ABSTRACT
Background Stroke prevention treatment guidance for patients with atrial fibrillation (AF) uses evidence generated from randomised controlled trials (RCTs). However, applicability to patient groups excluded from trials remains unknown. Real-world patient data provides an opportunity to evaluate outcomes in a trial analogous population of direct oral anticoagulants (DOACs) users and in patients otherwise excluded from RCTs, however there remains uncertainty on the validity of the methods and suitability of the data.
Successful reference trial emulation can support the generation of evidence around treatment effects in groups excluded or underrepresented in the original trials.
We used linked UK primary care data to investigate whether we could emulate the pivotal ARISTOTLE trial (apixaban vs warfarin) and extend the analysis to investigate the impact of warfarin time in therapeutic range (TTR) on results.
Methods and findings Patients with AF in a UK primary care database Clinical Practice Research Datalink (CPRD Aurum) prescribed apixaban or warfarin from 1 Jan 2013 to 31 Jul 2019 were selected. ARISTOTLE eligibility criteria were applied to this population and matched to the RCT apixaban arm on baseline characteristics creating a trial-analogous apixaban cohort; this was propensity-score matched to warfarin users in the CPRD Aurum. ARISTOTLE outcomes were assessed using Cox proportional hazards regression stratified by prior warfarin exposure status during 2.5 years of patient follow-up and results benchmarked against the trial results before treatment effectiveness was further evaluated based on (warfarin) time in therapeutic range (TTR).
The analysis sample comprised 8734 apixaban users and propensity-score matched 8734 warfarin users in CPRD. Results [Hazard Ratio (95% Confidence Interval)] confirmed apixaban non-inferiority for stroke or systemic embolism (SE) [CPRD 0.98 (0.82,1.19) vs trial 0.79 (0.66,0.95)] and death from any cause [CPRD 1.03 (0.93,1.14) vs trial 0.89 (0.80,0.998)] but did not indicate apixaban superiority. Absolute event rates for Stroke/SE were similar for apixaban in CPRD Aurum and ARISTOTLE (1.27%/year) whereas a lower event rate was observed for warfarin (CPRD Aurum 1.29%/year, ARISTOTLE 1.60%/year)
Analysis by TTR suggested non-inferiority of apixaban in those with TTR < 0.75 [Stroke/SE 0.94 (0.75,1.19), all-cause death 0.99 (0.87,1.12)]. However, apixaban was associated with increased hazards compared with well-controlled warfarin treatment (TTR ≥ 0.75) [Stroke/SE 1.49 (1.13,1.97), all-cause death 1.75 (1.49,2.06)]. The main limitation of the study’s methodology are the risk of residual confounding, channelling bias and attrition bias in the warfarin arm.
Conclusions Analysis of non-interventional data generated results demonstrating non-inferiority of apixaban vs warfarin consistent with the pre-specified benchmarking criteria. Unlike in ARISTOTLE superiority of apixaban vs warfarin was not seen which may be linked to the lower proportion of Asian patients and higher proportion of patients with well-controlled warfarin compared to ARISTOTLE. The methodological template developed can be used to investigate treatment effects of oral anticoagulants in patient groups excluded from or under-represented in trials and also provides a framework which can be adapted to investigate treatment effects for other conditions.
Why Was This Study Done?
Stroke prevention treatment guidelines for patients with atrial fibrillation (AF) are based on results from randomised controlled trials (RCTs), we do not know if these results are relevant to patients that would not have been eligible to be included in the RCTs.
This study used routinely collected health data from the UK to emulate an RCT that compared apixaban to warfarin, ARISTOTLE, and also looked at whether the benefit of apixban compared with warfarin was impacted by the quality of warfarin therapy (measured by time in therapeutic range, TTR).
Emulating an RCT for stroke prevention in patients with AF should help to understand how transferable RCT results are to ‘real-world’ practices and whether this methodological approach can help to improve treatment options and outcomes for patient groups currently underrepresented in clinical trials.
What Did the Researchers Do and Find?
The researchers looked at patients with AF in a UK primary care data prescribed apixaban or warfarin and applied a “reference trial emulation” approach, in which the ARISTOTLE trial eligibility, selection and analysis approaches were applied to UK primary care data and results benchmarked against those of ARISTOTLE.
Patients prescribed apixaban had similar rates of outcomes to those prescribed warfarin in our cohort and our results were successfully benchmarked against ARISTOTLE. Unlike ARISTOTLE we did not see superiority of apixaban vs warfarin [Hazard ratio (95% confidence interval)] for time to stroke or systemic embolism: 0.98 (0.82,1.19) in our cohort vs 0.79 (0.66,0.95) in ARISTOTLE.
We also found the benefit of apixaban vs warfarin differed depending on the quality of warfarin therapy with apixaban superior to poorly controlled warfarin therapy (TTR < 0.75) [Stroke/SE 0.94 (0.75,1.19), Death 0.99 (0.87,1.12)] whereas apixaban was not superior to well-controlled warfarin (TTR ≥ 0.75) [Stroke/SE 1.49 (1.13,1.97), Death 1.75 (1.49,2.06)].
What Do These Findings Mean?
ur results support the NICE guidelines on selecting treatment for stroke prevention in patients with AF and also provide reassurance on continuing warfarin in patients with high TTR.
can use UK primary health care data to emulate a reference trial of treatments for the prevention of stroke in AF.
can use the data and methods to look at how well treatments work in patients that would not have been included in RCTs such as those with multimorbidity or patient groups under-represented in RCTs such as ethnic minority groups and older patients.
limitations include the possibility of residual confounding, a risk patients doing well on warfarin were over-represented in our cohort, and a lower proportion of Asian participants in our cohort compared with ARISTOTLE.
Competing Interest Statement
I have read the journal's policy and the authors of this manuscript have the following competing interests: EMP was funded by a Medical Research Council studentship for this work, and is an employee of and holds stock in Compass Pathways outside the submitted work. UG is an employee of and holds stock in GSK. JT reports no conflict of interest and is supported by an unrestricted grant from GSK. LS reports grants from Wellcome, MRC, NIHR, BHF, Diabetes UK, ESRC, EU and GSK, personal fees from GSK and AstraZeneca, and is a trustee of the British Heart Foundation, outside the submitted work. PJB is supported by Barts Charity (MGU0504) and was supported by a GSK studentship at the time of writing. TBH reports no conflict of interest. AW reports no conflict of interest and is supported by a fellowship from British Heart Foundation (FS/19/19/34175). IJD reports grants, and holds stocks in GSK, outside the submitted work. KW has nothing to disclose.
Clinical Protocols
https://bmjopen.bmj.com/content/bmjopen/11/4/e042947.full.pdf
Funding Statement
This study was supported by the Medical Research Council (grant number MR/N013638/1 for EMP). All other authors report no relevant funding. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author Declarations
I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained.
Not Applicable
The details of the IRB/oversight body that provided approval or exemption for the research described are given below:
Scientific approval was provided by the London School of Hygiene and Tropical Medicine research ethics committee (ref 17682) and the independent scientific advisory committee of the Medicines and Healthcare Products Regulatory Agency (protocol no. 19_066R). CPRD data are already approved via a national research ethics committee for purely non-interventional research of this type. CPRD data are analysed anonmymously therefore individual patient consent is not sought by contributing medical practices when data is shared with CPRD however, patients are able to opt out of their patient information being shared for research.
I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals.
Not Applicable
I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance).
Not Applicable
I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable.
Not Applicable
Footnotes
The views expressed in this paper are those of the author and not do not necessarily reflect those of the SFDA or its stakeholders. Guaranteeing the accuracy and the validity of the data is a sole responsibility of the research team.
Data Availability
Data are not publicly available but are available subject to protocol approval via CPRD’s Research Data Governance Process (https://cprd.com/data-access) for researchers who meet the criteria for access to confidential data. The data underlying the results presented in the study are available from CPRD (https://www.cprd.com).