving OACs and underscore a need for a contemporary update to the status of non-anticoagulated AF patients and their clinical outcomes
Risk of bleeding, especially gastrointestinal bleeding, continues to be a major consideration when deciding on an OAC treatment strategy [28,29]. The risk of bleeding is further compounded in older and frail AF patients, and in those with renal dysfunction and other risk factors such as diabetes and hypertension. The GARDENIA study indicated that, in 81% of cases, the reason for not receiving an OAC was based on the physician’s decision or a joint decision between the patient and physician. Furthermore, while non-significant, physician-assessed frailty, history of bleeding, and history of cancer treatment were the strongest predictors of non-treatment with OAC. This highlights a larger perceived risk of bleeding by the treating physician and patient, and an elevated risk aversion behaviour. A similar outcome was also seen in the untreated cohort of the PINNACLE registry in the US. After re-review, 27.1% of the physicians would go on to reconsider prescribing OACs [30].
In GARDENIA, ~ 10% of the physicians changed their decision and prescribed OAC after the study began. Further, among patients who received OAC after enrolment, 22.6% had incident AF. This indicates that the study may have served as a platform for identifying high-risk OAC-naïve AF patients who eventually benefited from anticoagulation, a trend which has been reported in other studies [31]. The ESC and ACC/AHA/ACCP/HRS guidelines for AF management recommend periodic re-evaluation of newly diagnosed AF patients to determine the evolving risk of thromboembolism [1,32]. The result from GARDENIA provides further evidence thatperiodical re-evaluation of the anticoagualtion strategy of high-risk AF patients is required.
The 100-person year all-cause mortality rate seen in this study (13.22) was higher than the 2-year outcomes seen in GARFIELD-AF (3.79) registry and the two-year outcomes of ORBIT-AF registry (5.46 in women and 5.74 in men per 100-patient years, respectively) [33,34]. However, the proportion of OAC-untreated patients in these studies was lower than in GARDENIA. In GARDENIA ~ 91% (640/697) of the participants were untreated with an OAC for the duration of the study, and 58.7% of the participants were considered to be frail by the treating physician. The mean age and median CHA2DS2-VAsc score were 79 and 4, respectively. The population in the ELDERCARE-AF study is a closer comparator to this study. In the placebo arm of the ELDERCARE-AF study, the mean age and CHA2DS2-VASc score were 86.4 and 5.0, respectively. The all-cause mortality 100-person year rate in this arm was 10.2, which trends closer to that seen in the GARDENIA study [25].
With respect to the stroke rates, the 100-person-year event rate for stroke/SE estimated in GARDENIA at 2.12 was lower than the placebo arm of ELDERCARE-AF. We reason that the estimated stroke rate is lower due to limited enrolment and the short duration of follow-up (4-months). There were 6 stroke/SE/TIA events in the first 4-months of the study, and ~9% of the patients had started on OAC treatment after enrolment. This is likely to reduce the expected rate of stroke.
Finally, the estimated ISTH major bleeding 100-person-year rates observed in the OAC-untreated patients in this study (2.21%) were comparable to those in the placebo arm of the ELDERCARE-AF study (1.8%).
To predict outcomes in these patients, we used the GARFIELD risk score. While other mortality, stroke and bleeding risk predictors are available, we chose the GARFIELD risk score as it allowed calculation of risk for each day of the 2-year period, by treatment, rather than providing a single integer per patient [19]. The analysis of these patients indicated that they indeed corroborate with the predicted risk of mortality, stroke and bleeding in the OAC untreated patient population and the risk is reduced if they are assumed to have initiated a DOAC regimen. Further investigation is warranted, with a longer follow-up period to better understand the clinical outcomes in this population. Nevertheless, we believe clinicians would benefit from having access to such risk prediction tools that help them determine the evidence-based risk of outcomes in their patients.
Strengths and limitations
This is the first international registry conducted in the DOAC era that focussed on high-risk AF patients who were OAC naïve at the time of enrolment. Epidemiological studies and analyses of claims databases have suggested that a substantial proportion of older and high-risk AF patients do not receive appropriate OAC treatment, and hence are not attaining the net clinical benefit of anticoagulation. This study furthered our understanding of the characteristics of the OAC untreated population, studied the reasons why they remain untreated, and highlighted the resource burden they impose on the healthcare system, due to the high hospitalisation rate that was recorded. Finally, the GARFIELD risk assessment showed that the predicted stroke and mortality rate in these patients would have been lower had they received DOACs.
This study has limitations: since only OAC untreated patients were considered, the patients were not consecutively enrolled; therefore, it is not a comprehensive reflection of the patient population presenting at the clinic. Further, due to low enrolment, detailed statistical modelling regarding clinical outcomes in association with the treatment status was not performed. In addition, because the study follow-up was truncated to 4-months, long-term follow-up of the participants was not conducted. There were limited number of patients to determine the event rate at one year and to allow a closer comparison with the event rates from other studies which had a longer follow-up. The challenges in enrolment also signalled that while the untreated population existed, there were difficulties in accessing these patients for registries and clinical trials. Closer evaluation of strategies to enrol untreated AF patients must be conducted to improve recruitment and retention in future studies. Finally, the GAFIELD risk analysis was a post-hoc decision.
Conclusion
The GARDENIA registry was designed to provide a contemporary understanding of the characteristics and treatment of patients who are OAC untreated. The study demonstrated the urgent need to further evaluate this population, who can receive net clinical benefit from anticoagulation. Further, the use of an individualised risk score to predict outcomes provides a precedent for future studies and clinical practice. Finally, the study highlights a clear unmet need for safer anticoagulants with reduced bleeding risk. Factor XI inhibitors have consistently shown a safer bleeding profile [35]. However, their efficacy in stroke prevention warrants further evidence. The patient population studied in this registry may represent a key subset that could benefit from these emerging anticoagulants once stronger evidence of stroke-prevention efficacy is available.
Key learning points
What is already known:
- Older AF patients and those with elevated stroke risk factors are prone to under-treatment or being untreated with anticoagulants in routine clinical practice.
- Concerns about bleeding risk, cognitive impairment, and adverse reactions from polypharmacy are the main reasons for withholding oral anticoagulants (OACs), leaving high-risk AF patients without the net clinical benefit of anticoagulation.
- The introduction of DOACs has contributed to a marginal improvement in treatment frequency among high-risk AF patients, but a substantial proportion of patients remain untreated.
What this study adds:
- The Global prospective observationAl study of Real-worlD managemEnt of patieNts with atrIal fibrillAtion at high risk of stroke (GARDENIA) study is the first international registry conducted in the DOAC era that focused on high-risk AF patients who were OAC naïve at the time of enrolment.
- The registry highlights the contemporary reasons why many high-risk AF patients remain untreated with OACs. An underlying condition associated with bleeding risk remains the top reason for not initiating OAC treatment.
- Assessment of the registry data using the GARFIELD risk prediction tool showed that the predicted stroke and mortality rates in these patients would have been lower had they received DOACs.
Supporting information
S1 Data. GARDENIA supplementary materials - 26Jun2026.
https://doi.org/10.1371/journal.pone.0354382.s001
(DOCX)
S2 Table. The person who decided that OAC would not be used.
https://doi.org/10.1371/journal.pone.0354382.s003
(DOCX)
S3 Table. OAC uptake by patients after enrolment.
https://doi.org/10.1371/journal.pone.0354382.s004
(DOCX)
S4 Table. Reason given for starting OAC treatment.
https://doi.org/10.1371/journal.pone.0354382.s005
(DOCX)
S5 Table. Baseline characteristics associated with those who started OAC.
https://doi.org/10.1371/journal.pone.0354382.s006
(DOCX)
S6 Table. Distribution of baseline characteristics in GARDENIA: Atrial Fibrillation Strategy.
https://doi.org/10.1371/journal.pone.0354382.s007
(DOCX)
S9 Table. ISTH Bleeding rates while not on an OAC.
Patients who start an OAC are censored at the time of the OAC.
https://doi.org/10.1371/journal.pone.0354382.s010
(DOCX)
S10 Table. Clinical events based on main reason for not being treated with anticoagulants (per 100-person-year rates for end of observation time).
https://doi.org/10.1371/journal.pone.0354382.s011
(DOCX)
S11 Table. Disposition status at different time points during follow-up (697 with end of study information).
https://doi.org/10.1371/journal.pone.0354382.s012
(DOCX)
S13 Table. Comparing the GARDENIA patients to the GARFIELD-AF patients with CHA2DS2-VA of 2 or greater who did not receive any OAC as their initial treatment.
GARFIELD-AF is patients with newly diagnosed AF.
https://doi.org/10.1371/journal.pone.0354382.s014
(DOCX)
S14 Table. GARFIELD estimated rates of outcomes.
https://doi.org/10.1371/journal.pone.0354382.s015
(DOCX)
S1 Fig. Cumulative incidence of OAC uptake in those patients who went on to take an OAC during the follow-up period.
Note: Of the 697 patients with follow-up information, 57 were subsequently started on an OAC. (59 patients have an observation of OAC use but two do not have a start date for the OAC so we cannot confirm if it was before or after the start of enrolment). The median time to the start of an OAC was 42 days (11, 73).
https://doi.org/10.1371/journal.pone.0354382.s016
(JPG)
Acknowledgments
We thank the physicians, nurses and patients involved in the GARDENIA registry, and the staff at Anthos Therapeutics and CYTE Global for study execution.
^GARDENIA National Coordinating Investigators
Cecilia Bahit MD1*, Petr Janský MD, PhD2; Uwe Zeymer MD3; Matyas Keltai MD, PhD4; Giuseppe Ambrosio MD, PhD5; Janina Stepinska MD, PhD6; Richard Hobbs MBBS, MA 7; Jose Lopez-Sendon MD8; Charles Pollack MD9; Antonio Barretto MD, PhD10; Carlos Jerjes Sanchez Diaz MD11; Shaun Goodman MD, MS12
*Lead author
1INECO Neurociencias Oroño, Fundación INECO, Rosario, Santa Fe, Argentina (ceciliabahit@gmail.com).
2Department of Cardiovascular Surgery, Motol University Hospital, Prague, Czech Republic,
3Klinikum der Stadt Ludwigshafen am Rhein, Department of Cardiology, Ludwigshafen, Germany,
4Semmelweis University, Hungarian Cardiovascular Institute, Budapest, Hungary,
5Division of Cardiology, University of Perugia School of Medicine Cardiology, Italy,
6Institute of Cardiology, Warsaw, Poland
7Oxford Primary Care, Radcliffe Observatory Quarter, University of Oxford, Oxford, UK,
8Centro de Investigación Biomédica en Red-Enfermedades Cardiovasculares, Madrid, Spain, 9Department of Emergency Medicine, University of Mississippi Medical Centre, Jackson, MS, USA, 10 Heart Institute of the University of Sao Paulo,
11Tecnológico de Monterrey, Escuela de Medicina y Ciencias de la Salud, Instituto de Cardiología y Medicina Vascular, TecSalud, Monterrey, Mexico,
12St. Michael’s Hospital, Toronto, Canada;
References
- 1. Hindricks G, Potpara T, Dagres N, Arbelo E, Bax JJ, Blomström-Lundqvist C, et al. Corrigendum to: 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): The Task Force for the diagnosis and management of atrial fibrillation of the European Society of Cardiology (ESC) Developed with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. European Heart Journal. 2021;42(40):4194–4194.
- 2. Proietti M, Romiti GF, Vitolo M, Harrison SL, Lane DA, Fauchier L, et al. Epidemiology and impact of frailty in patients with atrial fibrillation in Europe. Age Ageing. 2022;51(8):afac192. pmid:35997262
- 3. Romiti GF, Proietti M, Bonini N, Ding WY, Boriani G, Huisman MV, et al. Clinical Complexity Domains, Anticoagulation, and Outcomes in Patients with Atrial Fibrillation: A Report from the GLORIA-AF Registry Phase II and III. Thromb Haemost. 2022;122(12):2030–41. pmid:36037828
- 4. Diemberger I, Fumagalli S, Mazzone AM, Bakhai A, Reimitz PE, Pecen L, et al. Perceived vs. objective frailty in patients with atrial fibrillation and impact on anticoagulant dosing: an ETNA-AF-Europe sub-analysis. Europace. 2022;24(9):1404–11. pmid:35512229
- 5. Johansson C, Hägg L, Johansson L, Jansson J-H. Characterization of patients with atrial fibrillation not treated with oral anticoagulants. Scand J Prim Health Care. 2014;32(4):226–31. pmid:25464863
- 6. Søgaard M, Jensen M, Højen AA, Larsen TB, Lip GYH, Ording AG, et al. Net Clinical Benefit of Oral Anticoagulation Among Frail Patients With Atrial Fibrillation: Nationwide Cohort Study. Stroke. 2024;55(2):413–22. pmid:38252753
- 7. Paquette M, Fadahunsi O, Wang M, Healey JS, Parkash R, Quayyum S, et al. Systematic review of treatment gaps in oral anticoagulant use in atrial fibrillation. Thrombosis Update. 2021;4:100053.
- 8. Williams BA, Blankenship JC, Voyce S, Chang AR. Trends over time in oral anticoagulation and stroke rates in atrial fibrillation: A community-based study. J Stroke Cerebrovasc Dis. 2024;33(12):108081. pmid:39396659
- 9. Navar AM, Kolkailah AA, Overton R, Shah NP, Rousseau JF, Flaker GC, et al. Trends in oral anticoagulant use among 436 864 patients with atrial fibrillation in community practice, 2011 to 2020. J Am Heart Assoc. 2022;11(22):e026723. pmid:36346063
- 10. Al-Hussainy N, Kragholm KH, Lundbye-Christensen S, Torp-Pedersen C, Pareek M, Therkelsen SK, et al. Gastrointestinal bleeding with direct oral anticoagulants in patients with atrial fibrillation and anaemia. Thromb Res. 2023;232:62–9. pmid:37939578
- 11. Giugliano RP, Braunwald E, Murphy SA, Wiviott SD, Halperin JL, Waldo AL, et al. N Engl J Med. 2013;369(22):2093–104.
- 12. Joosten LPT, van Doorn S, van de Ven PM, Köhlen BTG, Nierman MC, Koek HL, et al. Safety of Switching From a Vitamin K Antagonist to a Non-Vitamin K Antagonist Oral Anticoagulant in Frail Older Patients With Atrial Fibrillation: Results of the FRAIL-AF Randomized Controlled Trial. Circulation. 2024;149(4):279–89. pmid:37634130
- 13. Held C, Hylek EM, Alexander JH, Hanna M, Lopes RD, Wojdyla DM, et al. Clinical outcomes and management associated with major bleeding in patients with atrial fibrillation treated with apixaban or warfarin: insights from the ARISTOTLE trial. Eur Heart J. 2015;36(20):1264–72. pmid:25499871
- 14. Hsu JC, Maddox TM, Kennedy KF, Katz DF, Marzec LN, Lubitz SA, et al. Oral Anticoagulant Therapy Prescription in Patients With Atrial Fibrillation Across the Spectrum of Stroke Risk: Insights From the NCDR PINNACLE Registry. JAMA Cardiol. 2016;1(1):55–62. pmid:27437655
- 15. Sugrue A, Sanborn D, Amin M, Farwati M, Sridhar H, Ahmed A, et al. Inappropriate Dosing of Direct Oral Anticoagulants in Patients with Atrial Fibrillation. Am J Cardiol. 2021;144:52–9. pmid:33385355
- 16. Abdel-Qadir H, Singh SM, Pang A, Austin PC, Jackevicius CA, Tu K, et al. Evaluation of the Risk of Stroke Without Anticoagulation Therapy in Men and Women With Atrial Fibrillation Aged 66 to 74 Years Without Other CHA2DS2-VASc Factors. JAMA Cardiol. 2021;6(8):918–25. pmid:34009232
- 17. Hankey GJ, Patel MR, Stevens SR, Becker RC, Breithardt G, Carolei A, et al. Rivaroxaban compared with warfarin in patients with atrial fibrillation and previous stroke or transient ischaemic attack: a subgroup analysis of ROCKET AF. Lancet Neurol. 2012;11(4):315–22. pmid:22402056
- 18. Kakkar A, Bonaca M, Giugliano RP, Bloomfield D, Pieper K, Yi BA, et al. CO8 Design and Rationale of a Global Prospective Observational Study of Real-World Management of Patients with Atrial Fibrillation at High Risk of Stroke- Gardenia. Value in Health. 2023;26(6):S15.
- 19. Fox KAA, Virdone S, Pieper KS, Bassand J-P, Camm AJ, Fitzmaurice DA, et al. GARFIELD-AF risk score for mortality, stroke, and bleeding within 2 years in patients with atrial fibrillation. Eur Heart J Qual Care Clin Outcomes. 2022;8(2):214–27. pmid:33892489
- 20. Kakkar AK, Mueller I, Bassand J-P, Fitzmaurice DA, Goldhaber SZ, Goto S, et al. International longitudinal registry of patients with atrial fibrillation at risk of stroke: Global Anticoagulant Registry in the FIELD (GARFIELD). Am Heart J. 2012;163(1):13-19.e1. pmid:22172431
- 21. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron. 1976;16(1):31–41. pmid:1244564
- 22. Friberg L, Rosenqvist M, Lip GYH. Evaluation of risk stratification schemes for ischaemic stroke and bleeding in 182 678 patients with atrial fibrillation: the Swedish Atrial Fibrillation cohort study. Eur Heart J. 2012;33(12):1500–10. pmid:22246443
- 23. Pisters R, Lane DA, Nieuwlaat R, de Vos CB, Crijns HJGM, Lip GYH. A novel user-friendly score (HAS-BLED) to assess 1-year risk of major bleeding in patients with atrial fibrillation: the Euro Heart Survey. Chest. 2010;138(5):1093–100. pmid:20299623
- 24. McGrath ER, Go AS, Chang Y, Borowsky LH, Fang MC, Reynolds K, et al. Use of Oral Anticoagulant Therapy in Older Adults with Atrial Fibrillation After Acute Ischemic Stroke. J Am Geriatr Soc. 2017;65(2):241–8. pmid:28039855
- 25. Okumura K, Akao M, Yoshida T, Kawata M, Okazaki O, Akashi S, et al. Low-Dose Edoxaban in Very Elderly Patients with Atrial Fibrillation. N Engl J Med. 2020;383(18):1735–45. pmid:32865374
- 26. Ko D, Lin KJ, Bessette LG, Lee SB, Walkey AJ, Cheng S, et al. Trends in use of oral anticoagulants in older adults with newly diagnosed atrial fibrillation, 2010-2020. JAMA Netw Open. 2022;5(11):e2242964. pmid:36399339
- 27. Sussman M, Barnes GD, Guo JD, Tao CY, Gillespie JA, Ferri M, et al. The burden of undertreatment and non-treatment among patients with non-valvular atrial fibrillation and elevated stroke risk: a systematic review. Curr Med Res Opin. 2022;38(1):7–18. pmid:34632887
- 28. Ruff CT, Giugliano RP, Braunwald E, Hoffman EB, Deenadayalu N, Ezekowitz MD, 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(9921):955–62. pmid:24315724
- 29. Deitelzweig S, Keshishian A, Kang A, Dhamane AD, Luo X, Balachander N, et al. Burden of major gastrointestinal bleeding among oral anticoagulant-treated non-valvular atrial fibrillation patients. Ther Adv Gastroenterol. 2021;14:1756284821997352. pmid:33815568
- 30. Cannon CP, Kim JM, Lee JJ, Sutherland J, Bachireddy R, Valentine CM, et al. Patients and Their Physician’s Perspectives About Oral Anticoagulation in Patients With Atrial Fibrillation Not Receiving an Anticoagulant. JAMA Netw Open. 2023;6(4):e239638. pmid:37093601
- 31. Svennberg E, Engdahl J, Al-Khalili F, Friberg L, Frykman V, Rosenqvist M. Mass Screening for Untreated Atrial Fibrillation: The STROKESTOP Study. Circulation. 2015;131(25):2176–84. pmid:25910800
- 32. Joglar JA, Chung MK, Armbruster AL, Benjamin EJ, Chyou JY, Cronin EM, et al. 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2024;149(1):e1–156. pmid:38033089
- 33. Atar D, Berge E, Le Heuzey J-Y, Virdone S, Camm AJ, Steffel J, et al. The association between patterns of atrial fibrillation, anticoagulation, and cardiovascular events. Europace. 2020;22(2):195–204. pmid:31747004
- 34. Piccini JP, Simon DN, Steinberg BA, Thomas L, Allen LA, Fonarow GC, et al. Differences in Clinical and Functional Outcomes of Atrial Fibrillation in Women and Men: Two-Year Results From the ORBIT-AF Registry. JAMA Cardiol. 2016;1(3):282–91. pmid:27438106
- 35. Patel SM, Giugliano RP, Ruff CT. Factor XI inhibition for atrial fibrillation: where do we stand?. Circulation. 2025;151(20):1433–5. pmid:40388506