Emulating a target trial to assess effect modification: an application to obesity in the comparative effectiveness and safety of apixaban versus warfarin in non-valvular atrial fibrillation using electronic health records

Turki M Bin Hammad, Emma Powell, Paris J Baptiste, Ian Douglas, Kevin Wing
doi: https://doi.org/10.1101/2023.10.12.23296756

Abstract

Anticoagulation therapy is recommended for patients with non-valvular atrial fibrillation (NVAF) and an increased risk of stroke. Although apixaban showed superiority over warfarin in the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) study, little is known about their effects in overweight and obese patients, with concerns that obesity might undermine apixaban’s effects due to the fixed dosing across body mass index (BMI) groups. We emulated a target trial similar to the ARISTOTLE study using Clinical Practice Research Datalink (CPRD) linked data to estimate the 36-month risk ratios (RR) and risk differences of the effects of apixaban compared to warfarin in NVAF patients in a composite of stroke/systemic embolism (SE), major bleeding and all-cause mortality. In 55,826 patients, apixaban did not differ across groups of BMI in stroke/SE with RR (95% CI) of 1.15 (0.81, 1.62) in normal weight, 1.06 (0.70, 1.61) in overweight and 1.23 (0.69, 2.17) in obese patients. In major bleeding, apixaban was not superior to warfarin in the normal weight group (RR (95% CI) 1.10 (0.76, 1.60) but superior in overweight (RR (95% CI) 0.73 (0.57, 0.93) and obese (RR (95 % CI) 0.67 (0.52, 0.87) groups. In NVAF, the effectiveness and safety of apixaban compared to warfarin were consistent across BMI groups.

Introduction

Atrial fibrillation (AF) is a disease of irregular heart rhythm and the most common form of arrhythmia. It affects more than 46 million people around the world increasing the risk of stroke and mortality1–4. Patients at increased risk of stroke are recommended to start anticoagulation therapy. Warfarin, a vitamin K antagonist (VKA), was the anticoagulant of choice for a long period before a direct oral anticoagulant (DOAC), apixaban, proved to be superior in reducing the risk of a composite of stroke and systemic embolism and the risk of bleeding in the Apixaban for Reduction in Stroke and Other Thromboembolic Events in Atrial Fibrillation (ARISTOTLE) trial in 20115–7.

Randomised controlled trials (RCTs) are usually conducted at a high cost limiting their representation of important subpopulations8. Analysis of electronic health records (EHRs) addresses this issue but comes at the expense of estimating treatments’ effects in the absence of randomisation rendering any observed associations prone to confounding9. Statistical techniques (e.g., standardisation or outcome-regression) can only account for known and collected variables while those unmeasured can still cause biased causal treatment estimates10.

One way to improve the validity of inferences from observational data is to benchmark or validate an observational study against an existing RCT. That is, to design an observational study that mirrors the reference RCT in terms of design and statistical analysis and then compare the results from the two studies11,12. If results are comparable, one can have more trust in this data source to explore treatment effects in under-represented populations. Powell et al.(2021,2022) replicated the ARISTOTLE trial using a trial-analogous cohort from the Clinical Practice Research Datalink (CPRD) linked to Hospital Episodes Statistics (HES) and Office for National Statistics (ONS) databases in the United Kingdom13,14.

Obesity can play an important role in the treatment response of the two anticoagulants. Warfarin is a highly lipophilic drug with only a small fraction of the total dose unbounded (the active part of the drug) from plasma protein. This is affected by changes triggered by obesity15–17. For apixaban, a lower dose is used in underweight patients but there is no dose modification for obese patients possibly leading to sub-optimal dosing. Recent literature identified a gap in the evidence for optimal dosing of patients with increasing body weight which results in altered haemostasis and increased thrombosis18,19.

We emulated a target trial similar to ARISTOTLE to investigate whether obesity is an effect modifier of the effectiveness and safety of apixaban compared to warfarin in NVAF patients.

Methods

Study design

We carried out a cohort analysis emulating a target trial highly similar to the ARISTOTLE study. We had two primary clinical questions: What is the effect of apixaban (5mg/2.5mg twice daily) compared to warfarin (targeting an INR of (2.0-3.0)) in NVAF patients across each BMI group on the 3-year risk of 1) a composite of ischaemic/haemorrhagic stroke or systemic embolism regardless of treatment discontinuation or switching. 2) bleeding had no patient discontinued or switched treatment.

Our estimands targeted 1) treatment effects of apixaban to warfarin in NVAF patients regardless of treatment discontinuation or switching for each effectiveness outcome (Table S1, Estimands 1,3-7). 2) An estimand that differed in handling intercurrent events of treatment switching and discontinuation by using a hypothetical strategy through censoring for the safety outcome (Table S1, Estimands 2). We considered death as a mediator for other outcomes as our interest was the total effect of treatments20 .Figure 1 shows timelines of data collection21.

Figure 1.
Figure 1.

Study scheme and timelines showing windows and time-periods for collecting study variables

The reference RCT: ARISTOTLE study5

Study design

ARISTOTLE was a randomised, double-blind, double-dummy, active-controlled study that compared apixaban (5mg/2.5mg) to warfarin (targeting INR of 2.0 to 3.0) in AF patients from 19/12/2006 to 30/01/2011.

Patients were 18 ≥ years, had permanent or persistent AF or atrial flutter, had at least one additional risk factor for stroke. Inclusion and exclusion criteria are detailed in Table S2.

The primary objective was to establish the non-inferiority of apixaban to warfarin in terms of the primary outcome of a composite of ischaemic/haemorrhagic stroke or systemic embolism.

Study results

In 18,201 patients, the hazard ratio (HR) (apixaban/warfarin) for the primary composite outcome was 0.79 (95% CI: 0.66, 0.95; p: <0.001 for non-inferiority; p: 0.01 for superiority). The HR for major bleeding was 0.69 (95% CI: 0.60, 0.80; p: <0.001) and for death from any cause 0.89 (95% CI: 0.80, 0.998; p: 0.047).

The target trial protocol

Our target trial design consisted of two steps: 1) defining a protocol and an analysis plan of the target trial that would have been conducted if feasible; 2) emulating the target trial using observational analysis of CPRD-Aurum linked data apart from aspects that are limited by the nature of the data22,23. Table S2 shows the protocol of the target trial. This would be a pragmatic, open-label, randomized controlled trial closely mirroring the design and analysis of the ARISTOTLE study while considering the settings of the current study.

The main deviations in our target trial: 1) An open-label trial design since it is not feasible to emulate a double-blind trial in routine care23 ; 2) Study duration was from 01/01/2013 to 31/07/2019; 3) It included incident users of treatments; 4) Randomization was stratified by BMI to have a more efficient stratified analysis; 5) It did not assess some of the secondary outcomes. A detailed comparison can be found in Table S2.

Observational cohort analysis

We emulated the target trial using Clinical Practice Research Datalink Aurum (CPRD-Aurum) linked data with deviations such as the lack of informed consent and timeline of assessing baseline variables.Table S2 highlights the differences between the target trial and the observational analysis.

Data sources

The CPRD-Aurum contains anonymized primary care EHRs universally accessed across the United Kingdom (UK). It currently covers around ∼20% of the total population, with about 38.4 million patients available for linkage24.

The CPRD data is representative of the UK population in terms of age, sex, ethnicity, and BMI25,26 with only 5.4% of individuals registered with a general practice opting-out of data sharing27.

The database contains information on demographic characteristics, diagnoses and symptoms, drug prescriptions, vaccination history, laboratory tests, referrals to hospital and specialist care, and lifestyle factors (e.g., alcohol consumption).28.

The database was linked to the index of multiple deprivation, a measure of socioeconomic status in the UK29.

The database is also linked to Hospital Episodes Statistics-Admitted Patient Care (HES-APC) and mortality data from the Office for National Statistics (ONS)30,31.

Setting and participants

We included adult (≥ 18 years old) patients registered for ≥ 6 months in CPRD-Aurum participating general practices (to ensure they are incident users) with available linkage to HES-APC and ONS. The study was restricted to patients residing in England because of the availability of linkage to HES-APC32. The study period started on 01/01/2013 (after apixaban gained regulatory approvals) to 31/07/2019. Participants were followed from the date of their first prescription of study treatments to the earliest of first occurrence of study outcomes or censoring events.

Detailed eligibility criteria can be found in Table S2. Participants with a first prescription of apixaban or warfarin during the study period had to have documented NVAF diagnosis and scoring ≥ 2 in CHA2DS2-VASc. Patients were excluded if they had a history of clinically significant mitral stenosis, increased bleeding risk that contraindicates oral anticoagulation, other conditions requiring chronic anticoagulation, active infective endocarditis, use of aspirin >165 mg/day or simultaneous treatment of both aspirin (any dose) and clopidogrel or ticlopidine, stroke within seven days before the index date.

Variables

Details of data sources, codes and levels of study variables can be found in Table S3.

Treatment Strategies

We compared two treatment strategies 1) apixaban (2.5 mg/5 mg); 2) warfarin targeting an INR level of 2.0 to 3.0. We used the CPRD-Aurum to classify participants to a treatment strategy based on their earliest treatment prescription.

Outcomes

The primary outcomes in the study were risk of 1) composite outcome of stroke or SE and 2) major bleeding (defined as haemorrhagic stroke, gastrointestinal, retinal, respiratory and other unspecified sites haemorrhages).

Secondary outcomes were risk of the individual components of stroke/SE and death from any cause.

Study outcomes were identified using HES-APC and ONS based on the International Statistical Classification of Diseases and Related Health Problems version 10 (ICD-10) clinical codes.

Linkage to HES-APC reduces the chance of under-ascertainment of study outcomes because they are usually managed at secondary care points33,34.

Covariates

We used directed acyclic graphs (DAGs)35 to depict our assumptions about the underlying confounding (Figures S1, S2, S3). We assumed that patients were randomly assigned to a strategy conditioning on their baseline covariates and the DAGs were used for confounder selection in our statistical models.

For all outcomes, we accounted for age, gender, ethnicity, individual-level and practice-level indices of multiple deprivation (IMD2015), alcohol, cancer, smoking, renal function, liver disease, persistent uncontrolled hypertension (based on two readings in the prior year), low platelet count (≤ 100,000/mm3), hemoglobin level <9 g/dL, concomitant use of low-dose aspirin or clopidogrel, and index year (to account for changes in clinical care over the years).36–56

For effectiveness outcomes, we additionally included the remaining CHA2DS2-VASc score components and concomitant use of medications (statins, beta-blockers, amiodarone, angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs)).47,57,58,58–62

For mortality we added the components of the Charlson comorbidity index. These were chronic obstructive pulmonary disease, connective tissue disease, peptic ulcer disease, hemiplegia, solid and blood tumors.63–65

For bleeding, we included components of the ORBIT Bleeding Risk Score for Atrial Fibrillation and HAS-BLED Score for Major Bleeding Risk. These factors included any history of GI bleeding, intracranial bleeding, or haemorrhagic stroke, hypertension66,67. These were dealt with at the design stage through restriction.

We considered BMI as an effect modifier and was included in the causal graphs68. Although obesity can impact treatment response, it is unlikely to affect treatment prescription through paths other than one of the CHA2DS2-VASc components, which was blocked (Figures S1, S2, S3). BMI was recorded using the latest observation before study entry based on height and weight or direct BMI records and categorised as either healthy weight (≤ 24.9 kg/m2), overweight (25-29.9 kg/m2) or obese: (≥ 30kg/m2).

Statistical methods

We used a complete case analysis as our analysis population. We used stabilized inverse probability treatment weighting (IPTW) to achieve exchangeability between study arms based on baseline confounders. The weights were estimated using logistic regression models with treatment assignment as the outcome and confounders as predictors. We then calculated the predicted probability of treatment for each person and assigned it as the denominator of the IPT weights. The numerator for IPT weights targeting effect modification was estimated from the predicted probability using a logistic regression model with treatment assignment as the outcome and BMI groups as the only predictor. We also accounted for bias arising from censoring due to loss to follow-up using inverse probability censoring weighting. We obtained the weights using a pooled over time logistic regression model to estimate the time-varying probability of censoring with treatment and the same covariates in the IPT model. To get the final weights, we multiplied the weights from the inverse probability treatment model to those from the censoring model (Section F). We then used the weights in an IPW estimator based on weighted estimates of the observed cause-specific hazards of the outcome and death as a competing event (coincides with a weighted Aalen-Johansen estimator for competing risks) stroke/SE and bleeding outcomes and a complement of a weighted product-limit (Kaplan–Meier) estimator for death to estimate the effects of treatments on the 3-year cumulative incidence of outcome (36 months) with time since first prescription as the time scale and the risk of outcome estimated each month in the study20. We reported BMI-stratum-specific risk differences per 100 people, risk ratios and ratios of risk ratios. We derived 95% confidence intervals (CI) using the 2.5 and 97.5 percentiles of a non-parametric bootstrap distribution using 500 samples. We assessed both additive and multiplicative effect modification using the warfarin’s normal weight group as the reference arm but did not test for interaction69,70. Participants were censored in case of leaving the practice, opting-out from data sharing, or at the end of the study period.

We conducted one additional analysis using a hypothetical estimand for the primary effectiveness outcome given the potential importance for clinicians (Table S1, Estimands 8). We also conducted four sensitivity analyses: 1) Using 99% truncated IP weights; 2) Restricting the analysis to patients with available BMI measurements in the 3 years prior to study entry 3) Estimating the direct effect using an IPW estimator that corresponds to the complement of a weighted product-limit (Kaplan–Meier) estimator by censoring for death 4) Using IP weighted Cox proportional hazard model.

Results

Participants

55,826 (28.8%) out of 194,210 participants with a diagnosis of NVAF and first prescription of apixaban or warfarin between January 1, 2013, to July 31, 2019, were eligible for the study. There were 30,420 apixaban users contributing 43,472 person-years of follow-up and 25,406 warfarin users contributing 57,752 person-years of follow-up. Figure 2 shows the flow of participants selection in the study.

Figure 2.
Figure 2.

Flowchart of eligibility of patients diagnosed with non-valvular atrial fibrillation and had a first prescription of an anticoagulant during study period

Table 1 shows the baseline characteristics of the study population by exposure. In general, the characteristics were well-balanced between the two treatment arms. There were, however, noted differences in terms of some prognostic factors. Apixaban was given more frequently to patients with a history of stroke, transient ischemic attack, or systemic embolism (27% compared to 21%) but less frequently in other vascular diseases compared to warfarin. Standardised mean differences in baseline characteristics were highly balanced after IPT weighting Table S4. A summary of IP weights can be found in Section F.

View this table:
Table 1. Baseline characteristics of eligible study participants using CPRD linked data by treatment group

Comparative effectiveness

Primary outcomes

Stroke/systemic embolism

For the primary composite endpoint, the risk ratio (95% CI) of apixaban to warfarin was 1.15 (0.81, 1.62) in the normal weight group, 1.06 (0.70, 1.61) in the overweight group and 1.23 (0.69, 2.17) in the obese group. The estimated risk difference per 100 people (95% CI) was 0.6 (−0.7,1.9) in the normal weight group, 0.2 (−1.6, 2.0) in the overweight group and 0.7 (−1.7, 3.2) in the obese group. Figures (3, 4) show the 3-year BMI stratum-specific risks, risk differences per 100 people, and risk ratios and the associated cumulative incidences stratified by BMI. The stroke (any type), ischemic stroke, and SE outcomes were consistent in showing no difference between apixaban and warfarin. In haemorrhagic stroke, apixaban was better than warfarin in overweight patients but failed to show superiority in normal weight and obese patients. All of the individual components were consistent with the composite endpoint in terms of effect modification Section G.

Figure 3.
Figure 3.

Inverse probability weighted estimates of the 3-year BMI stratum-specific risks, risk differences per 100 people, and risk ratios of the comparative effectiveness for the total effect of apixaban versus warfarin in stroke/systemic embolism using a complete-case analysis (Estimand 1)

Figure 4.
Figure 4.

Inverse probability weighted cumulative incidence of the 3-year risk of the total effect of apixaban versus warfarin in stroke/systemic embolism stratified by BMI using a complete-case analysis (Estimand 1)

For death from any cause, the risk ratio (95% CI) of apixaban to warfarin was 1.14 (0.82, 1.60) in the normal weight group, 0.87 (0.61, 1.24) in the overweight group and 1.05 (0.74, 1.50) in the obese group. The estimated risk difference per 100 people (95% CI) was 3.2 (−4.6, 11.0) in the normal weight group, -2.1 (−8.2, 3.9) in the overweight group and 0.8 (−5.1, 6.6) in the obese group Figures (5, 6).

Figure 5.
Figure 5.

Inverse probability weighted estimates of 3-year BMI stratum-specific risks, risk differences per 100 people, and risk ratios of the comparative effectiveness for the total effect of apixaban versus warfarin in all-cause mortality using a complete-case analysis (Estimand 7)

Figure 6.
Figure 6.

Inverse probability weighted cumulative incidence of the 3-year risk of the total effect of apixaban versus warfarin in all-cause mortality stratified by BMI using a complete-case analysis (Estimand 7)

Major bleeding

The risk ratio (95% CI) of apixaban to warfarin was 1.10 (0.76, 1.60) in the normal weight group, 0.73 (0.57, 0.93) in the overweight group and 0.67 (0.52, 0.87) in the obese group. The estimated risk difference per 100 people (95% CI) was 0.9 (−2.4, 4.2) in the normal weight group, -2.3 (−4.1, -0.5) in the overweight group and -3.2 (−5.7, -0.8) in the obese group Figures (7, 8).

Figure 7.
Figure 7.

Inverse probability weighted estimates of 3-year BMI stratum-specific risks, risk differences per 100 people, and risk ratios of the comparative safety for the total effect of apixaban versus warfarin in major bleeding using a complete-case analysis (Estimand 2)

Figure 8.
Figure 8.

Inverse probability weighted cumulative incidence of the 3-year risk of the total effect of apixaban versus warfarin in major bleeding stratified by BMI using a complete-case analysis (Estimand 2)

Supplementary and sensitivity analyses

The hypothetical estimand for primary effectiveness outcome is reported in Figure S9. Sensitivity analyses were overall consistent with the main results (Figures S10, S11, S12, S13, S14, S15 and Section H.4).

Discussion

In this large cohort study using observational data from the CPRD-Aurum linked to HES APC and ONS to emulate a target trial, we did not find evidence of effect modification by obesity, using BMI as a proxy, on the effectiveness and safety of apixaban compared to warfarin in non-valvular atrial fibrillation patients.

In the analysis, apixaban was not superior to warfarin in reducing the risk of stroke/SE or all-cause mortality in the overall and across any BMI category. For major bleeding, apixaban showed superiority over warfarin overall and in overweight and obese patients but not in patients with normal weight. Apixaban was also not superior in reducing stroke (any type) or ischemic stroke. For haemorrhagic stroke, apixaban was not superior over warfarin except for obese patients but the study was not powered to answer this question. Events of SE were scarce, resulting in very wide 95% CIs and hindering any meaningful interpretations of the results. The results were robust in sensitivity analyses as none meaningfully deviated from the main analysis.

The similar comparative treatment response across BMI groups in the study was contrary to previous suggestions in the literature that obesity plays a major role in treatment effects of anticoagulants18,19.

Interpretation

Our BMI-stratified analysis was different to the reported results from the ARISTOTLE trial, which investigated the effects in the overall groups rather than within BMI groups5. In ARISTOTLE, apixaban was found to be superior to warfarin in both composite of stroke/SE (HR 0.79; 95% CI: 0.66, 0.95) and major bleeding (HR 0.69; 95% CI: 0.60, 0.80). However, our findings were consistent with the subgroup analysis of EU randomised patients of ARISTOTLE showing an HR of 0.92 (95% CI: 0.56,1.52) for stroke/SE71. Our findings were consistent with the results of the recent study by Deitelzweig et al.(2022) in the United States, which investigated the effects across BMI groups and showed similar conclusions to our study72. Another US-based study by Deitelzweig et al.(2020) investigated the effects in the obese population only and had similar results to the obese group in our study73.

In our study, and similarly, in both US-based cohort studies, the main difference from ARISTOTLE was the lack of superiority for the primary effectiveness outcome (stroke/SE) which could have arisen due to multiple reasons. First, we cannot rule out the possibility of unmeasured confounding. Despite accounting for several confounders in the analysis, there could be psychological and personal factors affecting the choice of prescribed treatment and the risk of the outcome. As the results of ARISTOTLE were published before our study start date and clinical trials usually change the clinical practice, it is possible that treatment choice relied on non-measurable behaviors (e.g., more willingness to comply with INR monitoring despite the availability of other effective treatments).

Second, this could be due to better warfarin control since the median percentage of time in therapeutic range (TTR) (INR 2-3) was 66% for the trial population compared to 76% in the CPRD cohort74.

Third, this could be due to deviations from ARISTOTLE protocol, such as lack of blinding or differences in warfarin monitoring. In ARISTOTLE, nonetheless, apixaban was not superior to warfarin in ischemic stroke which comprised a higher proportion of stroke events, in each BMI group, compared to ARISTOTLE. This could have driven the composite endpoint towards the null.

In terms of major bleeding, our study agreed with previous findings for overweight and obese groups but not for normal weight group. Apixaban appeared to be superior to warfarin in major bleeding events, with the effect being consistent across BMI groups. The rate of major bleeding was similar to that reported in ARISTOTLE.

Our study also employed a slightly different causal contrast compared to the other studies. This study was mainly concerned with the effect of prescribing either apixaban or warfarin on the rates of Stroke/SE or bleeding regardless of treatment switching except for safety.

Strengths and limitations

We used a validated and representative routinely collected primary care data with linkage to secondary care, which improved the completeness of ascertaining study outcomes. We designed the study to emulate a target trial and restricted the enrollment to incident users to minimize bias. Our study also collected and accounted for multiple confounding factors, including demographic, prognostic, social, lifestyle, and socioeconomic factors, including both individual level and practice level deprivation indices. We included a study population similar to those trial-eligible allowing a better comparison with the reference RCT.

Moreover, another strength was the ability to look at BMI groups with acceptable precision. The large sample size (> 50,000 participants) could be looked at from two perspectives. Although this is not largely distinguished in the literature, this study focused on effect modification rather than interaction70. That is, what is the average treatment effect within BMI strata rather than, for example, what is the joint effect of anticoagulation and weight loss (i.e., interaction effect). If one would consider the first case, the study was well-powered. However, to detect an interaction effect, a much larger sample size is needed, which would, probably, render our study underpowered75.

However, our study had some limitations. We had a large proportion of patients being censored with incomplete follow-up. This resulted in more warfarin patients having a complete follow-up compared to those on apixaban leading to differential follow-up, which can be a source of selection bias in the study22. A more comparable approach would have been to split the study into two periods: a recruitment period followed by a follow-up period to allow time for follow-up of the last recruited patient, as in RCTs. This was not feasible in our study since apixaban prescription increased dramatically during the last three years of the study6.

Furthermore, we reported both treatment policy and hypothetical estimands but did not measure treatment compliance, which can affect study estimates76. Lower compliance can attenuate differences between treatments or, in extreme cases, favor one treatment over another. It is more sensible to assume that compliance would be mainly affected in the warfarin arm due to the need of continuous monitoring, which, if true, would make warfarin effects underestimated. The effects in this study can be considered the effect of prescribing apixaban or warfarin at baseline rather than the effect of treatment22,76.

The effect modifier, BMI, can be subjected to measurement error with the categorization of BMI introducing further misclassification77. Although we have repeated the analysis restricting BMI observations to ≤ 3 years without a meaningful difference, BMI is expected to be a time-varying effect modifier that can change with time. Especially if this coincides with a diagnosis of NVAF, for which increased body weight is a major risk factor to sustain78. We excluded patients with missing data when estimating the inverse probability weights. This might impacted our study estimates despite the data were likely missing at random given the routine clinical care setting.

Conclusion

We believe that the results provide evidence of comparative treatment effects between apixaban and warfarin in the investigated BMI groups. Furthermore, our results were similar to the other studies investigating either the overall or BMI-stratum-specific effects of treatments. This provides robust data for evidence-based decisions for both patients and health practitioners regarding treatment choice. This could be more valuable also in times of pandemics where many patients might not be able to maintain the continuous monitoring for warfarin due to their frailty as NVAF patients or where many under-resourced healthcare systems might not have the capability for it.

In conclusion, after emulating the target trial, BMI, as a proxy for obesity, was not an effect modifier of the effectiveness and safety of apixaban compared to warfarin. The effects of the two treatments in the study were similar in obese and overweight patients to normal weight patients across all study outcomes. This provides reassurance about the use of apixaban in these patient populations. However, future research is needed to investigate the generalisability of the results, compare the different available direct oral anticoagulants (DOACs), and assess the effect in extremely obese patients.

Data Availability

Anonymised data from the Clinical Practice Research Datalink (CPRD), Hospital Episodes Statistics-Admitted Patient Care (HES-APC) and mortality data from the Office for National Statistics (ONS) were obtained after ethical approval and cannot be made publicly available.

Conflict of interest

T.Bin Hammad has nothing to disclose. E. Powell is employed at Compass Pathways and is funded by the MRC for this work. P. Baptiste is supported by a GSK studentship. I. Douglas reports grants, and holds stocks in GSK, outside the submitted work. K. Wing has nothing to disclose.

Acknowledgements

Author affiliations: Saudi Food and Drug Authority, Riyadh, Saudi Arabia (Turki Bin Hammad); London School of Hygiene and Tropical Medicine, London, United Kingdom (Turki Bin Hammad, Emma Powell,, Ian Douglas, Kevin Wing); Queen Mary University of London, London, United Kingdom (Paris J Baptiste).

This work was partly supported by the Saudi Food and Drug Authority (SFDA) as a part of a Master of Science (MSc) project for Turki Bin Hammad. The SFDA had no role in the design, review or writing of the manuscript.

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.

Anonymised data from the Clinical Practice Research Datalink (CPRD), Hospital Episodes Statistics-Admitted Patient Care (HES-APC) and mortality data from the Office for National Statistics (ONS) were obtained after ethical approval and cannot be made publicly available. All diagnostic codes can be accessed from https://datacompass.lshtm.ac.uk/id/eprint/3590/. Analysis code is available at: https://github.com/turkimo/apixaban_warfarin_bmi.

The study was approved by the MSc Research Ethics Committee (Ethics Ref: 27537) at the London School of Hygiene & Tropical Medicine and the Independent Scientific Advisory Committee for MHRA Database Research (approved amendment to 19_066).

Footnotes

  • Supplemental files updated

References

  1. 1.↵
    Benjamin EJ, Muntner P, Alonso A, et al. Heart Disease and Stroke Statistics—2019 Update: A Report From the American Heart Association. Circulation. 2019;139(10):e56–e528. doi:10.1161/CIR.0000000000000659
    OpenUrlCrossRefPubMed
  2. 2.
    Benjamin EJ, Wolf PA, D’Agostino RB, Silbershatz H, Kannel WB, Levy D. Impact of Atrial Fibrillation on the Risk of Death. Circulation. 1998;98(10):946–952. doi:10.1161/01.CIR.98.10.946
    OpenUrlAbstract/FREE Full Text
  3. 3.
    Wolf PA, Abbott RD, Kannel WB. Atrial fibrillation as an independent risk factor for stroke: The Framingham Study. Stroke. 1991;22(8):983–988. doi:10.1161/01.str.22.8.983
    OpenUrlAbstract/FREE Full Text
  4. 4.↵
    Odutayo A, Wong CX, Hsiao AJ, Hopewell S, Altman DG, Emdin CA. Atrial fibrillation and risks of cardiovascular disease, renal disease, and death: Systematic review and meta-analysis. BMJ (Clinical research ed). 2016;354:i4482. doi:10.1136/bmj.i4482
    OpenUrlAbstract/FREE Full Text
  5. 5.↵
    Granger CB, Alexander JH, McMurray JJV, et al. Apixaban versus Warfarin in Patients with Atrial Fibrillation. New England Journal of Medicine. 2011;365(11):981–992. doi:10.1056/NEJMoa1107039
    OpenUrlCrossRefPubMedWeb of Science
  6. 6.↵
    Apixaban: BNF Code 0208020Z0 | OpenPrescribing. Accessed July 29, 2022. https://openprescribing.net/chemical/0208020Z0/
  7. 7.↵
    Warfarin sodium: BNF Code 0208020V0 | OpenPrescribing. Accessed July 29, 2022. https://openprescribing.net/chemical/0208020V0/
  8. 8.↵
    Silverman SL. From Randomized Controlled Trials to Observational Studies. The American Journal of Medicine. 2009;122(2):114–120. doi:10.1016/j.amjmed.2008.09.030
    OpenUrlCrossRefPubMedWeb of Science
  9. 9.↵
    Rubin DB. The design versus the analysis of observational studies for causal effects: Parallels with the design of randomized trials. Statistics in Medicine. 2007;26(1):20–36. doi:10.1002/sim.2739
    OpenUrlCrossRefPubMed
  10. 10.↵
    VanderWeele TJ, Arah OA. Unmeasured Confounding for General Outcomes, Treatments, and Confounders. Epidemiology (Cambridge, Mass). 2011;22(1):42–52. doi:10.1097/EDE.0b013e3181f74493
    OpenUrlCrossRefPubMedWeb of Science
  11. 11.↵
    Dahabreh IJ, Robins JM, Hernán MA. Benchmarking Observational Methods by Comparing Randomized Trials and Their Emulations. Epidemiology. 2020;31(5):614–619. doi:10.1097/EDE.0000000000001231
    OpenUrlCrossRef
  12. 12.↵
    Wing K, Williamson E, Carpenter JR, et al. Real world effects of COPD medications: A cohort study with validation against results from randomised controlled trials. European Respiratory Journal. 2021;57(3). doi:10.1183/13993003.01586-2020
    OpenUrlAbstract/FREE Full Text
  13. 13.↵
    Powell EM, Douglas IJ, Gungabissoon U, Smeeth L, Wing K. Real-world effects of medications for stroke prevention in atrial fibrillation: Protocol for a UK population-based non-interventional cohort study with validation against randomised trial results. BMJ Open. 2021;11(4):e042947. doi:10.1136/bmjopen-2020-042947
    OpenUrlAbstract/FREE Full Text
  14. 14.↵
    Emma Powell, Usha Gungabissoon, John Tazare, et al. Comparative treatment effects of oral anticoagulants for stroke prevention in atrial fibrillation: Analysis of routinely collected data with validation against results from a randomised controlled trial. Published online 2023.
  15. 15.↵
    DailyMed - WARFARIN SODIUM tablet. Accessed July 22, 2022. https://dailymed.nlm.nih.gov/dailymed/
  16. 16.
    Blouin RA, Kolpek JH, Mann HJ. Influence of obesity on drug disposition. Clinical Pharmacy. 1987;6(9):706–714.
    OpenUrlPubMedWeb of Science
  17. 17.↵
    Wallace JL, Reaves AB, Tolley EA, et al. Comparison of initial warfarin response in obese patients versus non-obese patients. Journal of Thrombosis and Thrombolysis. 2013;36(1):96–101. doi:10.1007/s11239-012-0811-x
    OpenUrlCrossRefPubMed
  18. 18.↵
    Fender AC, Gawalko M, Dobrev D. Direct oral anticoagulation and severe obesity – One size fits all? IJC Heart & Vasculature. 2021;37:100923. doi:10.1016/j.ijcha.2021.100923
    OpenUrlCrossRef
  19. 19.↵
    Sebaaly J, Kelley D. Direct Oral Anticoagulants in Obesity: An Updated Literature Review. Annals of Pharmacotherapy. 2020;54(11):1144–1158. doi:10.1177/1060028020923584
    OpenUrlCrossRef
  20. 20.↵
    Young JG, Stensrud MJ, Tchetgen Tchetgen EJ, Hernán MA. A causal framework for classical statistical estimands in failure-time settings with competing events. Statistics in Medicine. 2020;39(8):1199–1236. doi:10.1002/sim.8471
    OpenUrlCrossRefPubMed
  21. 21.↵
    Schneeweiss S, Rassen JA, Brown JS, et al. Graphical Depiction of Longitudinal Study Designs in Health Care Databases. Annals of Internal Medicine. 2019;170(6):398–406. doi:10.7326/M18-3079
    OpenUrlCrossRefPubMed
  22. 22.↵
    Ma H, Jm R. Causal Inference: What If (the Book). Boca Raton: Chapman & Hall/CRC.; 2020. Accessed July 6, 2022. https://www.hsph.harvard.edu/miguel-hernan/causal-inference-book/
  23. 23.↵
    Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. American Journal of Epidemiology. 2016;183(8):758–764. doi:10.1093/aje/kwv254
    OpenUrlCrossRefPubMed
  24. 24.↵
    NHS Digital. CPRD Aurum March 2022 dataset | CPRD. Accessed July 2, 2022. https://www.cprd.com/cprd-aurum-march-2022-dataset
  25. 25.↵
    Bhaskaran K, Forbes HJ, Douglas I, Leon DA, Smeeth L. Representativeness and optimal use of body mass index (BMI) in the UK Clinical Practice Research Datalink (CPRD). BMJ Open. 2013;3(9):e003389. doi:10.1136/bmjopen-2013-003389
    OpenUrlAbstract/FREE Full Text
  26. 26.↵
    Herrett E, Gallagher AM, Bhaskaran K, et al. Data Resource Profile: Clinical Practice Research Datalink (CPRD). International Journal of Epidemiology. 2015;44(3):827–836. doi:10.1093/ije/dyv098
    OpenUrlCrossRefPubMed
  27. 27.↵
    NHS Digital. [MI] National Data Opt-out, April 2022. NHS Digital. Accessed July 31, 2022. https://digital.nhs.uk/data-and-information/publications/statistical/national-data-opt-out/april-2022
  28. 28.↵
    Wolf A, Dedman D, Campbell J, et al. Data resource profile: Clinical Practice Research Datalink (CPRD) Aurum. International Journal of Epidemiology. 2019;48(6):1740–1740g. doi:10.1093/ije/dyz034
    OpenUrlCrossRefPubMed
  29. 29.↵
    English indices of deprivation 2015. GOVUK. Accessed July 28, 2022. https://www.gov.uk/government/statistics/english-indices-of-deprivation-2015
  30. 30.↵
    Herbert A, Wijlaars L, Zylbersztejn A, Cromwell D, Hardelid P. Data Resource Profile: Hospital Episode Statistics Admitted Patient Care (HES APC). International Journal of Epidemiology. 2017;46(4):1093–1093i. doi:10.1093/ije/dyx015
    OpenUrlCrossRefPubMed
  31. 31.↵
  32. 32.↵
    NHS Digital. Hospital Episode Statistics (HES). NHS Digital. Accessed August 1, 2022. https://digital.nhs.uk/data-and-information/data-tools-and-services/data-services/hospital-episode-statistics
  33. 33.↵
    Saine ME, Carbonari DM, Newcomb CW, et al. Concordance of hospitalizations between Clinical Practice Research Datalink and linked Hospital Episode Statistics among patients treated with oral antidiabetic therapies. Pharmacoepidemiology and Drug Safety. 2019;28(10):1328–1335. doi:10.1002/pds.4853
    OpenUrlCrossRef
  34. 34.↵
    McDonald L, Schultze A, Carroll R, Ramagopalan SV. Performing studies using the UK Clinical Practice Research Datalink: To link or not to link? European Journal of Epidemiology. 2018;33(6):601–605. doi:10.1007/s10654-018-0389-5
    OpenUrlCrossRef
  35. 35.↵
    Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology (Cambridge, Mass). 1999;10(1):37–48.
    OpenUrlCrossRefPubMedWeb of Science
  36. 36.↵
    Essien UR, Holmes DN, Jackson LR II, et al. Association of Race/Ethnicity With Oral Anticoagulant Use in Patients With Atrial Fibrillation: Findings From the Outcomes Registry for Better Informed Treatment of Atrial Fibrillation II. JAMA Cardiology. 2018;3(12):1174–1182. doi:10.1001/jamacardio.2018.3945
    OpenUrlCrossRef
  37. 37.
    Essien UR, Kim N, Hausmann LRM, et al. Disparities in Anticoagulant Therapy Initiation for Incident Atrial Fibrillation by Race/Ethnicity Among Patients in the Veterans Health Administration System. JAMA Network Open. 2021;4(7):e2114234. doi:10.1001/jamanetworkopen.2021.14234
    OpenUrlCrossRef
  38. 38.
    Lip GYH, Kamath S, Jafri M, Mohammed A, Bareford D. Ethnic Differences in Patient Perceptions of Atrial Fibrillation and Anticoagulation Therapy. Stroke. 2002;33(1):238–242. doi:10.1161/hs0102.101817
    OpenUrlAbstract/FREE Full Text
  39. 39.
    Mathur R, Pollara E, Hull S, Schofield P, Ashworth M, Robson J. Ethnicity and stroke risk in patients with atrial fibrillation. Heart. 2013;99(15):1087–1092. doi:10.1136/heartjnl-2013-303767
    OpenUrlAbstract/FREE Full Text
  40. 40.
    Chung SC, Sofat R, Acosta-Mena D, et al. Atrial fibrillation epidemiology, disparity and healthcare contacts: A population-wide study of 5.6 million individuals. The Lancet Regional Health - Europe. 2021;7:100157. doi:10.1016/j.lanepe.2021.100157
    OpenUrlCrossRef
  41. 41.
    Marshall IJ, Wang Y, Crichton S, McKevitt C, Rudd AG, Wolfe CDA. The effects of socioeconomic status on stroke risk and outcomes. The Lancet Neurology. 2015;14(12):1206–1218. doi:10.1016/S1474-4422(15)00200-8
    OpenUrlCrossRef
  42. 42.
    R F Dunn P, Asaria M, Thorlby R. Level or not? The Health Foundation. Published online 2020. doi:10.37829/HF-2020-RC13
    OpenUrlCrossRef
  43. 43.
    Walton E, Ahmed A, Burton C, Mathers N. Influences of socioeconomic deprivation on GPs’ decisions to refer patients to cardiology: A qualitative study. The British Journal of General Practice. 2018;68(677):e826–e834. doi:10.3399/bjgp18X699785
    OpenUrlAbstract/FREE Full Text
  44. 44.
    Nussbaum C, Massou E, Fisher R, Morciano M, Harmer R, Ford J. Inequalities in the distribution of the general practice workforce in England: A practice-level longitudinal analysis. BJGP Open. 2021;5(5). doi:10.3399/BJGPO.2021.0066
    OpenUrlAbstract/FREE Full Text
  45. 45.
    Overvad TF, Rasmussen LH, Skjøth F, et al. Alcohol intake and prognosis of atrial fibrillation. Heart. 2013;99(15):1093–1099. doi:10.1136/heartjnl-2013-304036
    OpenUrlAbstract/FREE Full Text
  46. 46.
    Patra J, Taylor B, Irving H, et al. Alcohol consumption and the risk of morbidity and mortality for different stroke types - a systematic review and meta-analysis. BMC Public Health. 2010;10(1):258. doi:10.1186/1471-2458-10-258
    OpenUrlCrossRefPubMed
  47. 47.↵
    Apostolakis S, Sullivan RM, Olshansky B, Lip GYH. Factors Affecting Quality of Anticoagulation Control Among Patients With Atrial Fibrillation on Warfarin: The SAMe-TT2R2 Score. Chest. 2013;144(5):1555–1563. doi:10.1378/chest.13-0054
    OpenUrlCrossRefPubMedWeb of Science
  48. 48.
    Arai H, Ueda S, Uchida K, et al. Effect of current smoking on ischemic events in patients with atrial fibrillation taking vitamin K antagonist. International Journal of Cardiology Cardiovascular Risk and Prevention. 2022;14:200135. doi:10.1016/j.ijcrp.2022.200135
    OpenUrlCrossRef
  49. 49.
    Bonita R, Scragg R, Stewart A, Jackson R, Beaglehole R. Cigarette smoking and risk of premature stroke in men and women. Br Med J (Clin Res Ed). 1986;293(6538):6–8. doi:10.1136/bmj.293.6538.6
    OpenUrlAbstract/FREE Full Text
  50. 50.
    Grisold W, Oberndorfer S, Struhal W. Stroke and cancer: A review. Acta Neurologica Scandinavica. 2009;119(1):1–16. doi:10.1111/j.1600-0404.2008.01059.x
    OpenUrlCrossRefPubMedWeb of Science
  51. 51.
    Lee YJ, Park J, Uhm JS, et al. Bleeding risk and major adverse events in patients with cancer on oral anticoagulation therapy. International Journal of Cardiology. 2016;203:372–378. doi:10.1016/j.ijcard.2015.10.166
    OpenUrlCrossRef
  52. 52.
    Navi BB, Kasner SE, Elkind MSV, Cushman M, Bang OY, DeAngelis LM. Cancer and Embolic Stroke of Undetermined Source. Stroke. 2021;52(3):1121–1130. doi:10.1161/STROKEAHA.120.032002
    OpenUrlCrossRefPubMed
  53. 53.
    Bonde AN, Lip GYH, Kamper AL, et al. Renal Function and the Risk of Stroke and Bleeding in Patients With Atrial Fibrillation. Stroke. 2016;47(11):2707–2713. doi:10.1161/STROKEAHA.116.014422
    OpenUrlAbstract/FREE Full Text
  54. 54.
    Del-Carpio Munoz F, Gharacholou SM, Munger TM, et al. Meta-Analysis of Renal Function on the Safety and Efficacy of Novel Oral Anticoagulants for Atrial Fibrillation. The American Journal of Cardiology. 2016;117(1):69–75. doi:10.1016/j.amjcard.2015.09.046
    OpenUrlCrossRefPubMed
  55. 55.
    Alonso A, MacLehose RF, Chen LY, et al. Prospective study of oral anticoagulants and risk of liver injury in patients with atrial fibrillation. Heart. 2017;103(11):834–839. doi:10.1136/heartjnl-2016-310586
    OpenUrlAbstract/FREE Full Text
  56. 56.↵
    Hamaguchi M, Kojima T, Takeda N, et al. Nonalcoholic fatty liver disease is a novel predictor of cardiovascular disease. World Journal of Gastroenterology : WJG. 2007;13(10):1579–1584. doi:10.3748/wjg.v13.i10.1579
    OpenUrlCrossRefPubMedWeb of Science
  57. 57.↵
    Engell AE, Svendsen ALO, Lind BS, et al. Drug-drug interaction between warfarin and statins: A Danish cohort study. British Journal of Clinical Pharmacology. 2021;87(2):694–699. doi:10.1111/bcp.14428
    OpenUrlCrossRef
  58. 58.↵
    Holbrook AM, Pereira JA, Labiris R, et al. Systematic Overview of Warfarin and Its Drug and Food Interactions. Archives of Internal Medicine. 2005;165(10):1095–1106. doi:10.1001/archinte.165.10.1095
    OpenUrlCrossRefPubMedWeb of Science
  59. 59.
    Berlowitz DR, Miller DR, Oliveria SA, Cunningham F, Gomez-Caminero A, Rothendler JA. Differential associations of beta-blockers with hemorrhagic events for chronic heart failure patients on warfarin. Pharmacoepidemiology and Drug Safety. 2006;15(11):799–807. doi:10.1002/pds.1301
    OpenUrlCrossRefPubMedWeb of Science
  60. 60.
    Flaker G, Lopes RD, Hylek E, et al. Amiodarone, anticoagulation, and clinical events in patients with atrial fibrillation: Insights from the ARISTOTLE trial. Journal of the American College of Cardiology. 2014;64(15):1541–1550. doi:10.1016/j.jacc.2014.07.967
    OpenUrlFREE Full Text
  61. 61.
    Hankey GJ. Angiotensin-Converting Enzyme Inhibitors for Stroke Prevention. Stroke. 2003;34(2):354–356. doi:10.1161/01.STR.0000054261.97525.4B
    OpenUrlFREE Full Text
  62. 62.↵
    Healey JS, Baranchuk A, Crystal E, et al. Prevention of atrial fibrillation with angiotensin-converting enzyme inhibitors and angiotensin receptor blockers: A meta-analysis. Journal of the American College of Cardiology. 2005;45(11):1832–1839. doi:10.1016/j.jacc.2004.11.070
    OpenUrlFREE Full Text
  63. 63.↵
    Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. Journal of Chronic Diseases. 1987;40(5):373–383. doi:10.1016/0021-9681(87)90171-8
    OpenUrlCrossRefPubMedWeb of Science
  64. 64.
    Mongkhon P, Alwafi H, Fanning L, et al. Patterns and factors influencing oral anticoagulant prescription in people with atrial fibrillation and dementia: Results from UK primary care. British Journal of Clinical Pharmacology. 2021;87(3):1056–1068. doi:10.1111/bcp.14464
    OpenUrlCrossRef
  65. 65.↵
    Rouaud A, Hanon O, Boureau AS, Chapelet GG, Decker L de. Comorbidities against Quality Control of VKA Therapy in Non-Valvular Atrial Fibrillation: A French National Cross-Sectional Study. PLoS ONE. 2015;10(3):e0119043. doi:10.1371/journal.pone.0119043
    OpenUrlCrossRef
  66. 66.↵
    O’Brien EC, Simon DN, Thomas LE, et al. The ORBIT bleeding score: A simple bedside score to assess bleeding risk in atrial fibrillation. European Heart Journal. 2015;36(46):3258–3264. doi:10.1093/eurheartj/ehv476
    OpenUrlCrossRefPubMed
  67. 67.↵
    Pisters R, Lane DA, Nieuwlaat R, Vos CB de, 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):1093–1100. doi:10.1378/chest.10-0134
    OpenUrlCrossRefPubMedWeb of Science
  68. 68.↵
    VanderWeele TJ, Robins JM. Four types of effect modification: A classification based on directed acyclic graphs. Epidemiology (Cambridge, Mass). 2007;18(5):561–568. doi:10.1097/EDE.0b013e318127181b
    OpenUrlCrossRefPubMedWeb of Science
  69. 69.↵
    Knol MJ, VanderWeele TJ. Recommendations for presenting analyses of effect modification and interaction. International Journal of Epidemiology. 2012;41(2):514–520. doi:10.1093/ije/dyr218
    OpenUrlCrossRefPubMedWeb of Science
  70. 70.↵
    VanderWeele TJ. On the Distinction Between Interaction and Effect Modification. Epidemiology. 2009;20(6):863–871. doi:10.1097/EDE.0b013e3181ba333c
    OpenUrlCrossRefPubMedWeb of Science
  71. 71.↵
    European Medicines Agency. EMEA/H/C/002148/X/04/G Eliquis Assessment report. Published online September 2012. https://www.ema.europa.eu/en/documents/variation-report/eliquis-h-c-2148-x-0004-g-epar-assessment-report-extension_en.pdf
  72. 72.↵
    Deitelzweig S, Sah J, Kang A, et al. Effectiveness and Safety of Apixaban Versus Warfarin in Obese Patients with Nonvalvular Atrial Fibrillation Enrolled in Medicare and Veteran Affairs. The American Journal of Cardiology. 2022;163:43–49. doi:10.1016/j.amjcard.2021.09.047
    OpenUrlCrossRef
  73. 73.↵
    Deitelzweig S, Keshishian A, Kang A, et al. Effectiveness and Safety of Oral Anticoagulants among NVAF Patients with Obesity: Insights from the ARISTOPHANES Study. Journal of Clinical Medicine. 2020;9(6):E1633. doi:10.3390/jcm9061633
    OpenUrlCrossRef
  74. 74.↵
    Powell, E, Gungabissoon, U, Tazare, J, Smeeth, L, Douglas, I., Wing, K. Can we replicate an RCT in CPRD Aurum? Real world effects of oral anticoagulants in AF: A cohort study with validation against results from a randomised controlled trial. In: Spotlight poster presentation at the 38th International Conference on Pharmacoepidemiology; Therapeutic Risk Management, Copenhagen, Denmark.
  75. 75.↵
    Hsieh FY, Lavori PW, Cohen HJ, Feussner JR. An Overview of Variance Inflation Factors for Sample-Size Calculation. Evaluation & the Health Professions. 2003;26(3):239–257. doi:10.1177/0163278703255230
    OpenUrlCrossRefPubMedWeb of Science
  76. 76.↵
    Hernán MA, Hernández-Díaz S. Beyond the intention to treat in comparative effectiveness research. Clinical trials (London, England). 2012;9(1):48–55. doi:10.1177/1740774511420743
    OpenUrlCrossRefPubMedWeb of Science
  77. 77.↵
    Groenwold RHH, Klungel OH, Altman DG, Graaf Y van der, Hoes AW, Moons KGM. Adjustment for continuous confounders: An example of how to prevent residual confounding. CMAJ : Canadian Medical Association Journal. 2013;185(5):401–406. doi:10.1503/cmaj.120592
    OpenUrlFREE Full Text
  78. 78.↵
    Lavie CJ, Pandey A, Lau DH, Alpert MA, Sanders P. Obesity and Atrial Fibrillation Prevalence, Pathogenesis, and Prognosis. Journal of the American College of Cardiology. 2017;70(16):2022–2035. doi:10.1016/j.jacc.2017.09.002
    OpenUrlFREE Full Text
  79. 79.
    NICE. Identification and classification | Diagnosis | Obesity | CKS | NICE. Accessed July 30, 2022. https://cks.nice.org.uk/topics/obesity/diagnosis/identification-classification/
Back to top
Posted October 16, 2023.
Download PDF

Supplementary Material

Data/Code
Email
Emulating a target trial to assess effect modification: an application to obesity in the comparative effectiveness and safety of apixaban versus warfarin in non-valvular atrial fibrillation using electronic health records
Turki M Bin Hammad, Emma Powell, Paris J Baptiste, Ian Douglas, Kevin Wing
medRxiv 2023.10.12.23296756; doi: https://doi.org/10.1101/2023.10.12.23296756
Twitter logo Facebook logo LinkedIn logo Mendeley logo
Citation Tools

Subject Area