Article Text

Original research
Initiation patterns of anticoagulants for atrial fibrillation among older UK adults with and without chronic kidney disease, 2010–2020
  1. Julia Liaw1,2,
  2. Deborah Liaw3 and
  3. Chintan Dave2,4
  1. 1Centers for Pharmacoepidemiology and Treatment Sciences, Rutgers University, New Brunswick, New Jersey, USA
  2. 2Rutgers University, New Brunswick, New Jersey, USA
  3. 3University of South Florida Morsani College of Medicine, Tampa, Florida, USA
  4. 4Department of Pharmacy Practice and Administration, Rutgers University, New Brunswick, New Jersey, USA
  1. Correspondence to Dr Chintan Dave; cdave{at}


Background There is a paucity of data on the initiation patterns of anticoagulants among older atrial fibrillation patients with and without chronic kidney disease (CKD).

Setting and methods We used the UK Clinical Practice Research Datalink (2010–2020) to conduct a retrospective cohort study to evaluate anticoagulant initiation patterns for older adults (≥65 years) with CKD (N=18 421) and without CKD (N=41 901), categorised by severity of CKD: stages 3a, 3b and 4, and initiation dose by respective direct oral anticoagulant (DOAC).

Results Over the study period, warfarin initiations sharply declined and were replaced by DOACs regardless of CKD status or stage. By 2020, patients with CKD were modestly more likely (8.8% difference) to initiate apixaban compared with those without CKD (58.8% vs 50.0%; p<0.01). Among patients with CKD, those with stages 3a and 3b CKD had higher apixaban initiations compared with stage 4 CKD (56.9% and 64.6% vs 52.9%, respectively; p<0.01). Conversely, patients with stage 4 CKD were over three times more likely to initiate warfarin (14.7%) compared with those with stage 3a (2.6%) and 3b (4.0%) CKD (p<0.01). Throughout the study period, there was a rise in the proportion of patients initiating the higher 10 mg daily dose for apixaban, with an increase of 20.6% (from 64.3% in 2013 to 84.9% in 2020; p value for trend <0.01) among patients without CKD, and 21.8% (53.1% to 74.9%; p<0.01), 24.4% (18.8% to 43.2%; p<0.01) and 18.5% (0.0% to 18.2%; p<0.01) among patients with stages 3a, 3b and 4 CKD, respectively.

Conclusions and relevance Initiation of DOACs increased regardless of CKD status and stage, although with a reduced magnitude in severe CKD. Apixaban emerged as the preferred agent, with a secular trend towards the higher initiation dose in all subgroups. These findings illuminate evolving trends and priorities in anticoagulant preferences among patients with and without CKD.

  • Atrial Fibrillation
  • Epidemiology
  • Electronic Health Records

Data availability statement

Data are available on reasonable request. Data are available on reasonable request. Deidentified patient data are available through the primary author, ORCID 0000-0002-7160-9345, although reuse would be permitted only through researchers at the primary institution.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • Clinical guidelines have favoured direct-acting oral anticoagulants (DOACs) over warfarin for atrial fibrillation (AF) treatment due to their safety and convenience. Limited data exist regarding the adoption of DOACs among AF patients, especially those with chronic kidney disease (CKD).


  • The study analysed a 10-year period and found a significant shift from warfarin to DOACs for AF treatment in both patients with CKD and non-CKD. Apixaban emerged as the preferred DOAC, and there was a trend towards higher initiation doses of apixaban in all patient subgroups. Patients with severe CKD were less likely to initiate DOACs compared with those with less severe CKD or no CKD.


  • Clinicians should be aware of the changing trends in anticoagulant preferences, with a growing adoption of DOACs, especially apixaban. This study highlights the evolving landscape of AF treatment and emphasises the importance of individualised therapy based on CKD status.


Chronic kidney disease (CKD) affects more than 4 in 10 older adults over the age of 65 and is the primary contributor to the global rise in the incidence of end-stage kidney disease (ESKD), dialysis and kidney transplantations.1 Due to the shared pathophysiological risk factors (eg, diabetes) and mechanisms (eg, atrial remodelling secondary to CKD) underpinning the development of both atrial fibrillation (AF) and CKD, the prevalence of AF rises from 1% in the general population to 10% in patients with CKD, reaching 33% in those with severe kidney impairment.2 3 Older adults have the highest prevalence of both AF and CKD of any age group.4 5

Thromboembolism is the most severe consequence of AF, and the worsening kidney function can further—and independently—exacerbate such risks.6 Fortunately, oral anticoagulants can reduce the incidence of ischaemic stroke regardless of baseline risk.7 Warfarin has been the drug of choice for decades, but direct-acting oral anticoagulants (DOACs) are now increasingly recommended owing to their convenience, safety and effectiveness.8–10 Nevertheless, determining the optimal antithrombotic strategy for older adults with comorbid AF and CKD poses complex clinical challenges due to several reasons. First, although DOACs have demonstrated a favourable benefit–risk profile compared with warfarin in early-stage CKD, their effects in patients with advanced CKD are less clear due to their routine exclusion from clinical trials. Second, all anticoagulants increase bleeding risks, particularly intracerebral and articular bleeding in older patients, and gastrointestinal bleeding in those with CKD.11–13 Third, as all DOACs undergo varying degrees of renal metabolism, careful dosage adjustment is warranted in patients with CKD to avoid potential adverse reactions.

Currently, most national14 15 and international16 17 guidelines recommend DOACs in patients dually diagnosed with AF and CKD, although advising caution among those with severe CKD, marked by an estimated glomerular filtration rate (eGFR) of <30 mL/min/1.73 m2. Despite the rapidly changing landscape in the availability of newer anticoagulants and the corresponding changes in clinical guidelines, there is a paucity of data comparing the use and uptake of anticoagulants among AF patients with and without CKD. Accordingly, we employed a large electronic healthcare database from the UK from 2010 to 2020 to examine the trends in the initiation patterns of anticoagulants among AF patients with and without CKD. Further, we analysed medication use across varying degrees of CKD and initiation doses.


Data sources and study population

We used Clinical Practice Research Datalink (CPRD) Gold data from 2010 to 2020. CPRD is an integrated, widely used electronic healthcare database containing longitudinal health records for >15 million patients sourced from >700 practices in the UK. The data are extensively validated, of research quality, and highly generalisable to the UK population.18–20 Medical information, including patient symptoms, diagnoses and lifestyle measures such as smoking status, is recorded using Read codes (V.2). All prescriptions issues by the provider are documented as product names and British National Formulary Codes.

The study cohort was composed of patients ≥65 years initiating the following anticoagulants: warfarin, dabigatran, rivaroxaban, edoxaban and apixaban. Initiation was defined as non-use of an individual study medication for 365 days prior to the initiation. Cohort membership was further restricted to initiation episodes with a diagnosis of AF, documented serum creatinine within 1 year of the index date and to practices that were up-to-standard for a minimum of 1 year prior to the index date. Patients with evidence of hyperthyroidism, valvular disease, ESKD (defined as GFR<15 mL/min/1.73 m2), recent hip fracture/replacement (ie, within 6 weeks of index date) and recent venous thromboembolism (VTE) (ie, within 6 months of index date) were excluded from analyses. Patients were allowed to contribute multiple initiation episodes as long as the eligibility criteria were met at the time of initiation.

Kidney function and baseline characteristics

We calculated eGFR using the most recent serum creatinine value prior to treatment initiation using the validated CKD Epidemiology Collaboration equations, that did not consider race.21 22 An eGFR threshold of <60 mL/min/1.73m2 was used to classify patients as having CKD, and patients were further categorised by CKD stages (eGFR in mL/min/1.73m2): 3a (45–59), 3b (30–44) and 4 (15–29).

Along with kidney function, we assessed baseline characteristics overall, by CKD status, and by individual anticoagulant initiated. These included variables corresponding to patient demographics (eg, age at index date, biological sex, region), medical characteristics (eg, myocardial infarction, heart failure and CHA2DS2-VASc scores) and lifestyle measures (eg, smoking status, alcohol abuse).

Statistical analysis

All analyses were performed by using SAS V.9.4 (SAS Institute). Initiation episodes were assigned to a calendar year from 2010 to 2020 based on the date of medication initiation. For our primary analysis evaluating trends in medication initiation patterns among patients with and without CKD, we calculated the proportion of individual anticoagulants initiated (numerator) over all anticoagulants initiations (denominator) for each calendar year.

We also reported initiation patterns by CKD stages, examined initiation dosages for individual DOACs and performed a subgroup analysis for individuals aged ≥75 years to determine the impact on our findings. We calculated the number of stroke or transient ischaemic attack (TIA) and major bleeding events along with the corresponding crude incidence rates, categorised by CKD status and anticoagulant type. Finally, we also conducted a sensitivity analysis examining a cohort of treatment naïve patients using anticoagulants for the first time, thereby excluding patients switching to another anticoagulant. Trends were assessed with the Cochrane Armitage test, and χ2 tests were used to assess significant changes in anticoagulant initiation across subgroups (eg, CKD presence or stage), with corresponding p values reported.


We identified 60 322 eligible episodes of anticoagulant initiation, out of which 18 421 had CKD (table 1). Compared with their counterparts without kidney impairment, patients with CKD were older, more likely to be female, have elevated CHA2DS2-VASc scores, lower eGFR and higher prevalence of cardiometabolic diseases. The serum creatinine values (mean (SD) and median (IQR)) were obtained on an average time frame of 89.2 (93.5) days, and a median of 54 (14–139) days preceding the date of anticoagulant initiation. In a crude analysis, patients with CKD initiating anticoagulants demonstrated a higher risk of stroke or TIA and major bleeding compared with those without CKD (table 1). Moreover, while initiators of warfarin and DOACs exhibited relatively similar incidence rates for stroke or TIA, apixaban users demonstrated the lowest risk of major bleeding among all anticoagulant initiators (table 2).

Table 1

Baseline characteristics for initiation episodes of anticoagulant medications by CKD status from 2010 to 2020*

Table 2

Baseline characteristics for initiation episodes of anticoagulant medications by class from 2010 to 2020*

The distribution of baseline characteristics was mostly similar among patients initiating the five anticoagulants (table 2), with the notable exception of prior anticoagulant use. Patients initiating warfarin had minimal prior anticoagulant use in the year preceding initiation compared with DOAC users, among whom between 21.9% and 30.6.5% had previously used warfarin. Among DOAC initiators, we noted a higher prevalence of prior rivaroxaban use compared with other DOAC therapies, indicating a tendency for patients initially treated with rivaroxaban to transition to alternative DOACs.

Trends by CKD status

Over the study period and among patients with CKD, warfarin initiations as a proportion of overall anticoagulant initiations decreased by 96% from 99.9% in 2010 to 3.9% in 2020 (p value for trend <0.01%), becoming one of least frequently initiated anticoagulant by 2020 (figure 1A). This fall in utilisation for warfarin coincided initially with the availability of rivaroxaban and apixaban thereafter. Use of apixaban rose by 56.1% from 2.7% in 2013 to 58.8% of initiations in 2020; p<0.01), becoming one of the most frequently initiated anticoagulant. By contrast, while the use of rivaroxaban increased initially from 3.6% in 2012, and peaking in 2015 to 33.3%, it declined steadily thereafter to only comprise of 12.7% of initiations by 2020. Edoxaban initiations rose from 0.1% in 2015 to 23.8% in 2020, surpassing rivaroxaban. Dabigatran use remained low (<0.8% of all initiations) throughout the study period.

Figure 1

Changes in prescribing patterns of anticoagulant medications by CKD status. Prescribing patterns among patients with CKD. Prescribing patterns among patients without CKD. The figures describe the proportion of anticoagulant initiation episodes between 2010 and 2020 by patients with CKD and non-CKD. (1) The European Society of Cardiology guidelines were published, which recommend DOACs over warfarin. (2) The American Heart Association/American College of Cardiology/Heart Rhythm Society guidelines also recommend DOACs over warfarin. CKD, chronic kidney disease; DOAC, direct oral anticoagulant.

These initiation trends were qualitatively similar among patients without CKD (figure 1B). Warfarin use declined by 96.5% (from 100% in 2010 to 3.5% in 2020; p<0.01), and increased for apixaban by 47.9% (from 2.1% in 2013 to 50.0% in 2020; p<0.01). Meanwhile, edoxaban use also rose by 31.6% (from 0.1% in 2015 to 31.7% in 2020; p<0.05), securing its position as the second most initiated anticoagulant.

When directly comparing prescribing preferences across CKD status and by 2020, initiation rates for apixaban were higher by 8.8% (58.8% with CKD vs 50.0% without CKD; p<0.01) and lower for edoxaban by 7.9% (23.8 vs 31.7%; p<0.01). Meanwhile, the use of warfarin was slightly more prevalent among patients with CKD (3.9% vs 3.5%; p<0.01).

Findings for sensitivity analysis restricting to first-time initiators of any anticoagulant were mostly similar to the primary analysis (online supplemental table 2), with some evidence of an even more pronounced increase in the use of apixaban. A subgroup analysis, restricted to those aged ≥75 years, closely resembled the primary analysis (online supplemental table 7).

Trends by CKD stage

By the end of the study period in 2020, the decrease in warfarin was most prominent among patients with stage 3a CKD (2.6% of initiations in 2020), followed by stage 3b (4.0%) and stage 4 (14.7%) (figure 2); in other words, patients with stage 4 CKD were over three times more likely to be initiated on warfarin compared with those with stage 3a or 3b CKD (14.7% vs 2.6% and 4.0%, respectively; p<0.01). While apixaban remained the most frequently initiated anticoagulant across all CKD stages by 2020, its use was lower among patients with stage 4 CKD, offset by higher warfarin prescribing. Notably, there was a decreasing trend in the use of rivaroxaban with worsening renal function, observed in patients with CKD with stage 3a (14.9% of initiations by 2020), stage 3b (9.2%) and stage 4 (5.9%). The use of dabigatran remained negligible across all CKD stages.

Figure 2

Changes in prescribing patterns by CKD stage. Prescribing patterns among patients with stage 3A CKD. Prescribing patterns among patients with stage 3B CKD. Prescribing patterns among patients with stage 4 CKD. The figures describe the proportion of anticoagulant initiation episodes between 2010 and 2020 by stages 3A, 3B and 4 CKD. CKD, chronic kidney disease.

Initiation dose for DOACs

We observed substantial differences in initiation dosing preferences across the various CKD groups. There was an inverse correlation between use of the apixaban 10 mg daily dose and higher CKD stages, indicating dose adjustment based on renal function. Notably, a secular trend of higher initiation dose of apixaban was evident for all subgroups (figure 3). For instance, over the course of the study period, the use of the 10 mg dose of apixaban increased by 20.6% among patients without CKD (from 64.3% in 2013 to 84.9% in 2020), 21.8% in patients with stage 3a CKD (from 53.1% to 74.9%), 24.4% in patients with stage 3b CKD (from 18.8% to 43.2%) and 18.5% in patients with stage 4 CKD (from 0.0% to 18.5%) (online supplemental table 4). To address variations in dose reduction criteria among DOACs, we analysed, for each specific DOAC, the proportion of patients who met the respective dose reduction criteria and were subsequently dose-reduced, relative to the total number of cases eligible for dose reduction. For apixaban, this proportion was lower at 27.5%, whereas for edoxaban and rivaroxaban, it was notably higher at 85.4% and 68.1%, respectively.

Figure 3

Prescribing patterns of apixaban 10 mg initiators by CKD stage. The figure describes the proportion of apixaban initiation episodes involving a daily dose of 10 mg (as opposed to 5 mg) from 2013 to 2020, categorised by patients without CKD, stages 3A, 3B and 4 CKD. CKD, chronic kidney disease.

However, when compared with apixaban, edoxaban and rivaroxaban displayed no evidence of a secular trend in the preferential initiation of higher doses (online supplemental tables 5,6). Nonetheless, like apixaban, these medications also demonstrated a similar stepwise reduction in the use of higher dosing regimen with worsening CKD stages.


Despite the clinical complexities imposed by CKD, to date, there have been limited data evaluating patterns of anticoagulant use considering either the presence or severity of CKD. In this study, we leveraged a large database of UK electronic medical records between 2010 and 2020 to compare anticoagulant initiation among AF patients with and without CKD. We observed a decline in warfarin use over time, coinciding with the approval of DOACs. Notably, among DOACs, apixaban emerged as the preferred choice, regardless of CKD status or stage. Patients with stage 4 CKD were three times more likely to initiate warfarin compared with those with mild to moderate renal impairment and exhibited the lowest relative use of apixaban within any subgroup. Additionally, there was a secular trend towards higher initiation doses of apixaban across all patient subgroups over the study period.

Although current clinical guidelines advocate for DOACs over warfarin for AF,16 17 real-world prescribing decisions are influenced by multiple factors including efficacy, safety, tolerability, treatment costs and available evidence among patients with CKD. Moreover, consideration must be given to the variable significance of CKD or its stages based on the specific disease context, which may impact prescribers’ decisions on anticoagulant initiation or dosage adjustment. For example, while apixaban requires renal dose adjustment in AF, with caution in severe kidney disease, its use in VTE does not require dosage adjustment or restrictions based on kidney function. This variability underscores the complexity of prescribing practices of anticoagulants. Our study revealed that DOACs, particularly apixaban, were more commonly prescribed than warfarin regardless of CKD status, reflecting evolving patient and clinician priorities. Notably, these trends contrast with the therapeutic inertia observed in other clinical contexts for newer therapies among patients with CKD.23 24 However, the use of warfarin remained higher in patients with stage 4 CKD compared with those with stages 3a and 3b CKD, despite the European Society of Cardiology and the recent American Heart Association/American College of Cardiology/Heart Rhythm Society recommendations.14 17 The disparity may stem from the absence of clinical trial data to fully establish DOAC’s efficacy and safety among patients with eGFR below 25–30 mL/min.25 While warfarin has demonstrated efficacy in AF patients with severe CKD,26 concerns remain over its safety profile, marked by an elevated risk of bleeding and a rare yet life-threatening effect for vascular calcification with prolonged use.27

While the current National Institute for Health and Care Excellence (NICE) guidelines do not recommend a particular DOAC among patients with CKD and AF,15 they acknowledge the more favourable data for apixaban in this patient population. The preferential prescribing preference for apixaban may also have been attributed to two factors. First, meta-analyses of clinical trials have shown superiority of apixaban over other DOACs, especially for risk of major bleeding in advanced CKD. Our analysis of major bleeding crude incidence rates aligns with findings from large-scale trials comparing different DOACs, which indicate an additive benefit of apixaban in reducing the risk of major bleeding relative to other anticoagulants. Second, among all DOACs, apixaban has the lowest renal excretion (27%), compared with rivaroxaban (35%), edoxaban (50%) and dabigatran (80%), making it a more attractive option for patients with CKD.

In our study, we observed that the higher 10 mg daily dose of apixaban was the primary initiation dose among patients without CKD and those with stage 3a CKD. Further, there was a notable secular trend in the increasing use of this higher dose among patients with stages 3b and 4 CKD. Despite meeting dose reduction criteria, the majority of apixaban initiators were prescribed the standard 10 mg daily dose, with only 27% undergoing dose reduction. While we cannot attribute this trend to any specific cause, certain factors such as growing familiarity of apixaban among healthcare providers may have played a role. Further, the availability of data demonstrating a low risk of bleeding and a perceived improvement in stroke outcomes with higher doses may have also influenced prescribing decisions.

Our study findings are aligned with the existing literature, showing a decline in warfarin use in favour of DOACs in the general AF population,28–30 and generally consistent with the limited available literature examining AF patients with CKD. A prior study examining anticoagulant use in newly admitted nursing home residents with AF and CKD (2011–2018) noted a clear preference for warfarin among patients with CKD compared with those without CKD31; whereas, our study found a more modest difference in warfarin use between patients with CKD and non-CKD. This variation could be attributed to differences in the underlying study populations and the study period considered. Additionally, an older Danish study conducted from 2011 to 2013, in AF patients newly initiating anticoagulants (3.2% of whom had CKD),32 the presence of CKD was associated with a 67% lower odds of DOAC (vs warfarin) initiation.

Our study represents the first systematic effort to assess anticoagulant initiation patterns among patients with and without CKD, as well as across CKD stages. This was accomplished by leveraging a large, highly generalisable, and comprehensive electronic health record database from the UK. Laboratory data to estimate serum creatinine were well recorded in the data, allowing for an unbiased appraisal of medication use. However, study findings should be viewed in light of certain limitations. First, we had information on prescribing—rather than dispensing, so it is possible that patients may not have filled their prescriptions. Second, given that our study ended in 2020, we were unable to evaluate the impact of 2021 NICE AF guidelines nor study the impact of the COVID-19 pandemic on treatment disruptions or switches. Third, CPRD data do not identify the prescription’s origin (eg, primary or specialty care) and do not include information on prescriptions originating from inpatient hospitalisations or emergency department visits. Fourth, as our study spanned from 2010 to 2020, variations in approval times for anticoagulants, such as edoxaban in 2015, may have limited our ability to assess trends, particularly in comparison to subsequent guideline updates in 2016 and 2019. Finally, as we employed a 12-month washout window to define new medication initiation, it is possible that patients may have initiated their medications prior to this 12-month window.

In conclusion, patients with stage 4 CKD were still more likely to be initiated on warfarin compared with those with stage 3a and 3b CKD, despite minimal warfarin use overall. Apixaban emerged as the preferred anticoagulant regardless of CKD status or severity, with a strong secular trend towards the use of higher doses over time. These findings ultimately shed light on the evolving trends and shifting patient and prescriber priorities in anticoagulant therapy initiation.

Data availability statement

Data are available on reasonable request. Data are available on reasonable request. Deidentified patient data are available through the primary author, ORCID 0000-0002-7160-9345, although reuse would be permitted only through researchers at the primary institution.

Ethics statements

Patient consent for publication

Ethics approval

The study was approved by Rutgers Institutional Review Board (IRB) and Clinical Practice Research Datalink (CPRD) Independent Scientific Advisory Committee.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Presented at This work was presented as an oral presentation at the 39th International Conference on Pharmacoepidemiology, Halifax, Nova Scotia, Canada, from 23 August 2023 to 27 August 2023.

  • Contributors All authors were involved in the conception and design of the study, analysis and interpretation of data, critical revision of the article for important intellectual content and final approval of the article. JL and DL drafted the article, and CD provided administrative and logistical support, statistical expertise and overall study supervision. JL assembled and analysed the data. JL and CD are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

  • Funding This study was conducted with the support of 1R01HL163163.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.