Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Risk of heart failure in elderly patients with atrial fibrillation and diabetes taking different oral anticoagulants: a nationwide cohort study

Risk of heart failure in elderly patients with atrial fibrillation and diabetes taking different... Background Heart failure (HF) is a critical complication in elderly patients with atrial fibrillation (AF) and diabetes mellitus (DM). Recent preclinical studies suggested that non‑ vitamin K antagonist oral anticoagulants (NOACs) can potentially suppress the progression of cardiac fibrosis and ischemic cardiomyopathy. Whether different oral antico ‑ agulants influence the risk of HF in older adults with AF and DM is unknown. This study aimed to evaluate the risk of HF in elderly patients with AF and DM who were administered NOACs or warfarin. Methods A nationwide retrospective cohort study was conducted based on claims data from the entire Taiwan‑ ese population. Target trial emulation design was applied to strengthen causal inference using observational data. Patients aged ≥ 65 years with AF and DM on NOAC or warfarin treatment between 2012 and 2019 were included and followed up until 2020. The primary outcome was newly diagnosed HF. Propensity score‑based fine stratification weightings were used to balance patient characteristics between NOAC and warfarin groups. Hazard ratios (HRs) were estimated using Cox proportional hazard models. Results The study included a total of 24,835 individuals (19,710 NOAC and 5,125 warfarin users). Patients taking NOACs had a significantly lower risk of HF than those taking warfarin (HR = 0.80, 95% CI 0.74–0.86, p < 0.001). Sub‑ group analyses for individual NOACs suggested that dabigatran (HR = 0.86, 95% CI 0.80–0.93, p < 0.001), rivaroxaban (HR = 0.80, 95% CI 0.74–0.86, p < 0.001), apixaban (HR = 0.78, 95% CI 0.68–0.90, p < 0.001), and edoxaban (HR = 0.72, 95% CI 0.60–0.86, p < 0.001) were associated with lower risks of HF than warfarin. The findings were consistent regard‑ less of age and sex subgroups and were more prominent in those with high medication possession ratios. Several sensitivity analyses further supported the robustness of our findings. Conclusions This nationwide cohort study demonstrated that elderly patients with AF and DM taking NOACs had a lower risk of incident HF than those taking warfarin. Our findings suggested that NOACs may be the preferred Huei‑Kai Huang and Ching‑Hui Loh have contributed equally to this work *Correspondence: Huei‑Kai Huang drhkhuang@gmail.com Ching‑Hui Loh twdoc1960@gmail.com Full list of author information is available at the end of the article © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 2 of 11 oral anticoagulant treatment when considering the prevention of heart failure in this vulnerable population. Future research is warranted to elucidate causation and investigate the underlying mechanisms. Keywords Oral anticoagulant, Heart failure, Atrial fibrillation, Diabetes mellitus, Elderly Background However, to date, the evidence comparing the risk of In the elderly population, atrial fibrillation (AF) and HF between NOAC and warfarin use is still lacking, even diabetes mellitus (DM) are both global epidemics and though this issue is critical for improving patient progno- important public health problems [1, 2]. Due to their sis in elderly adults already with AF and DM. Therefore, high prevalence and incidence, these two chronic con- we used nationwide cohort data to investigate the risk of ditions commonly coexist. Heart failure (HF), another HF development in elderly patients with AF and DM tak- global epidemic, affects at least 26 million people world - ing NOAC versus warfarin. wide and is one of the leading causes of morbidity, hos- pitalization, and mortality in older adults, placing a huge Methods financial burden on the health care system [3, 4]. Current Data sources evidence indicates that HF is a critical complication in We conducted a nationwide retrospective cohort study patients with AF and DM. Hyperglycemia, insulin resist- using data from the National Health Insurance Research ance, and hyperinsulinemia in DM can trigger a cascade Database (NHIRD) in Taiwan. The National Health of deleterious effects contributing to development of HF Insurance program, a mandatory single-payer program and effort intolerance [5–7]. The tachycardia, irregularity, administered by Taiwan’s government, covered more loss of atrial systole, and cardiac fibrosis in patients with than 99% of the entire population in Taiwan (approxi- AF also contribute to HF development [4, 8]. Since AF, mately 23.6 million individuals) [21, 22]. The NHIRD DM, and aging are all major risk factors of HF [4, 5, 9] contains patient demographic information and medical and concomitant HF in elderly patients with AF and DM claims for all inpatient, outpatient, and emergency care could increase their risk of stroke, worsen patient prog- services in Taiwan. The diagnostic and procedure codes noses, and increase the healthcare cost burden [5, 10, 11], in NHIRD were derived using the International Classifi - the prevention of HF development in the elderly popula- cation of Diseases, Ninth Revision, Clinical Modification tion with AF and DM is crucial. (ICD-9-CM) codes before 2016 and the International Long-term oral anticoagulant treatment is an essential Classification of Diseases, Tenth Revision, Clinical Modi - medication for stroke prevention in elderly patients with fication (ICD-10-CM) codes since 2016. Information on AF and DM [12, 13]. Warfarin, a vitamin K antagonist, mortality was obtained by cross-referencing the NHIRD has been used to prevent stroke for decades. Non-vita- with the Taiwan National Register of Deaths. The NHIRD min K antagonist oral anticoagulants (NOACs) have been is maintained by the Health and Welfare Data Science approved as another choice of oral anticoagulants and Center, Ministry of Health and Welfare, Taiwan, and the have been found to offer comparable efficacy and safety anonymized data has been made available for research for stroke prevention [14]. In addition to its anticoagu- purposes by formal application. Our study was approved lation effect, recent studies have suggested that NOACs, by the Research Ethics Committee of Hualien Tzu Chi compared with warfarin, were linked to better glycemic Hospital (REC No: IRB107-152-C); the requirement for control and lower diabetes complication risks [15–17]. informed consent was waived due to the retrospective Additionally, recent preclinical studies suggested that use of anonymized data. This study was conducted in NOACs have potential anti-inflammatory effects and accordance with the World Medical Association Declara- may suppress the progression of cardiac fibrosis and tion of Helsinki. ischemic cardiomyopathy, all of which are related to the pathophysiology of HF [18–20]. Thus, it is reasonable to Study population suppose that NOACs may have a beneficial effect on pre - We conducted the observational study with target trial venting HF compared with warfarin in patients with AF emulation to strengthen causal inference [23, 24]; the and DM. As HF is also an independent and potent risk details of how we emulated a target trial are described in factor for stroke development [11], of which oral antico- Additional file  1: Table  S1. We applied similar selection agulants are mainly prescribed for prevention, choosing criteria to those of the target trial to include all adults appropriate oral anticoagulant types to decrease incident aged ≥ 65 years with diagnoses of both AF and DM who HF risks is crucial. Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 3 of 11 had been treated with oral anticoagulants between 2012 outpatient basis within the year prior to the index date. and 2019 in NHIRD. The ICD-9-CM code 427.31 and The Charlson comorbidity index was calculated to quan - ICD-10-CM codes I48.0, I48.1, I48.2, and I48.91 were tify the overall comorbidity status [26]. We also calcu- used to identify AF diagnosis; the ICD-9-CM code 250 or lated the CHA2DS2-VASc score, which is used to predict ICD-10-CM codes E08-E13 were used for DM diagnosis. stroke risk and determine whether an oral anticoagulant Both diagnoses should be made at least once in an inpa- should be used in clinical practice [27, 28]. We defined tient service or twice in outpatient services. We restricted baseline antidiabetic drugs based on the treatment pre- our study population to patients aged ≥ 65 years because scribed within 1 month prior to the index date; the num- both AF and HF developed mainly in older adults. ber of diabetes medication types were also calculated. We excluded patients without AF and DM diagnoses Other baseline medications were defined as a drug pre - at baseline. We excluded patients with end-stage renal scribed for at least 30  days within the year prior to the disease (ESRD), rheumatic heart disease, congenital index date. The duration of AF and DM were defined heart disease, or having valve replacement surgery before as the period from the date of first diagnosis of AF or the index date because those patients are more likely to DM to the date of initiating oral anticoagulants (index receive warfarin over NOACs [25], and their exclusion date). The index year, monthly income (derived from helped minimize a potential confounding-by-indication income-related insurance premiums), physician’s medical bias. To apply the new-user design, those with a prescrip- specialty, and the hospital level of oral anticoagulant ini- tion of any oral anticoagulants in 2011 were excluded, tiation were also retrieved as covariates [17]. enhancing the likelihood of identifying new oral antico- agulant users since 2012 when NOACs were introduced Propensity score‑based fine stratification weighting in Taiwan’s National Health Insurance program. We We calculated the propensity score for each patient to excluded those with index dates in 2020, ensuring at least estimate the probability of initiating NOACs using mul- 1-year follow-up for each patient. Finally, we excluded tivariable logistic regression models, including all covari- patients with any prior HF diagnoses before the index ates shown in Table 1. We used fine stratification weights date (Additional file 1: Figure S1). based on propensity scores to create more exchangeable groups with balanced characteristics for comparisons. Exposures, outcomes, and follow‑up Two fine stratification weighting methods were applied to To emulate a target trial with intention-to-treat analysis, cover both targets of inference: estimation of the average we used an as-started design that divided patients into treatment effect in the whole population (ATE) and esti - NOAC and warfarin groups according to their first oral mation of the average treatment effect among the treated anticoagulant use regardless of subsequent prescriptions population (ATT) [29]. The individuals were stratified [24]. The index date (time zero of follow-up) was defined into 50 strata by the propensity score distribution; how as the date of initiation of oral anticoagulant treatment, the weights were calculated in each stratum is described and follow-up began since then. elsewhere [29]. The propensity score-based fine stratifica - The primary outcome was the incident HF diagnosed tion weighting was conducted individually for each com- in an inpatient service or at least three times in an out- parison set, including that of overall analyses, subgroup patient service (ICD-9-CM code: 428; ICD-10-CM code: analyses, stratified analyses, or sensitivity analyses. I50). The date of the first HF diagnosis was assigned as the date of event occurrence. We followed up with each Statistical analyses patient from their index date until an occurrence of the The difference in baseline characteristics was determined outcome event, death, or December 31, 2020 (the last by standardized difference, with a value of < 0.1 consid- date in our database), whichever came first. ered negligible. The standardized difference is preferred In our main analyses, we compared HF risk between to significance testing of covariates between study groups overall NOACs versus warfarin. We further per- because it is not confounded by sample sizes or the statis- formed subgroup analyses that subclassified NOACs tical power [30]. We estimated the cumulative incidences into four subgroups (dabigatran, rivaroxaban, apixa- and cause-specific hazard ratios (HRs) of HF using cause- ban, and edoxaban) and compared each with warfarin. specific Cox proportional hazard models with death We also performed analyses stratified by age (65–74 treated as a censoring event [31]. To address the poten- and ≥ 75 years), sex, and hospital levels. tial cluster effect and variation from each different hos - pital or clinic (where oral anticoagulant treatment was Covariates and confounders initiated), we included shared frailty, estimating the clus- Pre-existing comorbidity was defined as a condition diag - ter random effect of hospital/clinic, into the regression nosed at least once on an inpatient basis or twice on an model [32, 33]. A two-tailed probability (p) value < 0.05 Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 4 of 11 Table 1 Baseline characteristics of older patients with atrial fibrillation and diabetes receiving NOAC or warfarin after propensity score‑based fine stratification weighting * ** Population with fine stratification weights (ATE) Population with fine stratification weights (ATT) † † NOAC (N = 19,591) Warfarin (N = 5,117) SMD NOAC (N = 19,591) Warfarin (N = 5,117) SMD Age (years) 76.6 ± 7.3 76.7 ± 7.6 0.013 76.8 ± 7.4 76.9 ± 7.7 0.013 Sex Male 52.3 53.0 0.014 52.3 53.6 0.026 Female 47.7 47.0 0.014 47.7 46.4 0.026 ‡§ Charlson comorbidity index 2.7 ± 2.0 2.7 ± 1.9 0.000 2.7 ± 2.0 2.6 ± 1.9 0.051 ‡# CHA2DS2‑VASc score 4.3 ± 1.5 4.2 ± 1.5 0.067 4.2 ± 1.5 4.1 ± 1.6 0.065 Comorbidities Hypertension 78.1 77.1 0.024 77.5 76.3 0.029 Coronary artery disease 30.5 30.6 0.002 30.0 30.2 0.004 COPD 12.9 13.6 0.021 12.7 13.6 0.027 Chronic kidney disease 13.7 14.3 0.017 13.2 14.1 0.026 Cirrhosis 4.0 4.5 0.025 3.5 4.2 0.036 Hyperlipidemia 39.8 39.3 0.010 40.3 39.8 0.010 Stroke 32.2 29.9 0.050 32.0 28.6 0.074 Rheumatoid arthritis 0.8 0.7 0.012 0.8 0.7 0.012 Gout 9.6 9.5 0.003 9.2 9.1 0.004 Dementia 7.3 6.8 0.020 7.6 6.9 0.027 Malignancy 9.4 9.3 0.003 9.5 9.4 0.003 Medication use Statins 40.2 39.7 0.010 41.3 40.8 0.010 ACEI or ARB 61.7 61.8 0.002 61.9 62.0 0.002 β blockers 44.5 45.8 0.026 44.3 45.9 0.032 Calcium channel blockers 46.9 47.4 0.010 45.9 46.6 0.014 Diuretics 22.3 23.7 0.033 21.0 23.0 0.048 NSAID 33.4 32.6 0.017 33.4 32.3 0.023 Corticosteroids 5.8 5.7 0.004 5.7 5.6 0.004 Antipsychotics 5.3 5.2 0.005 5.1 5.1 0.000 Proton pump inhibitors 9.0 8.2 0.029 9.1 8.0 0.039 Baseline diabetes medications Metformin 48.6 47.4 0.024 49.3 47.5 0.036 Sulfonylurea 31.4 30.4 0.022 30.1 28.9 0.026 Meglitinide 5.7 5.5 0.009 5.2 5.0 0.009 AGI 7.7 8.7 0.037 7.3 8.6 0.048 TZD 4.8 4.6 0.010 4.8 4.4 0.019 DPP‑4i 26.7 27.7 0.023 27.6 29.1 0.033 SGLT‑2i 1.7 2.0 0.022 2.1 2.4 0.020 GLP‑1 RA 0.2 0.2 0.000 0.2 0.2 0.000 Insulin 10.2 10.0 0.007 9.9 9.8 0.003 Numbers of diabetes medications Without medications 33.3 33.8 0.011 33.7 34.5 0.017 1 type 24.4 23.3 0.026 24.2 22.8 0.033 2 types 21.4 22.4 0.024 21.3 22.0 0.017 ≥ 3 types 20.9 20.5 0.010 20.8 20.7 0.003 Duration of diabetes < 2 years 21.8 20.6 0.029 18.0 15.8 0.059 ≥ 2 years 78.2 79.4 0.029 82.0 84.2 0.059 Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 5 of 11 Table 1 (continued) * ** Population with fine stratification weights (ATE) Population with fine stratification weights (ATT) † † NOAC (N = 19,591) Warfarin (N = 5,117) SMD NOAC (N = 19,591) Warfarin (N = 5,117) SMD Duration of AF < 2 years 71.2 68.5 0.059 69.9 66.0 0.084 ≥ 2 years 28.8 31.5 0.059 30.1 34.0 0.084 Index year 2012–2013 13.3 13.5 0.006 6.5 6.6 0.004 2014–2015 23.6 24.0 0.009 21.2 21.7 0.012 2016–2017 30.4 28.9 0.033 33.9 32.1 0.038 2018–2019 32.7 33.6 0.019 38.4 39.5 0.023 Income level (NTD) Financially dependent 29.5 29.2 0.007 29.5 29.1 0.009 15,840–29,999 47.8 47.9 0.002 47.2 47.6 0.008 30,000–44,999 11.2 11.1 0.003 11.2 11.0 0.006 ≥ 45,000 11.6 11.7 0.003 12.2 12.2 0.000 Hospital level of OAC initiation Medical center 36.2 34.1 0.044 38.2 35.2 0.062 Regional hospital 44.7 47.0 0.046 44.4 47.3 0.058 District hospital or clinic 19.1 19.0 0.003 17.5 17.5 0.000 Physician specialty Cardiologist 63.7 66.2 0.052 65.2 68.5 0.070 Neurologist 19.5 16.7 0.073 20.3 16.4 0.101 Others 16.8 17.2 0.011 14.5 15.0 0.014 Data are presented as percentages unless otherwise noted ACEI angiotensin-converting enzyme inhibitors, AF atrial fibrillation, AGI alpha-glucosidase inhibitors, ARB angiotensin II receptor blockers, ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, COPD chronic obstructive pulmonary disease, DPP-4i dipeptidyl peptidase-4 inhibitors, GLP-1 RA glucagon-like peptide-1 receptor agonists, IPTW inverse probability of treatment weighting, NOAC non-vitamin K antagonist oral anticoagulant, NSAID nonsteroidal anti-inflammatory drugs, NTD New Taiwan Dollar, OAC oral anticoagulant, PSM propensity score matching, SGLT-2i sodium-glucose cotransporter-2 inhibitors, SMD standardized mean difference, TZD thiazolidinedione The pseudo-population constructed by propensity score-based fine stratification weighting to estimate the average treatment effect in the whole population ** The pseudo-population constructed by propensity score-based fine stratification weighting to estimate the average treatment effect among the treated population A standardized mean difference of < 0.1 indicates a negligible difference Presented as mean ± standard deviation Calculated without scores for age Congestive heart failure, hypertension, age ≥ 75 years, diabetes mellitus, stroke or transient ischemic attack, vascular disease, age 65–74 years, sex category (CHA2DS2-VASc) score The period from the date of first diagnosis of diabetes or AF to the index date was considered statistically significant. We managed data discontinued. Discontinuation was defined as patients and performed statistical analyses using SAS software, without a refilled prescription of the index oral anti - version 9.4 (SAS Institute, Inc., Cary, NC, USA) and coagulant 90  days after the last prescription. Second, STATA, version 15 (Stata Corporation LLC, College Sta- we restricted our analysis to those taking the index oral tion, TX, USA). anticoagulant with a high medication possession ratio, defined as ≥ 80%. The medication possession ratio was Sensitivity analyses calculated by dividing the number of days with prescrip- Various sensitivity analyses were conducted to determine tion of oral anticoagulants by the days of the follow-up the robustness of our study results. First, to consider period [34]. Third, we excluded patients with any diag - the treatment adherence during follow-up, we applied noses of chronic kidney disease (CKD) before the index the on-treatment design (analog of per-protocol design date since we could not obtain individuals’ renal function in clinical trials) in which the follow-up would be cen- data, which may influence the choice of NOACs. Fourth, sored when the oral anticoagulant type was switched or to determine whether potential variations between Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 6 of 11 Risk of incident HF physicians who initiated the NOAC or warfarin prescrip- In the analysis with propensity score-based fine strati - tion influenced our results, we performed a sensitivity fication weighting for ATE estimation, NOAC use was analysis that included shared frailty, estimating the clus- significantly associated with a lower risk of develop - ter random effect of different physicians, into the regres - ing HF than warfarin use (HR = 0.80, 95% confidence sion model. Additionally, we performed two sensitivity interval CI 0.74–0.86, p < 0.001). In the ATT estimation analyses with different statistical designs. One sensitivity analysis, a similar result of lower HF risk in NOAC users analysis applied propensity  score matching (rather than was observed (HR = 0.77, 95% CI 0.70–0.84, p < 0.001) fine stratification) was performed to balance patient (Table  2). Figure  1 illustrates the curves for cumula- characteristics between groups; the matching was based tive HF incidences in patients taking NOACs and those on the nearest-neighbor matching algorithm without taking warfarin; a lower cumulative HF incidence was replacement, with a caliper width equal to 0.2 stand- observed in NOAC users. The curves for estimating ATE ard deviation of the logit of the propensity score [35, and ATT are shown in Fig. 1A and B, respectively. 36]. Another sensitivity analysis estimated the adjusted In the ATE estimation analyses for each NOAC, dabi- HRs by multivariable Cox regression models based on gatran (HR = 0.86, 95% CI 0.80–0.93, p < 0.001), rivaroxa- the original cohort without applying propensity score ban (HR = 0.80, 95% CI 0.74–0.86, p < 0.001), apixaban methods. (HR = 0.78, 95% CI 0.68–0.90, p < 0.001), and edoxaban (HR = 0.72, 95% CI 0.60–0.86, p < 0.001) were all associ- Results ated with a lower HF risk when compared with warfarin Patient characteristics (Table  3). The ATT estimation analyses demonstrated We initially included 24,835 patients (19,710 NOAC similar findings. and 5,125 warfarin users) after applying the inclusion In the analyses stratified by age, sex, and hospital levels, and exclusion criteria; the patient characteristics in the consistent findings were observed; the significantly lower original cohort are shown in Additional file  1: Table  S2. HF risk associated with NOAC use was observed in all For further analyses, we constructed pseudo-popula- stratified groups, regardless of age, sex, or hospital levels tions containing 19,591 NOAC and 5,117 warfarin users (Table 4). after applying propensity score-based fine stratification weighting. The patient characteristics in the weighted Results of sensitivity analyses population for ATE and ATT estimation are presented With the application of an on-treatment design, NOAC in Table  1. The mean age was approximately 76.6  years, users still demonstrated a lower HF risk than warfarin and female patients accounted for 47% of all participants. users (HR = 0.67, 95% CI 0.60–0.75, p < 0.001) in the ATE The mean follow-up duration was 3.0 years. Patient char - estimation analysis (Table  5). In the analysis restricted acteristics were balanced appropriately between groups to only patients with a high medication possession after fine stratification weighting, with standardized dif - ratio (≥ 80%), a more remarkable association between ferences < 0.1. The flowchart of patient selection is pre - NOAC use and lower HF risk was observed (HR = 0.47, sented in Additional file 1: Figure S1. Table 2 Risk of heart failure in older patients with atrial fibrillation and diabetes receiving NOAC versus warfarin Event no Person‑ yearsIncidence rate HR (95% CI) p‑ value Fine stratification weights estimating ATE NOAC (N = 19,591) 4512 59,298 76.1 0.80 (0.74–0.86) < 0.001 Warfarin (N = 5,117) 1404 14,677 95.6 1 (ref.) ** Fine stratification weights estimating ATT NOAC (N = 19,591) 4158 55,059 75.5 0.77 (0.70–0.84) < 0.001 Warfarin (N = 5,117) 1343 13,576 98.9 1 (ref.) ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, CI confidence interval, HR hazard ratio, NOAC non- vitamin K antagonist oral anticoagulant, ref. reference Propensity score-based fine stratification weighting which estimated the average treatment effect in the whole population ** Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population Incidence rate, per 1000 person-years Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 7 of 11 Fig. 1 The cumulative incidence curves of HF in elderly patients with AF and DM taking NOACs and those taking warfarin. The curves were estimated according to the pseudo‑populations constructed by A propensity score ‑based fine stratification weighting estimating ATE and B that estimating ATT. AF atrial fibrillation, ATE average treatment effect in the whole population, AT T average treatment effect among the treated population, DM diabetes mellitus, NOAC non‑ vitamin K antagonist oral anticoagulant, HF heart failure Table 3 Risk of heart failure in older patients with atrial fibrillation and diabetes receiving each NOAC versus warfarin Fine stratification weights estimating ATE Fine stratification weights estimating ** ATT † † HR (95% CI) p‑ value HR (95% CI) p‑ value Dabigatran vs. warfarin 0.86 (0.80–0.93) < 0.001 0.81 (0.75–0.88) < 0.001 Rivaroxaban vs. warfarin 0.80 (0.74–0.86) < 0.001 0.77 (0.71–0.83) < 0.001 Apixaban vs. warfarin 0.78 (0.68–0.90) < 0.001 0.72 (0.62–0.83) < 0.001 Edoxaban vs. warfarin 0.72 (0.60–0.86) < 0.001 0.66 (0.54–0.81) < 0.001 ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, CI confidence interval, HR hazard ratio Propensity score-based fine stratification weighting which estimated the average treatment effect in the whole population ** Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population The HR is calculated using patients taking warfarin as the reference group 95% CI 0.40–0.56, p < 0.001) (Table  5). In the analysis Additional file  1: Table S2. Overall, all the sensitivity anal- that excluded patients with CKD, a similar result of a yses generated comparable results as our primary analy- lower HF risk in NOAC users was observed (HR = 0.79, ses, further supporting the robustness of our findings. 95% CI 0.72–0.87, p < 0.001) (Table  5). Additionally, the analysis including shared frailty to address the potential Discussion cluster random effect of different physicians also dem - This nationwide retrospective cohort study demonstrated onstrated a similar result (HR = 0.80, 95% CI 0.74–0.86, that elderly adults with AF and DM taking NOACs had an p < 0.001). The above sensitivity analyses for ATT esti - approximately 20% lower risk of incident HF than those mation demonstrated consistent results. In the analysis taking warfarin. The association between NOAC use and applying propensity  score matching or using multivari- decreased HF risk was consistent, regardless of age, sex, able regression models to adjust for covariates without hospital-level subgroups, or the estimations for ATE or propensity score methods, NOAC users still had a lower ATT. The findings were further supported by several sen - HF risk than warfarin users (Additional file  1: Table S3). sitivity analyses. Notably, the lower risk of HF associated The baseline patient characteristics in the analysis apply - with NOAC use versus warfarin use was more remark- ing propensity  score matching are shown in Additional able in patients taking oral anticoagulants with a high file  1: Table S4; the patient characteristics in the analysis using multivariable regression models only are shown in Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 8 of 11 Table 4 Risk of heart failure in older patients with atrial fibrillation and diabetes receiving NOAC versus warfarin, stratified for age, sex, and hospital levels Fine stratification weights estimating ATE Fine stratification weights estimating ** ATT † † HR (95% CI) p‑ value HR (95% CI) p‑ value Age 65–74 years 0.79 (0.68–0.92) 0.003 0.75 (0.62–0.90) 0.002 ≥ 75 years 0.80 (0.71–0.90) < 0.001 0.77 (0.67–0.89) < 0.001 Sex Male 0.71 (0.63–0.81) < 0.001 0.67 (0.57–0.78) < 0.001 Female 0.86 (0.78–0.96) 0.009 0.85 (0.75–0.96) 0.007 Hospital level Medical center 0.83 (0.73–0.94) 0.003 0.81 (0.70–0.93) 0.003 Regional hospital 0.84 (0.75–0.94) 0.002 0.81 (0.71–0.93) 0.002 District hospital or clinic 0.78 (0.65–0.94) 0.009 0.76 (0.61–0.94) 0.011 ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, CI confidence interval, HR hazard ratio, NOAC non- vitamin K antagonist oral anticoagulant, ref. reference Propensity score-based fine stratification weighting which estimated the average treatment effect in the whole population ** Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population The HR is calculated using patients taking warfarin as the reference group Table 5 Risk of heart failure in older patients with atrial medication possession ratio and when applying the on- fibrillation and diabetes receiving NOAC versus warfarin in the treatment design to the analysis, implying the robust asso- sensitivity analysis applying on‑treatment design, that restricting ciation between oral anticoagulant choices and HF risk. patients with MPR ≥ 80%, that excluding patients with CKD, and Although the exact mechanisms of lower HF risk in that considering cluster effects of different physicians NOAC users could not be determined in our study, sev- Fine stratification Fine stratification eral hypotheses could help explain our findings. Previous weights estimating ATE weights estimating preclinical evidence has suggested that both factor Xa ** ATT and thrombin have activities beyond coagulation, includ- † † HR (95% CI) p‑ value HR (95% CI) p‑ value ing involvement in inflammation, atherosclerotic plaque progression, atherothrombosis, vascular remodeling, Applying on‑treatment design and tissue fibrosis [18–20]. Among NOACs, rivaroxa - NOAC vs 0.67 (0.60–0.75) < 0.001 0.64 (0.57–0.72) < 0.001 warfarin ban, apixaban, and edoxaban are factor Xa inhibitors, and Restricting on patients with MPR ≥ 80% dabigatran is a direct thrombin inhibitor; the inhibition NOAC vs 0.47 (0.40–0.56) < 0.001 0.45 (0.38–0.55) < 0.001 of factor Xa or thrombin theoretically not only affects warfarin the function of coagulation but also the aforementioned Excluding patients with CKD activities. Recent preclinical and clinical studies have fur NOAC vs 0.79 (0.72–0.87) < 0.001 0.76 (0.69–0.85) < 0.001 ther supported that NOACs have potential anti-inflam - warfarin matory effects, reduce atherosclerosis, help prevent Considering cluster effects of different physicians ischemic heart disease, and suppress the progression of NOAC vs 0.80 (0.74–0.86) < 0.001 0.77 (0.70–0.84) < 0.001 cardiac fibrosis and ischemic cardiomyopathy [18–20, warfarin 37, 38], all of which may restrain the pathophysiology ATE average treatment effect in the whole population, ATT average treatment of cardiac dysfunction and HF, further decreasing the effect among the treated population, CI confidence interval, CKD chronic kidney disease, HR hazard ratio, MPR medication possession ratio, NOAC non-vitamin K risk of developing HF. In addition, previous studies have antagonist oral anticoagulant, ref. reference indicated that poor diabetes control increases the risk of Propensity score-based fine stratification weighting which estimated the developing HF [5, 39, 40]. Hyperglycemia, insulin resist average treatment effect in the whole population ** ance, and hyperinsulinemia could trigger a cascade of Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population deleterious effects, such as inflammation, dyslipidemia, The HR is calculated using patients taking warfarin as the reference group endothelial dysfunction, activation of the renin–angio- We included shared frailty, estimating the cluster random effect of different tensin–aldosterone system, autonomic dysfunction, and physicians, into the regression model to consider the potential variation from each different physician who initiated the NOAC/warfarin prescription Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 9 of 11 cardiac fibrosis, which further cause both ischemic car - medical records to confirm diagnostic accuracy due to diomyopathy and diabetic cardiomyopathy, predispos- the patient anonymity policy in the NHIRD; therefore, ing HF development [5]. Previous studies have found a potential misclassification errors may exist in the claims- beneficial role of vitamin K in improving insulin sensi - based data. However, misclassifications among patients tivity and glucose tolerance and reducing insulin resist- taking NOACs and those taking warfarin are non-differ - ance through several mechanisms [41–43]. In recent ential, thereby pushing the estimates towards the null [46, real-world studies, better blood glucose and diabetes 47]. Since we already observed a significant difference in control were suggested in patients taking NOACs than the HF risk between NOACs and warfarin in our study, in those taking warfarin due to the presence or absence the true effect sizes may be larger than we observed. of their mechanisms of antagonizing vitamin K [15–17]. Third, some patients could alter the types of oral antico - It is therefore plausible to support that one of the expla- agulants used during follow-up; hence, our main analysis nations for NOACs being associated with lower HF risk with an as-started design (emulating intention-to-treat than warfarin may be via their beneficial effects on glyce - analysis) may underestimate the true effect sizes for dif - mic and diabetes control. ferences in HF risk between NOAC and warfarin groups. Some existing studies have evaluated the efficacy and In the sensitivity analysis with an on-treatment design safety of NOACs versus those of warfarin for stroke pre- (analog of per-protocol) and that limited to patients vention in AF patients already coexisting with HF [44, with a high medication possession ratio of index antico- 45]. However, to our knowledge, evidence regarding the agulant treatment, we further obtained larger effect sizes risk of incident HF in those treated with NOACs versus with more significant results. Such results implied that those treated with warfarin is still lacking. Our study our findings of lower HF risk in NOAC users might be focused on elderly AF patients with DM, a vulnerable genuine and merits further confirmation in future stud - population prone to HF, and demonstrated that NOACs ies. Fourth, our study focused on a vulnerable population, were associated with a decreased risk of incident HF namely elderly patients with AF and DM; patient baseline compared with warfarin. Such findings have important characteristics revealed a significant comorbidity status clinical impacts because HF coexisting with AF and DM in our study population. However, it remains unclear could increase the risk of stroke, for which oral antico- whether the observed lower HF risk among NOAC users agulants are mainly prescribed for prevention, and sub- can be generalized to younger or healthier patients; more stantially deteriorate patient prognosis and quality of life research is required to answer this question. [5, 10, 11]. Our results suggested that NOACs are the preferred oral anticoagulant treatment among elderly AF Conclusions patients with DM when considering the prevention of HF In this nationwide retrospective cohort study, elderly development in this vulnerable population. patients with AF and DM taking NOACs had a lower risk The main strengths of our study were the use of a real- of incident HF than those taking warfarin. Our findings world nationwide database representing Taiwan’s entire suggest that NOACs may be the preferred oral antico- population, the target trial emulation design strengthen- agulant treatment to reduce the risk of HF in elderly AF ing causal inference using observational data, the novel patients with DM. Future research is warranted to elu- findings demonstrating the different risks of HF between cidate causation and investigate the underlying mecha- different oral anticoagulant users, and the study robust - nisms of our findings. ness supported by various sensitivity analyses. However, some limitations should be acknowledged. First, we Abbreviations could not gather data on lifestyle, smoking and drink- HF Heart failure ing history, and detailed laboratory examination results AF Atrial fibrillation DM Diabetes mellitus (e.g., blood glucose and renal function). Additionally, NOAC Non‑ vitamin K antagonist oral anticoagulant the indication for which the physicians had chosen war- HR Hazard ratio farin over NOAC (or vice versa) for each patient could CI Confidence interval NHIRD National Health Insurance Research Database not be obtained from the claims-based dataset. Although ESRD End‑stage renal disease we employed propensity score methods (including fine ATE A verage treatment effect in the whole population stratification weighting and matching) and multivariable ATT Average treatment effect in the treated population CKD Chronic kidney disease regressions to exclude potential confounders, there may still be some unknown or unmeasured confounders. Sec- Supplementary Information ond, we were unable to access patients’ comprehensive The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12933‑ 022‑ 01688‑1. Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 10 of 11 MD, USA. Section of Endocrinology, Diabetes, Nutrition & Weight Manage‑ Additional file 1: Figure S1. Flowchart of patient selection. Table S1. ment, Boston University School of Medicine, Boston, MA, USA. Specification and emulation of a target trial evaluating the effect of NOACs versus warfarin on the risk of incident heart failure using real‑ world Received: 1 September 2022 Accepted: 5 November 2022 data from Taiwan’s NHIRD. Table S2. Baseline characteristics of elderly patients with atrial fibrillation and diabetes receiving NOAC or warfarin in the original population, without weighting or matching. Table S3. Risk of heart failure in elderly patients with atrial fibrillation and diabetes receiv‑ ing NOAC versus warfarin in the sensitivity analysis applying propensity References score matching and that applying multivariable regression models with‑ 1. Kirkman MS, Briscoe VJ, Clark N, Florez H, Haas LB, Halter JB, Huang ES, out propensity score methods. Table S4. Baseline characteristics of elderly Korytkowski MT, Munshi MN, Odegard PS, et al. Diabetes in older adults. patients with atrial fibrillation and diabetes receiving NOAC or warfarin in Diabetes Care. 2012;35(12):2650–64. the population after propensity score matching. 2. Chugh SS, Havmoeller R, Narayanan K, Singh D, Rienstra M, Benjamin EJ, Gillum RF, Kim YH, McAnulty JH Jr, Zheng ZJ, et al. Worldwide epidemiol‑ ogy of atrial fibrillation: a Global Burden of Disease 2010 Study. Circula‑ Acknowledgements tion. 2014;129(8):837–47. The authors thank the Health and Welfare Data Science Center, Ministry of 3. Savarese G, Lund LH. Global public health burden of heart failure. Card Health and Welfare, Taiwan, for approving our access to the database, and the Fail Rev. 2017;3(1):7–11. Health and Welfare Data Science Center of Tzu Chi University for facilitating 4. Butrous H, Hummel SL. Heart failure in older adults. Can J Cardiol. data extraction. The authors thank Editage for English language editing. 2016;32(9):1140–7. 5. Dunlay SM, Givertz MM, Aguilar D, Allen LA, Chan M, Desai AS, Deswal Author contributions A, Dickson VV, Kosiborod MN, Lekavich CL, et al. Type 2 diabetes mellitus Concept and design: SL, PL, HH and CL. Acquisition, analysis, or interpretation and heart failure: a scientific statement from the american heart associa‑ of data: all authors. Drafting of the manuscript: SL and HH. Critical revision of tion and the heart failure society of America: this statement does not the manuscript for important intellectual content: all authors. Statistical analy‑ represent an update of the 2017 ACC/AHA/HFSA heart failure guideline sis: PL, EL, and HH. Administrative, technical, or material support: Y T, JY, and CL. update. Circulation. 2019;140(7):e294–324. Supervision: JY and CL. All authors read and approved the final manuscript. 6. Nesti L, Pugliese NR, Sciuto P, Natali A. Type 2 diabetes and reduced exer‑ cise tolerance: a review of the literature through an integrated physiology Funding approach. Cardiovasc Diabetol. 2020;19(1):134. This work was supported by a grant from the Hualien Tzu Chi Hospital 7. Nesti L, Pugliese NR, Sciuto P, De Biase N, Mazzola M, Fabiani I, Trico D, ( TCRD108‑21). The funder had no role in study design, data collection, data Masi S, Natali A. Mechanisms of reduced peak oxygen consumption analysis, data interpretation, writing of the report, decision to submit for in subjects with uncomplicated type 2 diabetes. Cardiovasc Diabetol. publication, or approval of the manuscript for publication. 2021;20(1):124. 8. Sugumar H, Nanayakkara S, Prabhu S, Voskoboinik A, Kaye DM, Ling LH, Availability of data and materials Kistler PM. Pathophysiology of atrial fibrillation and heart failure: danger ‑ The dataset used in this study is managed by the Taiwan Ministry of Health ous interactions. Cardiol Clin. 2019;37(2):131–8. and Welfare and thus cannot be made available publicly. Researchers inter‑ 9. Anter E, Jessup M, Callans DJ. Atrial fibrillation and heart failure: treatment ested in accessing this dataset can submit a formal application to the Ministry considerations for a dual epidemic. Circulation. 2009;119(18):2516–25. of Health and Welfare to request access ( Taiwan Ministry of Health and 10. Kotecha D, Piccini JP. Atrial fibrillation in heart failure: what should we do? Welfare, No. 488, Sect. 6, Zhongxiao E Rd, Nangang District, Taipei 115, Taiwan; Eur Heart J. 2015;36(46):3250–7. website: https:// dep. mohw. gov. tw/ DOS/ cp‑ 2516‑ 59203‑ 113. html). 11. Adelborg K, Szépligeti S, Sundbøll J, Horváth‑Puhó E, Henderson VW, Ord‑ ing A, Pedersen L, Sørensen HT. Risk of stroke in patients with heart fail‑ ure: a population‑based 30‑ year cohort study. Stroke. 2017;48(5):1161–8. Declarations 12. Eckman MH, Singer DE, Rosand J, Greenberg SM. Moving the tipping point: the decision to anticoagulate patients with atrial fibrillation. Circ Ethics approval and consent to participate Cardiovasc Qual Outcomes. 2011;4(1):14–21. Our study was approved by the Research Ethics Committee of Hualien Tzu Chi 13. Chao TF, Lip GYH, Lin YJ, Chang SL, Lo LW, Hu YF, Tuan TC, Liao JN, Chung Hospital (REC No: IRB107‑152‑ C); the requirement for informed consent was FP, Chen TJ, et al. Age threshold for the use of non‑ vitamin K antagonist waived due to the retrospective use of anonymized data. oral anticoagulants for stroke prevention in patients with atrial fibrillation: insights into the optimal assessment of age and incident comorbidities. Consent for publication Eur Heart J. 2019;40(19):1504–14. Not applicable. 14. Hindricks G, Potpara T, Dagres N, Arbelo E, Bax JJ, Blomström‑Lundqvist C, Boriani G, Castella M, Dan GA, Dilaveris PE, et al. 2020 ESC guidelines Competing interests for the diagnosis and management of atrial fibrillation developed in The authors declare that no competing interests exist. collaboration with the European Association for Cardio‑ Thoracic Surgery (EACTS): the task force for the diagnosis and management of atrial fibril‑ Author details lation of the European Society of Cardiology (ESC) Developed with the Department of Physical Medicine and Rehabilitation, Hualien Tzu Chi special contribution of the European Heart Rhythm Association (EHRA) of Hospital, Buddhist Tzu Chi Medical Foundation, Hualien, Taiwan. School the ESC. Eur Heart J. 2021;42(5):373–498. of Medicine, Tzu Chi University, Hualien, Taiwan. Center for Aging and Health, 15. Huang HK, Liu PP, Lin SM, Hsu JY, Peng CC, Munir KM, Wu TY, Yeh JI, Loh Hualien Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, No. 707, Sec. CH, Tu YK. Risk of developing diabetes in patients with atrial fibrillation 3, Chung Yang Rd., Hualien 97002, Taiwan. Institute of Medical Sciences, Tzu taking non‑ vitamin K antagonist oral anticoagulants or warfarin: a nation‑ Chi University, Hualien, Taiwan. Institute of Epidemiology and Preventive wide cohort study. Diabetes Obes Metab. 2021;23(2):499–507. Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan. 16. Cheung CL, Sing CW, Lau WCY, Li GHY, Lip GYH, Tan KCB, Cheung BMY, Department of Dentistry, National Taiwan University Hospital and School Chan EWY, Wong ICK. Treatment with direct oral anticoagulants or war‑ of Dentistry, National Taiwan University, Taipei, Taiwan. School of Pharmacy, farin and the risk for incident diabetes among patients with atrial fibrilla‑ Institute of Clinical Pharmacy and Pharmaceutical Sciences, College of Medi‑ tion: a population‑based cohort study. Cardiovasc Diabetol. 2021;20(1):71. cine, National Cheng Kung University, Tainan, Taiwan. Department of Family 17. Huang HK, Liu PP, Lin SM, Hsu JY, Yeh JI, Lai EC, Peng CC, Munir KM, Loh Medicine, Hualien Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, No. CH, Tu YK. Diabetes‑related complications and mortality in patients with 707, Sec. 3, Chung Yang Rd., Hualien 97002, Taiwan. Division of Endocrinology, Diabetes and Nutrition, University of Maryland School of Medicine, Baltimore, Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 11 of 11 atrial fibrillation receiving different oral anticoagulants : a nationwide 42. Manna P, Kalita J. Beneficial role of vitamin K supplementation on insulin analysis. Ann Intern Med. 2022;175(4):490–8. sensitivity, glucose metabolism, and the reduced risk of type 2 diabetes: a 18. Esmon CT. Targeting factor Xa and thrombin: impact on coagulation and review. Nutrition. 2016;32(7–8):732–9. beyond. Thromb Haemost. 2014;111(4):625–33. 43. Karamzad N, Maleki V, Carson‑ Chahhoud K, Azizi S, Sahebkar A, Gargari 19. Tsujino Y, Sakamoto T, Kinoshita K, Nakatani Y, Yamaguchi Y, Kataoka N, BP. A systematic review on the mechanisms of vitamin K effects on the Nishida K, Kinugawa K. Edoxaban suppresses the progression of atrial complications of diabetes and pre‑ diabetes. BioFactors. 2020;46(1):21–37. fibrosis and atrial fibrillation in a canine congestive heart failure model. 44. Zhao L, Wang WYS, Yang X. Anticoagulation in atrial fibrillation with heart Heart Vessels. 2019;34(8):1381–8. failure. Heart Fail Rev. 2018;23(4):563–71. 20. Liu J, Nishida M, Inui H, Chang J, Zhu Y, Kanno K, Matsuda H, Sairyo M, 45. Brown LAE, Boos CJ. Atrial fibrillation and heart failure: factors influencing Okada T, Nakaoka H, et al. Rivaroxaban suppresses the progression of the choice of oral anticoagulant. Int J Cardiol. 2017;227:863–8. ischemic cardiomyopathy in a murine model of diet‑induced myocardial 46. Copeland KT, Checkoway H, McMichael AJ, Holbrook RH. Bias due infarction. J Atheroscler Thromb. 2019;26(10):915–30. to misclassification in the estimation of relative risk. Am J Epidemiol. 21. Hsieh CY, Su CC, Shao SC, Sung SF, Lin SJ, Kao Yang YH, Lai EC. Taiwan’s 1977;105(5):488–95. national health insurance research database: past and future. Clin Epide‑ 47. Höfler M. The effect of misclassification on the estimation of association: miol. 2019;11:349–58. a review. Int J Methods Psychiatr Res. 2005;14(2):92–101. 22. Hsing AW, Ioannidis JP. Nationwide population science: lessons from the Taiwan national health insurance research database. JAMA Intern Med. Publisher’s Note 2015;175(9):1527–9. Springer Nature remains neutral with regard to jurisdictional claims in pub‑ 23. Hernán MA. Methods of public health research—strengthening causal lished maps and institutional affiliations. inference from observational data. N Engl J Med. 2021;385(15):1345–8. 24. Kutcher SA, Brophy JM, Banack HR, Kaufman JS, Samuel M. Emulating a randomised controlled trial with observational data: an introduction to the target trial framework. Can J Cardiol. 2021;37(9):1365–77. 25. Chen A, Stecker E, Warden BA. Direct oral anticoagulant use: a practical guide to common clinical challenges. J Am Heart Assoc. 2020;9(13):e017559. 26. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of clas‑ sifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83. 27. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibril‑ lation using a novel risk factor‑based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–72. 28. Pamukcu B, Lip GY, Lane DA. Simplifying stroke risk stratification in atrial fibrillation patients: implications of the CHA2DS2‑ VASc risk stratification scores. Age Ageing. 2010;39(5):533–5. 29. Desai RJ, Franklin JM. Alternative approaches for confounding adjustment in observational studies using weighting based on the propensity score: a primer for practitioners. BMJ. 2019;367:l5657. 30. Heinze G, Jüni P. An overview of the objectives of and the approaches to propensity score analyses. Eur Heart J. 2011;32(14):1704–8. 31. Austin PC, Lee DS, Fine JP. Introduction to the analysis of survival data in the presence of competing risks. Circulation. 2016;133(6):601–9. 32. Austin PC. A tutorial on multilevel survival analysis: methods models and applications. Int Stat Rev. 2017;85(2):185–203. 33. Balan TA, Putter H. A tutorial on frailty models. Stat Methods Med Res. 2020;29(11):3424–54. 34. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf. 2006;15(8):565–74. 35. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399–424. 36. Austin PC. Optimal caliper widths for propensity‑score matching when estimating differences in means and differences in proportions in obser ‑ vational studies. Pharm Stat. 2011;10(2):150–61. 37. Gradolí J, Vidal V, Brady AJ, Facila L. Anticoagulation in patients with Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : ischaemic heart disease and peripheral arterial disease: clinical implica‑ tions of COMPASS study. Eur Cardiol. 2018;13(2):115–8. fast, convenient online submission 38. Ferri LA, Bassanelli G, Savonitto S. Use of direct oral anticoagulant thorough peer review by experienced researchers in your field in ischaemic heart disease: the COMPASS study. Eur Heart J Suppl. 2019;21(Suppl B):B84‑b87. rapid publication on acceptance 39. van Melle JP, Bot M, de Jonge P, de Boer RA, van Veldhuisen DJ, Whooley support for research data, including large and complex data types MA. Diabetes, glycemic control, and new‑ onset heart failure in patients • gold Open Access which fosters wider collaboration and increased citations with stable coronary artery disease: data from the heart and soul study. Diabetes Care. 2010;33(9):2084–9. maximum visibility for your research: over 100M website views per year 40. Iribarren C, Karter AJ, Go AS, Ferrara A, Liu JY, Sidney S, Selby JV. Glycemic control and heart failure among adult patients with diabetes. Circulation. At BMC, research is always in progress. 2001;103(22):2668–73. Learn more biomedcentral.com/submissions 41. Li Y, Chen JP, Duan L, Li S. Eec ff t of vitamin K2 on type 2 diabetes mellitus: a review. Diabetes Res Clin Pract. 2018;136:39–51. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Cardiovascular Diabetology Springer Journals

Risk of heart failure in elderly patients with atrial fibrillation and diabetes taking different oral anticoagulants: a nationwide cohort study

Loading next page...
 
/lp/springer-journals/risk-of-heart-failure-in-elderly-patients-with-atrial-fibrillation-and-6O5pPw4CAs

References (56)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
eISSN
1475-2840
DOI
10.1186/s12933-022-01688-1
Publisher site
See Article on Publisher Site

Abstract

Background Heart failure (HF) is a critical complication in elderly patients with atrial fibrillation (AF) and diabetes mellitus (DM). Recent preclinical studies suggested that non‑ vitamin K antagonist oral anticoagulants (NOACs) can potentially suppress the progression of cardiac fibrosis and ischemic cardiomyopathy. Whether different oral antico ‑ agulants influence the risk of HF in older adults with AF and DM is unknown. This study aimed to evaluate the risk of HF in elderly patients with AF and DM who were administered NOACs or warfarin. Methods A nationwide retrospective cohort study was conducted based on claims data from the entire Taiwan‑ ese population. Target trial emulation design was applied to strengthen causal inference using observational data. Patients aged ≥ 65 years with AF and DM on NOAC or warfarin treatment between 2012 and 2019 were included and followed up until 2020. The primary outcome was newly diagnosed HF. Propensity score‑based fine stratification weightings were used to balance patient characteristics between NOAC and warfarin groups. Hazard ratios (HRs) were estimated using Cox proportional hazard models. Results The study included a total of 24,835 individuals (19,710 NOAC and 5,125 warfarin users). Patients taking NOACs had a significantly lower risk of HF than those taking warfarin (HR = 0.80, 95% CI 0.74–0.86, p < 0.001). Sub‑ group analyses for individual NOACs suggested that dabigatran (HR = 0.86, 95% CI 0.80–0.93, p < 0.001), rivaroxaban (HR = 0.80, 95% CI 0.74–0.86, p < 0.001), apixaban (HR = 0.78, 95% CI 0.68–0.90, p < 0.001), and edoxaban (HR = 0.72, 95% CI 0.60–0.86, p < 0.001) were associated with lower risks of HF than warfarin. The findings were consistent regard‑ less of age and sex subgroups and were more prominent in those with high medication possession ratios. Several sensitivity analyses further supported the robustness of our findings. Conclusions This nationwide cohort study demonstrated that elderly patients with AF and DM taking NOACs had a lower risk of incident HF than those taking warfarin. Our findings suggested that NOACs may be the preferred Huei‑Kai Huang and Ching‑Hui Loh have contributed equally to this work *Correspondence: Huei‑Kai Huang drhkhuang@gmail.com Ching‑Hui Loh twdoc1960@gmail.com Full list of author information is available at the end of the article © The Author(s) 2023. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 2 of 11 oral anticoagulant treatment when considering the prevention of heart failure in this vulnerable population. Future research is warranted to elucidate causation and investigate the underlying mechanisms. Keywords Oral anticoagulant, Heart failure, Atrial fibrillation, Diabetes mellitus, Elderly Background However, to date, the evidence comparing the risk of In the elderly population, atrial fibrillation (AF) and HF between NOAC and warfarin use is still lacking, even diabetes mellitus (DM) are both global epidemics and though this issue is critical for improving patient progno- important public health problems [1, 2]. Due to their sis in elderly adults already with AF and DM. Therefore, high prevalence and incidence, these two chronic con- we used nationwide cohort data to investigate the risk of ditions commonly coexist. Heart failure (HF), another HF development in elderly patients with AF and DM tak- global epidemic, affects at least 26 million people world - ing NOAC versus warfarin. wide and is one of the leading causes of morbidity, hos- pitalization, and mortality in older adults, placing a huge Methods financial burden on the health care system [3, 4]. Current Data sources evidence indicates that HF is a critical complication in We conducted a nationwide retrospective cohort study patients with AF and DM. Hyperglycemia, insulin resist- using data from the National Health Insurance Research ance, and hyperinsulinemia in DM can trigger a cascade Database (NHIRD) in Taiwan. The National Health of deleterious effects contributing to development of HF Insurance program, a mandatory single-payer program and effort intolerance [5–7]. The tachycardia, irregularity, administered by Taiwan’s government, covered more loss of atrial systole, and cardiac fibrosis in patients with than 99% of the entire population in Taiwan (approxi- AF also contribute to HF development [4, 8]. Since AF, mately 23.6 million individuals) [21, 22]. The NHIRD DM, and aging are all major risk factors of HF [4, 5, 9] contains patient demographic information and medical and concomitant HF in elderly patients with AF and DM claims for all inpatient, outpatient, and emergency care could increase their risk of stroke, worsen patient prog- services in Taiwan. The diagnostic and procedure codes noses, and increase the healthcare cost burden [5, 10, 11], in NHIRD were derived using the International Classifi - the prevention of HF development in the elderly popula- cation of Diseases, Ninth Revision, Clinical Modification tion with AF and DM is crucial. (ICD-9-CM) codes before 2016 and the International Long-term oral anticoagulant treatment is an essential Classification of Diseases, Tenth Revision, Clinical Modi - medication for stroke prevention in elderly patients with fication (ICD-10-CM) codes since 2016. Information on AF and DM [12, 13]. Warfarin, a vitamin K antagonist, mortality was obtained by cross-referencing the NHIRD has been used to prevent stroke for decades. Non-vita- with the Taiwan National Register of Deaths. The NHIRD min K antagonist oral anticoagulants (NOACs) have been is maintained by the Health and Welfare Data Science approved as another choice of oral anticoagulants and Center, Ministry of Health and Welfare, Taiwan, and the have been found to offer comparable efficacy and safety anonymized data has been made available for research for stroke prevention [14]. In addition to its anticoagu- purposes by formal application. Our study was approved lation effect, recent studies have suggested that NOACs, by the Research Ethics Committee of Hualien Tzu Chi compared with warfarin, were linked to better glycemic Hospital (REC No: IRB107-152-C); the requirement for control and lower diabetes complication risks [15–17]. informed consent was waived due to the retrospective Additionally, recent preclinical studies suggested that use of anonymized data. This study was conducted in NOACs have potential anti-inflammatory effects and accordance with the World Medical Association Declara- may suppress the progression of cardiac fibrosis and tion of Helsinki. ischemic cardiomyopathy, all of which are related to the pathophysiology of HF [18–20]. Thus, it is reasonable to Study population suppose that NOACs may have a beneficial effect on pre - We conducted the observational study with target trial venting HF compared with warfarin in patients with AF emulation to strengthen causal inference [23, 24]; the and DM. As HF is also an independent and potent risk details of how we emulated a target trial are described in factor for stroke development [11], of which oral antico- Additional file  1: Table  S1. We applied similar selection agulants are mainly prescribed for prevention, choosing criteria to those of the target trial to include all adults appropriate oral anticoagulant types to decrease incident aged ≥ 65 years with diagnoses of both AF and DM who HF risks is crucial. Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 3 of 11 had been treated with oral anticoagulants between 2012 outpatient basis within the year prior to the index date. and 2019 in NHIRD. The ICD-9-CM code 427.31 and The Charlson comorbidity index was calculated to quan - ICD-10-CM codes I48.0, I48.1, I48.2, and I48.91 were tify the overall comorbidity status [26]. We also calcu- used to identify AF diagnosis; the ICD-9-CM code 250 or lated the CHA2DS2-VASc score, which is used to predict ICD-10-CM codes E08-E13 were used for DM diagnosis. stroke risk and determine whether an oral anticoagulant Both diagnoses should be made at least once in an inpa- should be used in clinical practice [27, 28]. We defined tient service or twice in outpatient services. We restricted baseline antidiabetic drugs based on the treatment pre- our study population to patients aged ≥ 65 years because scribed within 1 month prior to the index date; the num- both AF and HF developed mainly in older adults. ber of diabetes medication types were also calculated. We excluded patients without AF and DM diagnoses Other baseline medications were defined as a drug pre - at baseline. We excluded patients with end-stage renal scribed for at least 30  days within the year prior to the disease (ESRD), rheumatic heart disease, congenital index date. The duration of AF and DM were defined heart disease, or having valve replacement surgery before as the period from the date of first diagnosis of AF or the index date because those patients are more likely to DM to the date of initiating oral anticoagulants (index receive warfarin over NOACs [25], and their exclusion date). The index year, monthly income (derived from helped minimize a potential confounding-by-indication income-related insurance premiums), physician’s medical bias. To apply the new-user design, those with a prescrip- specialty, and the hospital level of oral anticoagulant ini- tion of any oral anticoagulants in 2011 were excluded, tiation were also retrieved as covariates [17]. enhancing the likelihood of identifying new oral antico- agulant users since 2012 when NOACs were introduced Propensity score‑based fine stratification weighting in Taiwan’s National Health Insurance program. We We calculated the propensity score for each patient to excluded those with index dates in 2020, ensuring at least estimate the probability of initiating NOACs using mul- 1-year follow-up for each patient. Finally, we excluded tivariable logistic regression models, including all covari- patients with any prior HF diagnoses before the index ates shown in Table 1. We used fine stratification weights date (Additional file 1: Figure S1). based on propensity scores to create more exchangeable groups with balanced characteristics for comparisons. Exposures, outcomes, and follow‑up Two fine stratification weighting methods were applied to To emulate a target trial with intention-to-treat analysis, cover both targets of inference: estimation of the average we used an as-started design that divided patients into treatment effect in the whole population (ATE) and esti - NOAC and warfarin groups according to their first oral mation of the average treatment effect among the treated anticoagulant use regardless of subsequent prescriptions population (ATT) [29]. The individuals were stratified [24]. The index date (time zero of follow-up) was defined into 50 strata by the propensity score distribution; how as the date of initiation of oral anticoagulant treatment, the weights were calculated in each stratum is described and follow-up began since then. elsewhere [29]. The propensity score-based fine stratifica - The primary outcome was the incident HF diagnosed tion weighting was conducted individually for each com- in an inpatient service or at least three times in an out- parison set, including that of overall analyses, subgroup patient service (ICD-9-CM code: 428; ICD-10-CM code: analyses, stratified analyses, or sensitivity analyses. I50). The date of the first HF diagnosis was assigned as the date of event occurrence. We followed up with each Statistical analyses patient from their index date until an occurrence of the The difference in baseline characteristics was determined outcome event, death, or December 31, 2020 (the last by standardized difference, with a value of < 0.1 consid- date in our database), whichever came first. ered negligible. The standardized difference is preferred In our main analyses, we compared HF risk between to significance testing of covariates between study groups overall NOACs versus warfarin. We further per- because it is not confounded by sample sizes or the statis- formed subgroup analyses that subclassified NOACs tical power [30]. We estimated the cumulative incidences into four subgroups (dabigatran, rivaroxaban, apixa- and cause-specific hazard ratios (HRs) of HF using cause- ban, and edoxaban) and compared each with warfarin. specific Cox proportional hazard models with death We also performed analyses stratified by age (65–74 treated as a censoring event [31]. To address the poten- and ≥ 75 years), sex, and hospital levels. tial cluster effect and variation from each different hos - pital or clinic (where oral anticoagulant treatment was Covariates and confounders initiated), we included shared frailty, estimating the clus- Pre-existing comorbidity was defined as a condition diag - ter random effect of hospital/clinic, into the regression nosed at least once on an inpatient basis or twice on an model [32, 33]. A two-tailed probability (p) value < 0.05 Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 4 of 11 Table 1 Baseline characteristics of older patients with atrial fibrillation and diabetes receiving NOAC or warfarin after propensity score‑based fine stratification weighting * ** Population with fine stratification weights (ATE) Population with fine stratification weights (ATT) † † NOAC (N = 19,591) Warfarin (N = 5,117) SMD NOAC (N = 19,591) Warfarin (N = 5,117) SMD Age (years) 76.6 ± 7.3 76.7 ± 7.6 0.013 76.8 ± 7.4 76.9 ± 7.7 0.013 Sex Male 52.3 53.0 0.014 52.3 53.6 0.026 Female 47.7 47.0 0.014 47.7 46.4 0.026 ‡§ Charlson comorbidity index 2.7 ± 2.0 2.7 ± 1.9 0.000 2.7 ± 2.0 2.6 ± 1.9 0.051 ‡# CHA2DS2‑VASc score 4.3 ± 1.5 4.2 ± 1.5 0.067 4.2 ± 1.5 4.1 ± 1.6 0.065 Comorbidities Hypertension 78.1 77.1 0.024 77.5 76.3 0.029 Coronary artery disease 30.5 30.6 0.002 30.0 30.2 0.004 COPD 12.9 13.6 0.021 12.7 13.6 0.027 Chronic kidney disease 13.7 14.3 0.017 13.2 14.1 0.026 Cirrhosis 4.0 4.5 0.025 3.5 4.2 0.036 Hyperlipidemia 39.8 39.3 0.010 40.3 39.8 0.010 Stroke 32.2 29.9 0.050 32.0 28.6 0.074 Rheumatoid arthritis 0.8 0.7 0.012 0.8 0.7 0.012 Gout 9.6 9.5 0.003 9.2 9.1 0.004 Dementia 7.3 6.8 0.020 7.6 6.9 0.027 Malignancy 9.4 9.3 0.003 9.5 9.4 0.003 Medication use Statins 40.2 39.7 0.010 41.3 40.8 0.010 ACEI or ARB 61.7 61.8 0.002 61.9 62.0 0.002 β blockers 44.5 45.8 0.026 44.3 45.9 0.032 Calcium channel blockers 46.9 47.4 0.010 45.9 46.6 0.014 Diuretics 22.3 23.7 0.033 21.0 23.0 0.048 NSAID 33.4 32.6 0.017 33.4 32.3 0.023 Corticosteroids 5.8 5.7 0.004 5.7 5.6 0.004 Antipsychotics 5.3 5.2 0.005 5.1 5.1 0.000 Proton pump inhibitors 9.0 8.2 0.029 9.1 8.0 0.039 Baseline diabetes medications Metformin 48.6 47.4 0.024 49.3 47.5 0.036 Sulfonylurea 31.4 30.4 0.022 30.1 28.9 0.026 Meglitinide 5.7 5.5 0.009 5.2 5.0 0.009 AGI 7.7 8.7 0.037 7.3 8.6 0.048 TZD 4.8 4.6 0.010 4.8 4.4 0.019 DPP‑4i 26.7 27.7 0.023 27.6 29.1 0.033 SGLT‑2i 1.7 2.0 0.022 2.1 2.4 0.020 GLP‑1 RA 0.2 0.2 0.000 0.2 0.2 0.000 Insulin 10.2 10.0 0.007 9.9 9.8 0.003 Numbers of diabetes medications Without medications 33.3 33.8 0.011 33.7 34.5 0.017 1 type 24.4 23.3 0.026 24.2 22.8 0.033 2 types 21.4 22.4 0.024 21.3 22.0 0.017 ≥ 3 types 20.9 20.5 0.010 20.8 20.7 0.003 Duration of diabetes < 2 years 21.8 20.6 0.029 18.0 15.8 0.059 ≥ 2 years 78.2 79.4 0.029 82.0 84.2 0.059 Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 5 of 11 Table 1 (continued) * ** Population with fine stratification weights (ATE) Population with fine stratification weights (ATT) † † NOAC (N = 19,591) Warfarin (N = 5,117) SMD NOAC (N = 19,591) Warfarin (N = 5,117) SMD Duration of AF < 2 years 71.2 68.5 0.059 69.9 66.0 0.084 ≥ 2 years 28.8 31.5 0.059 30.1 34.0 0.084 Index year 2012–2013 13.3 13.5 0.006 6.5 6.6 0.004 2014–2015 23.6 24.0 0.009 21.2 21.7 0.012 2016–2017 30.4 28.9 0.033 33.9 32.1 0.038 2018–2019 32.7 33.6 0.019 38.4 39.5 0.023 Income level (NTD) Financially dependent 29.5 29.2 0.007 29.5 29.1 0.009 15,840–29,999 47.8 47.9 0.002 47.2 47.6 0.008 30,000–44,999 11.2 11.1 0.003 11.2 11.0 0.006 ≥ 45,000 11.6 11.7 0.003 12.2 12.2 0.000 Hospital level of OAC initiation Medical center 36.2 34.1 0.044 38.2 35.2 0.062 Regional hospital 44.7 47.0 0.046 44.4 47.3 0.058 District hospital or clinic 19.1 19.0 0.003 17.5 17.5 0.000 Physician specialty Cardiologist 63.7 66.2 0.052 65.2 68.5 0.070 Neurologist 19.5 16.7 0.073 20.3 16.4 0.101 Others 16.8 17.2 0.011 14.5 15.0 0.014 Data are presented as percentages unless otherwise noted ACEI angiotensin-converting enzyme inhibitors, AF atrial fibrillation, AGI alpha-glucosidase inhibitors, ARB angiotensin II receptor blockers, ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, COPD chronic obstructive pulmonary disease, DPP-4i dipeptidyl peptidase-4 inhibitors, GLP-1 RA glucagon-like peptide-1 receptor agonists, IPTW inverse probability of treatment weighting, NOAC non-vitamin K antagonist oral anticoagulant, NSAID nonsteroidal anti-inflammatory drugs, NTD New Taiwan Dollar, OAC oral anticoagulant, PSM propensity score matching, SGLT-2i sodium-glucose cotransporter-2 inhibitors, SMD standardized mean difference, TZD thiazolidinedione The pseudo-population constructed by propensity score-based fine stratification weighting to estimate the average treatment effect in the whole population ** The pseudo-population constructed by propensity score-based fine stratification weighting to estimate the average treatment effect among the treated population A standardized mean difference of < 0.1 indicates a negligible difference Presented as mean ± standard deviation Calculated without scores for age Congestive heart failure, hypertension, age ≥ 75 years, diabetes mellitus, stroke or transient ischemic attack, vascular disease, age 65–74 years, sex category (CHA2DS2-VASc) score The period from the date of first diagnosis of diabetes or AF to the index date was considered statistically significant. We managed data discontinued. Discontinuation was defined as patients and performed statistical analyses using SAS software, without a refilled prescription of the index oral anti - version 9.4 (SAS Institute, Inc., Cary, NC, USA) and coagulant 90  days after the last prescription. Second, STATA, version 15 (Stata Corporation LLC, College Sta- we restricted our analysis to those taking the index oral tion, TX, USA). anticoagulant with a high medication possession ratio, defined as ≥ 80%. The medication possession ratio was Sensitivity analyses calculated by dividing the number of days with prescrip- Various sensitivity analyses were conducted to determine tion of oral anticoagulants by the days of the follow-up the robustness of our study results. First, to consider period [34]. Third, we excluded patients with any diag - the treatment adherence during follow-up, we applied noses of chronic kidney disease (CKD) before the index the on-treatment design (analog of per-protocol design date since we could not obtain individuals’ renal function in clinical trials) in which the follow-up would be cen- data, which may influence the choice of NOACs. Fourth, sored when the oral anticoagulant type was switched or to determine whether potential variations between Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 6 of 11 Risk of incident HF physicians who initiated the NOAC or warfarin prescrip- In the analysis with propensity score-based fine strati - tion influenced our results, we performed a sensitivity fication weighting for ATE estimation, NOAC use was analysis that included shared frailty, estimating the clus- significantly associated with a lower risk of develop - ter random effect of different physicians, into the regres - ing HF than warfarin use (HR = 0.80, 95% confidence sion model. Additionally, we performed two sensitivity interval CI 0.74–0.86, p < 0.001). In the ATT estimation analyses with different statistical designs. One sensitivity analysis, a similar result of lower HF risk in NOAC users analysis applied propensity  score matching (rather than was observed (HR = 0.77, 95% CI 0.70–0.84, p < 0.001) fine stratification) was performed to balance patient (Table  2). Figure  1 illustrates the curves for cumula- characteristics between groups; the matching was based tive HF incidences in patients taking NOACs and those on the nearest-neighbor matching algorithm without taking warfarin; a lower cumulative HF incidence was replacement, with a caliper width equal to 0.2 stand- observed in NOAC users. The curves for estimating ATE ard deviation of the logit of the propensity score [35, and ATT are shown in Fig. 1A and B, respectively. 36]. Another sensitivity analysis estimated the adjusted In the ATE estimation analyses for each NOAC, dabi- HRs by multivariable Cox regression models based on gatran (HR = 0.86, 95% CI 0.80–0.93, p < 0.001), rivaroxa- the original cohort without applying propensity score ban (HR = 0.80, 95% CI 0.74–0.86, p < 0.001), apixaban methods. (HR = 0.78, 95% CI 0.68–0.90, p < 0.001), and edoxaban (HR = 0.72, 95% CI 0.60–0.86, p < 0.001) were all associ- Results ated with a lower HF risk when compared with warfarin Patient characteristics (Table  3). The ATT estimation analyses demonstrated We initially included 24,835 patients (19,710 NOAC similar findings. and 5,125 warfarin users) after applying the inclusion In the analyses stratified by age, sex, and hospital levels, and exclusion criteria; the patient characteristics in the consistent findings were observed; the significantly lower original cohort are shown in Additional file  1: Table  S2. HF risk associated with NOAC use was observed in all For further analyses, we constructed pseudo-popula- stratified groups, regardless of age, sex, or hospital levels tions containing 19,591 NOAC and 5,117 warfarin users (Table 4). after applying propensity score-based fine stratification weighting. The patient characteristics in the weighted Results of sensitivity analyses population for ATE and ATT estimation are presented With the application of an on-treatment design, NOAC in Table  1. The mean age was approximately 76.6  years, users still demonstrated a lower HF risk than warfarin and female patients accounted for 47% of all participants. users (HR = 0.67, 95% CI 0.60–0.75, p < 0.001) in the ATE The mean follow-up duration was 3.0 years. Patient char - estimation analysis (Table  5). In the analysis restricted acteristics were balanced appropriately between groups to only patients with a high medication possession after fine stratification weighting, with standardized dif - ratio (≥ 80%), a more remarkable association between ferences < 0.1. The flowchart of patient selection is pre - NOAC use and lower HF risk was observed (HR = 0.47, sented in Additional file 1: Figure S1. Table 2 Risk of heart failure in older patients with atrial fibrillation and diabetes receiving NOAC versus warfarin Event no Person‑ yearsIncidence rate HR (95% CI) p‑ value Fine stratification weights estimating ATE NOAC (N = 19,591) 4512 59,298 76.1 0.80 (0.74–0.86) < 0.001 Warfarin (N = 5,117) 1404 14,677 95.6 1 (ref.) ** Fine stratification weights estimating ATT NOAC (N = 19,591) 4158 55,059 75.5 0.77 (0.70–0.84) < 0.001 Warfarin (N = 5,117) 1343 13,576 98.9 1 (ref.) ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, CI confidence interval, HR hazard ratio, NOAC non- vitamin K antagonist oral anticoagulant, ref. reference Propensity score-based fine stratification weighting which estimated the average treatment effect in the whole population ** Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population Incidence rate, per 1000 person-years Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 7 of 11 Fig. 1 The cumulative incidence curves of HF in elderly patients with AF and DM taking NOACs and those taking warfarin. The curves were estimated according to the pseudo‑populations constructed by A propensity score ‑based fine stratification weighting estimating ATE and B that estimating ATT. AF atrial fibrillation, ATE average treatment effect in the whole population, AT T average treatment effect among the treated population, DM diabetes mellitus, NOAC non‑ vitamin K antagonist oral anticoagulant, HF heart failure Table 3 Risk of heart failure in older patients with atrial fibrillation and diabetes receiving each NOAC versus warfarin Fine stratification weights estimating ATE Fine stratification weights estimating ** ATT † † HR (95% CI) p‑ value HR (95% CI) p‑ value Dabigatran vs. warfarin 0.86 (0.80–0.93) < 0.001 0.81 (0.75–0.88) < 0.001 Rivaroxaban vs. warfarin 0.80 (0.74–0.86) < 0.001 0.77 (0.71–0.83) < 0.001 Apixaban vs. warfarin 0.78 (0.68–0.90) < 0.001 0.72 (0.62–0.83) < 0.001 Edoxaban vs. warfarin 0.72 (0.60–0.86) < 0.001 0.66 (0.54–0.81) < 0.001 ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, CI confidence interval, HR hazard ratio Propensity score-based fine stratification weighting which estimated the average treatment effect in the whole population ** Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population The HR is calculated using patients taking warfarin as the reference group 95% CI 0.40–0.56, p < 0.001) (Table  5). In the analysis Additional file  1: Table S2. Overall, all the sensitivity anal- that excluded patients with CKD, a similar result of a yses generated comparable results as our primary analy- lower HF risk in NOAC users was observed (HR = 0.79, ses, further supporting the robustness of our findings. 95% CI 0.72–0.87, p < 0.001) (Table  5). Additionally, the analysis including shared frailty to address the potential Discussion cluster random effect of different physicians also dem - This nationwide retrospective cohort study demonstrated onstrated a similar result (HR = 0.80, 95% CI 0.74–0.86, that elderly adults with AF and DM taking NOACs had an p < 0.001). The above sensitivity analyses for ATT esti - approximately 20% lower risk of incident HF than those mation demonstrated consistent results. In the analysis taking warfarin. The association between NOAC use and applying propensity  score matching or using multivari- decreased HF risk was consistent, regardless of age, sex, able regression models to adjust for covariates without hospital-level subgroups, or the estimations for ATE or propensity score methods, NOAC users still had a lower ATT. The findings were further supported by several sen - HF risk than warfarin users (Additional file  1: Table S3). sitivity analyses. Notably, the lower risk of HF associated The baseline patient characteristics in the analysis apply - with NOAC use versus warfarin use was more remark- ing propensity  score matching are shown in Additional able in patients taking oral anticoagulants with a high file  1: Table S4; the patient characteristics in the analysis using multivariable regression models only are shown in Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 8 of 11 Table 4 Risk of heart failure in older patients with atrial fibrillation and diabetes receiving NOAC versus warfarin, stratified for age, sex, and hospital levels Fine stratification weights estimating ATE Fine stratification weights estimating ** ATT † † HR (95% CI) p‑ value HR (95% CI) p‑ value Age 65–74 years 0.79 (0.68–0.92) 0.003 0.75 (0.62–0.90) 0.002 ≥ 75 years 0.80 (0.71–0.90) < 0.001 0.77 (0.67–0.89) < 0.001 Sex Male 0.71 (0.63–0.81) < 0.001 0.67 (0.57–0.78) < 0.001 Female 0.86 (0.78–0.96) 0.009 0.85 (0.75–0.96) 0.007 Hospital level Medical center 0.83 (0.73–0.94) 0.003 0.81 (0.70–0.93) 0.003 Regional hospital 0.84 (0.75–0.94) 0.002 0.81 (0.71–0.93) 0.002 District hospital or clinic 0.78 (0.65–0.94) 0.009 0.76 (0.61–0.94) 0.011 ATE average treatment effect in the whole population, ATT average treatment effect among the treated population, CI confidence interval, HR hazard ratio, NOAC non- vitamin K antagonist oral anticoagulant, ref. reference Propensity score-based fine stratification weighting which estimated the average treatment effect in the whole population ** Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population The HR is calculated using patients taking warfarin as the reference group Table 5 Risk of heart failure in older patients with atrial medication possession ratio and when applying the on- fibrillation and diabetes receiving NOAC versus warfarin in the treatment design to the analysis, implying the robust asso- sensitivity analysis applying on‑treatment design, that restricting ciation between oral anticoagulant choices and HF risk. patients with MPR ≥ 80%, that excluding patients with CKD, and Although the exact mechanisms of lower HF risk in that considering cluster effects of different physicians NOAC users could not be determined in our study, sev- Fine stratification Fine stratification eral hypotheses could help explain our findings. Previous weights estimating ATE weights estimating preclinical evidence has suggested that both factor Xa ** ATT and thrombin have activities beyond coagulation, includ- † † HR (95% CI) p‑ value HR (95% CI) p‑ value ing involvement in inflammation, atherosclerotic plaque progression, atherothrombosis, vascular remodeling, Applying on‑treatment design and tissue fibrosis [18–20]. Among NOACs, rivaroxa - NOAC vs 0.67 (0.60–0.75) < 0.001 0.64 (0.57–0.72) < 0.001 warfarin ban, apixaban, and edoxaban are factor Xa inhibitors, and Restricting on patients with MPR ≥ 80% dabigatran is a direct thrombin inhibitor; the inhibition NOAC vs 0.47 (0.40–0.56) < 0.001 0.45 (0.38–0.55) < 0.001 of factor Xa or thrombin theoretically not only affects warfarin the function of coagulation but also the aforementioned Excluding patients with CKD activities. Recent preclinical and clinical studies have fur NOAC vs 0.79 (0.72–0.87) < 0.001 0.76 (0.69–0.85) < 0.001 ther supported that NOACs have potential anti-inflam - warfarin matory effects, reduce atherosclerosis, help prevent Considering cluster effects of different physicians ischemic heart disease, and suppress the progression of NOAC vs 0.80 (0.74–0.86) < 0.001 0.77 (0.70–0.84) < 0.001 cardiac fibrosis and ischemic cardiomyopathy [18–20, warfarin 37, 38], all of which may restrain the pathophysiology ATE average treatment effect in the whole population, ATT average treatment of cardiac dysfunction and HF, further decreasing the effect among the treated population, CI confidence interval, CKD chronic kidney disease, HR hazard ratio, MPR medication possession ratio, NOAC non-vitamin K risk of developing HF. In addition, previous studies have antagonist oral anticoagulant, ref. reference indicated that poor diabetes control increases the risk of Propensity score-based fine stratification weighting which estimated the developing HF [5, 39, 40]. Hyperglycemia, insulin resist average treatment effect in the whole population ** ance, and hyperinsulinemia could trigger a cascade of Propensity score-based fine stratification weighting which estimated the average treatment effect among the treated population deleterious effects, such as inflammation, dyslipidemia, The HR is calculated using patients taking warfarin as the reference group endothelial dysfunction, activation of the renin–angio- We included shared frailty, estimating the cluster random effect of different tensin–aldosterone system, autonomic dysfunction, and physicians, into the regression model to consider the potential variation from each different physician who initiated the NOAC/warfarin prescription Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 9 of 11 cardiac fibrosis, which further cause both ischemic car - medical records to confirm diagnostic accuracy due to diomyopathy and diabetic cardiomyopathy, predispos- the patient anonymity policy in the NHIRD; therefore, ing HF development [5]. Previous studies have found a potential misclassification errors may exist in the claims- beneficial role of vitamin K in improving insulin sensi - based data. However, misclassifications among patients tivity and glucose tolerance and reducing insulin resist- taking NOACs and those taking warfarin are non-differ - ance through several mechanisms [41–43]. In recent ential, thereby pushing the estimates towards the null [46, real-world studies, better blood glucose and diabetes 47]. Since we already observed a significant difference in control were suggested in patients taking NOACs than the HF risk between NOACs and warfarin in our study, in those taking warfarin due to the presence or absence the true effect sizes may be larger than we observed. of their mechanisms of antagonizing vitamin K [15–17]. Third, some patients could alter the types of oral antico - It is therefore plausible to support that one of the expla- agulants used during follow-up; hence, our main analysis nations for NOACs being associated with lower HF risk with an as-started design (emulating intention-to-treat than warfarin may be via their beneficial effects on glyce - analysis) may underestimate the true effect sizes for dif - mic and diabetes control. ferences in HF risk between NOAC and warfarin groups. Some existing studies have evaluated the efficacy and In the sensitivity analysis with an on-treatment design safety of NOACs versus those of warfarin for stroke pre- (analog of per-protocol) and that limited to patients vention in AF patients already coexisting with HF [44, with a high medication possession ratio of index antico- 45]. However, to our knowledge, evidence regarding the agulant treatment, we further obtained larger effect sizes risk of incident HF in those treated with NOACs versus with more significant results. Such results implied that those treated with warfarin is still lacking. Our study our findings of lower HF risk in NOAC users might be focused on elderly AF patients with DM, a vulnerable genuine and merits further confirmation in future stud - population prone to HF, and demonstrated that NOACs ies. Fourth, our study focused on a vulnerable population, were associated with a decreased risk of incident HF namely elderly patients with AF and DM; patient baseline compared with warfarin. Such findings have important characteristics revealed a significant comorbidity status clinical impacts because HF coexisting with AF and DM in our study population. However, it remains unclear could increase the risk of stroke, for which oral antico- whether the observed lower HF risk among NOAC users agulants are mainly prescribed for prevention, and sub- can be generalized to younger or healthier patients; more stantially deteriorate patient prognosis and quality of life research is required to answer this question. [5, 10, 11]. Our results suggested that NOACs are the preferred oral anticoagulant treatment among elderly AF Conclusions patients with DM when considering the prevention of HF In this nationwide retrospective cohort study, elderly development in this vulnerable population. patients with AF and DM taking NOACs had a lower risk The main strengths of our study were the use of a real- of incident HF than those taking warfarin. Our findings world nationwide database representing Taiwan’s entire suggest that NOACs may be the preferred oral antico- population, the target trial emulation design strengthen- agulant treatment to reduce the risk of HF in elderly AF ing causal inference using observational data, the novel patients with DM. Future research is warranted to elu- findings demonstrating the different risks of HF between cidate causation and investigate the underlying mecha- different oral anticoagulant users, and the study robust - nisms of our findings. ness supported by various sensitivity analyses. However, some limitations should be acknowledged. First, we Abbreviations could not gather data on lifestyle, smoking and drink- HF Heart failure ing history, and detailed laboratory examination results AF Atrial fibrillation DM Diabetes mellitus (e.g., blood glucose and renal function). Additionally, NOAC Non‑ vitamin K antagonist oral anticoagulant the indication for which the physicians had chosen war- HR Hazard ratio farin over NOAC (or vice versa) for each patient could CI Confidence interval NHIRD National Health Insurance Research Database not be obtained from the claims-based dataset. Although ESRD End‑stage renal disease we employed propensity score methods (including fine ATE A verage treatment effect in the whole population stratification weighting and matching) and multivariable ATT Average treatment effect in the treated population CKD Chronic kidney disease regressions to exclude potential confounders, there may still be some unknown or unmeasured confounders. Sec- Supplementary Information ond, we were unable to access patients’ comprehensive The online version contains supplementary material available at https:// doi. org/ 10. 1186/ s12933‑ 022‑ 01688‑1. Lin et al. Cardiovascular Diabetology (2023) 22:1 Page 10 of 11 MD, USA. Section of Endocrinology, Diabetes, Nutrition & Weight Manage‑ Additional file 1: Figure S1. Flowchart of patient selection. Table S1. ment, Boston University School of Medicine, Boston, MA, USA. Specification and emulation of a target trial evaluating the effect of NOACs versus warfarin on the risk of incident heart failure using real‑ world Received: 1 September 2022 Accepted: 5 November 2022 data from Taiwan’s NHIRD. Table S2. Baseline characteristics of elderly patients with atrial fibrillation and diabetes receiving NOAC or warfarin in the original population, without weighting or matching. Table S3. Risk of heart failure in elderly patients with atrial fibrillation and diabetes receiv‑ ing NOAC versus warfarin in the sensitivity analysis applying propensity References score matching and that applying multivariable regression models with‑ 1. Kirkman MS, Briscoe VJ, Clark N, Florez H, Haas LB, Halter JB, Huang ES, out propensity score methods. Table S4. Baseline characteristics of elderly Korytkowski MT, Munshi MN, Odegard PS, et al. Diabetes in older adults. patients with atrial fibrillation and diabetes receiving NOAC or warfarin in Diabetes Care. 2012;35(12):2650–64. the population after propensity score matching. 2. Chugh SS, Havmoeller R, Narayanan K, Singh D, Rienstra M, Benjamin EJ, Gillum RF, Kim YH, McAnulty JH Jr, Zheng ZJ, et al. Worldwide epidemiol‑ ogy of atrial fibrillation: a Global Burden of Disease 2010 Study. Circula‑ Acknowledgements tion. 2014;129(8):837–47. The authors thank the Health and Welfare Data Science Center, Ministry of 3. Savarese G, Lund LH. Global public health burden of heart failure. Card Health and Welfare, Taiwan, for approving our access to the database, and the Fail Rev. 2017;3(1):7–11. Health and Welfare Data Science Center of Tzu Chi University for facilitating 4. Butrous H, Hummel SL. Heart failure in older adults. Can J Cardiol. data extraction. The authors thank Editage for English language editing. 2016;32(9):1140–7. 5. Dunlay SM, Givertz MM, Aguilar D, Allen LA, Chan M, Desai AS, Deswal Author contributions A, Dickson VV, Kosiborod MN, Lekavich CL, et al. Type 2 diabetes mellitus Concept and design: SL, PL, HH and CL. Acquisition, analysis, or interpretation and heart failure: a scientific statement from the american heart associa‑ of data: all authors. Drafting of the manuscript: SL and HH. Critical revision of tion and the heart failure society of America: this statement does not the manuscript for important intellectual content: all authors. Statistical analy‑ represent an update of the 2017 ACC/AHA/HFSA heart failure guideline sis: PL, EL, and HH. Administrative, technical, or material support: Y T, JY, and CL. update. Circulation. 2019;140(7):e294–324. Supervision: JY and CL. All authors read and approved the final manuscript. 6. Nesti L, Pugliese NR, Sciuto P, Natali A. Type 2 diabetes and reduced exer‑ cise tolerance: a review of the literature through an integrated physiology Funding approach. Cardiovasc Diabetol. 2020;19(1):134. This work was supported by a grant from the Hualien Tzu Chi Hospital 7. Nesti L, Pugliese NR, Sciuto P, De Biase N, Mazzola M, Fabiani I, Trico D, ( TCRD108‑21). The funder had no role in study design, data collection, data Masi S, Natali A. Mechanisms of reduced peak oxygen consumption analysis, data interpretation, writing of the report, decision to submit for in subjects with uncomplicated type 2 diabetes. Cardiovasc Diabetol. publication, or approval of the manuscript for publication. 2021;20(1):124. 8. Sugumar H, Nanayakkara S, Prabhu S, Voskoboinik A, Kaye DM, Ling LH, Availability of data and materials Kistler PM. Pathophysiology of atrial fibrillation and heart failure: danger ‑ The dataset used in this study is managed by the Taiwan Ministry of Health ous interactions. Cardiol Clin. 2019;37(2):131–8. and Welfare and thus cannot be made available publicly. Researchers inter‑ 9. Anter E, Jessup M, Callans DJ. Atrial fibrillation and heart failure: treatment ested in accessing this dataset can submit a formal application to the Ministry considerations for a dual epidemic. Circulation. 2009;119(18):2516–25. of Health and Welfare to request access ( Taiwan Ministry of Health and 10. Kotecha D, Piccini JP. Atrial fibrillation in heart failure: what should we do? Welfare, No. 488, Sect. 6, Zhongxiao E Rd, Nangang District, Taipei 115, Taiwan; Eur Heart J. 2015;36(46):3250–7. website: https:// dep. mohw. gov. tw/ DOS/ cp‑ 2516‑ 59203‑ 113. html). 11. Adelborg K, Szépligeti S, Sundbøll J, Horváth‑Puhó E, Henderson VW, Ord‑ ing A, Pedersen L, Sørensen HT. Risk of stroke in patients with heart fail‑ ure: a population‑based 30‑ year cohort study. Stroke. 2017;48(5):1161–8. Declarations 12. Eckman MH, Singer DE, Rosand J, Greenberg SM. Moving the tipping point: the decision to anticoagulate patients with atrial fibrillation. Circ Ethics approval and consent to participate Cardiovasc Qual Outcomes. 2011;4(1):14–21. Our study was approved by the Research Ethics Committee of Hualien Tzu Chi 13. Chao TF, Lip GYH, Lin YJ, Chang SL, Lo LW, Hu YF, Tuan TC, Liao JN, Chung Hospital (REC No: IRB107‑152‑ C); the requirement for informed consent was FP, Chen TJ, et al. Age threshold for the use of non‑ vitamin K antagonist waived due to the retrospective use of anonymized data. oral anticoagulants for stroke prevention in patients with atrial fibrillation: insights into the optimal assessment of age and incident comorbidities. Consent for publication Eur Heart J. 2019;40(19):1504–14. Not applicable. 14. Hindricks G, Potpara T, Dagres N, Arbelo E, Bax JJ, Blomström‑Lundqvist C, Boriani G, Castella M, Dan GA, Dilaveris PE, et al. 2020 ESC guidelines Competing interests for the diagnosis and management of atrial fibrillation developed in The authors declare that no competing interests exist. collaboration with the European Association for Cardio‑ Thoracic Surgery (EACTS): the task force for the diagnosis and management of atrial fibril‑ Author details lation of the European Society of Cardiology (ESC) Developed with the Department of Physical Medicine and Rehabilitation, Hualien Tzu Chi special contribution of the European Heart Rhythm Association (EHRA) of Hospital, Buddhist Tzu Chi Medical Foundation, Hualien, Taiwan. School the ESC. Eur Heart J. 2021;42(5):373–498. of Medicine, Tzu Chi University, Hualien, Taiwan. Center for Aging and Health, 15. Huang HK, Liu PP, Lin SM, Hsu JY, Peng CC, Munir KM, Wu TY, Yeh JI, Loh Hualien Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, No. 707, Sec. CH, Tu YK. Risk of developing diabetes in patients with atrial fibrillation 3, Chung Yang Rd., Hualien 97002, Taiwan. Institute of Medical Sciences, Tzu taking non‑ vitamin K antagonist oral anticoagulants or warfarin: a nation‑ Chi University, Hualien, Taiwan. Institute of Epidemiology and Preventive wide cohort study. Diabetes Obes Metab. 2021;23(2):499–507. Medicine, College of Public Health, National Taiwan University, Taipei, Taiwan. 16. Cheung CL, Sing CW, Lau WCY, Li GHY, Lip GYH, Tan KCB, Cheung BMY, Department of Dentistry, National Taiwan University Hospital and School Chan EWY, Wong ICK. Treatment with direct oral anticoagulants or war‑ of Dentistry, National Taiwan University, Taipei, Taiwan. School of Pharmacy, farin and the risk for incident diabetes among patients with atrial fibrilla‑ Institute of Clinical Pharmacy and Pharmaceutical Sciences, College of Medi‑ tion: a population‑based cohort study. Cardiovasc Diabetol. 2021;20(1):71. cine, National Cheng Kung University, Tainan, Taiwan. Department of Family 17. Huang HK, Liu PP, Lin SM, Hsu JY, Yeh JI, Lai EC, Peng CC, Munir KM, Loh Medicine, Hualien Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, No. CH, Tu YK. Diabetes‑related complications and mortality in patients with 707, Sec. 3, Chung Yang Rd., Hualien 97002, Taiwan. Division of Endocrinology, Diabetes and Nutrition, University of Maryland School of Medicine, Baltimore, Lin  et al. Cardiovascular Diabetology (2023) 22:1 Page 11 of 11 atrial fibrillation receiving different oral anticoagulants : a nationwide 42. Manna P, Kalita J. Beneficial role of vitamin K supplementation on insulin analysis. Ann Intern Med. 2022;175(4):490–8. sensitivity, glucose metabolism, and the reduced risk of type 2 diabetes: a 18. Esmon CT. Targeting factor Xa and thrombin: impact on coagulation and review. Nutrition. 2016;32(7–8):732–9. beyond. Thromb Haemost. 2014;111(4):625–33. 43. Karamzad N, Maleki V, Carson‑ Chahhoud K, Azizi S, Sahebkar A, Gargari 19. Tsujino Y, Sakamoto T, Kinoshita K, Nakatani Y, Yamaguchi Y, Kataoka N, BP. A systematic review on the mechanisms of vitamin K effects on the Nishida K, Kinugawa K. Edoxaban suppresses the progression of atrial complications of diabetes and pre‑ diabetes. BioFactors. 2020;46(1):21–37. fibrosis and atrial fibrillation in a canine congestive heart failure model. 44. Zhao L, Wang WYS, Yang X. Anticoagulation in atrial fibrillation with heart Heart Vessels. 2019;34(8):1381–8. failure. Heart Fail Rev. 2018;23(4):563–71. 20. Liu J, Nishida M, Inui H, Chang J, Zhu Y, Kanno K, Matsuda H, Sairyo M, 45. Brown LAE, Boos CJ. Atrial fibrillation and heart failure: factors influencing Okada T, Nakaoka H, et al. Rivaroxaban suppresses the progression of the choice of oral anticoagulant. Int J Cardiol. 2017;227:863–8. ischemic cardiomyopathy in a murine model of diet‑induced myocardial 46. Copeland KT, Checkoway H, McMichael AJ, Holbrook RH. Bias due infarction. J Atheroscler Thromb. 2019;26(10):915–30. to misclassification in the estimation of relative risk. Am J Epidemiol. 21. Hsieh CY, Su CC, Shao SC, Sung SF, Lin SJ, Kao Yang YH, Lai EC. Taiwan’s 1977;105(5):488–95. national health insurance research database: past and future. Clin Epide‑ 47. Höfler M. The effect of misclassification on the estimation of association: miol. 2019;11:349–58. a review. Int J Methods Psychiatr Res. 2005;14(2):92–101. 22. Hsing AW, Ioannidis JP. Nationwide population science: lessons from the Taiwan national health insurance research database. JAMA Intern Med. Publisher’s Note 2015;175(9):1527–9. Springer Nature remains neutral with regard to jurisdictional claims in pub‑ 23. Hernán MA. Methods of public health research—strengthening causal lished maps and institutional affiliations. inference from observational data. N Engl J Med. 2021;385(15):1345–8. 24. Kutcher SA, Brophy JM, Banack HR, Kaufman JS, Samuel M. Emulating a randomised controlled trial with observational data: an introduction to the target trial framework. Can J Cardiol. 2021;37(9):1365–77. 25. Chen A, Stecker E, Warden BA. Direct oral anticoagulant use: a practical guide to common clinical challenges. J Am Heart Assoc. 2020;9(13):e017559. 26. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of clas‑ sifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40(5):373–83. 27. Lip GY, Nieuwlaat R, Pisters R, Lane DA, Crijns HJ. Refining clinical risk stratification for predicting stroke and thromboembolism in atrial fibril‑ lation using a novel risk factor‑based approach: the euro heart survey on atrial fibrillation. Chest. 2010;137(2):263–72. 28. Pamukcu B, Lip GY, Lane DA. Simplifying stroke risk stratification in atrial fibrillation patients: implications of the CHA2DS2‑ VASc risk stratification scores. Age Ageing. 2010;39(5):533–5. 29. Desai RJ, Franklin JM. Alternative approaches for confounding adjustment in observational studies using weighting based on the propensity score: a primer for practitioners. BMJ. 2019;367:l5657. 30. Heinze G, Jüni P. An overview of the objectives of and the approaches to propensity score analyses. Eur Heart J. 2011;32(14):1704–8. 31. Austin PC, Lee DS, Fine JP. Introduction to the analysis of survival data in the presence of competing risks. Circulation. 2016;133(6):601–9. 32. Austin PC. A tutorial on multilevel survival analysis: methods models and applications. Int Stat Rev. 2017;85(2):185–203. 33. Balan TA, Putter H. A tutorial on frailty models. Stat Methods Med Res. 2020;29(11):3424–54. 34. Andrade SE, Kahler KH, Frech F, Chan KA. Methods for evaluation of medication adherence and persistence using automated databases. Pharmacoepidemiol Drug Saf. 2006;15(8):565–74. 35. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399–424. 36. Austin PC. Optimal caliper widths for propensity‑score matching when estimating differences in means and differences in proportions in obser ‑ vational studies. Pharm Stat. 2011;10(2):150–61. 37. Gradolí J, Vidal V, Brady AJ, Facila L. Anticoagulation in patients with Re Read ady y to to submit y submit your our re researc search h ? Choose BMC and benefit fr ? Choose BMC and benefit from om: : ischaemic heart disease and peripheral arterial disease: clinical implica‑ tions of COMPASS study. Eur Cardiol. 2018;13(2):115–8. fast, convenient online submission 38. Ferri LA, Bassanelli G, Savonitto S. Use of direct oral anticoagulant thorough peer review by experienced researchers in your field in ischaemic heart disease: the COMPASS study. Eur Heart J Suppl. 2019;21(Suppl B):B84‑b87. rapid publication on acceptance 39. van Melle JP, Bot M, de Jonge P, de Boer RA, van Veldhuisen DJ, Whooley support for research data, including large and complex data types MA. Diabetes, glycemic control, and new‑ onset heart failure in patients • gold Open Access which fosters wider collaboration and increased citations with stable coronary artery disease: data from the heart and soul study. Diabetes Care. 2010;33(9):2084–9. maximum visibility for your research: over 100M website views per year 40. Iribarren C, Karter AJ, Go AS, Ferrara A, Liu JY, Sidney S, Selby JV. Glycemic control and heart failure among adult patients with diabetes. Circulation. At BMC, research is always in progress. 2001;103(22):2668–73. Learn more biomedcentral.com/submissions 41. Li Y, Chen JP, Duan L, Li S. Eec ff t of vitamin K2 on type 2 diabetes mellitus: a review. Diabetes Res Clin Pract. 2018;136:39–51.

Journal

Cardiovascular DiabetologySpringer Journals

Published: Jan 6, 2023

Keywords: Oral anticoagulant; Heart failure; Atrial fibrillation; Diabetes mellitus; Elderly

There are no references for this article.