Introduction
Contemporary studies show that 20% to 30% of patients with ischemic stroke have undiagnosed atrial fibrillation (AF) and AF independently increases risk of all-cause mortality.1,2 Mortality due to ischemic stroke can largely be mitigated by anticoagulation, while other types of cardiovascular death, for example due to heart failure, remain common even in patients with AF treated according to the current guidelines.3 Not surprisingly, given increased morbidity in patients with AF, between 10% to 40% of patients with AF are hospitalized each year.4 The management of AF in routine practice differs significantly from the clinical trial setting due to differences in patient baseline characteristics as well as in care those patients receive. To investigate this issue, we aimed to compare long-term outcomes in patients with AF treated in both academic and district hospitals.
Methods
Study population
This retrospective observational cohort study included data from the Multicenter Experience in Atrial Fibrillation Patients Treated with Oral Anticoagulation (CRAFT) study (ClinicalTrials.gov identifier, NCT02987062). Details about the study design and main results were reported elsewhere.5 Briefly, the CRAFT study included patients aged 18 or older, with electrocardiogram-based diagnosis of AF, hospitalized between 2011 and 2016 at academic and district hospitals. Due to the retrospective nature of the study, the need for approval of a local ethics committee was waived. All patients provided written informed consent.
Primary and secondary end points
The primary end point was assigned as a major adverse event (MAE) defined as all-cause death and thromboembolic and hemorrhagic events during the median follow-up of 4 years. The end of follow-up was set at January 16, 2019. The secondary end points were defined as components of the primary end point. Thromboembolic events included ischemic stroke (different locations), transient ischemic attack, and peripheral thromboembolism (different locations). Hemorrhagic events included gastrointestinal, intracranial, and other types of bleeding. Data on long-term outcomes were obtained from the Polish National Health Fund (Narodowy Fundusz Zdrowia) that gathers data about medical services including primary diagnosis coded with the International Classification of Diseases, Tenth Revision (ICD-10). Supplementary material, Table S1 presents the list of the codes used in the study.
Statistical analysis
All continuous variables were tested for normality with the Shapiro–Wilk test. All continuous variables were assigned as nonparametric and expressed as median (interquartile range). Categorical variables were expressed as counts with percentages. The Fisher exact test (2-group comparison) or the χ2 test (3-group comparison) was used to compare categorical variables. Differences in continuous parameters were compared using the Mann–Whitney test (2-group comparison) and the Kruskal–Wallis test (3-group comparison). The Bonferroni correction was applied to address the multiple comparison issue. To determine predictors of primary and secondary end points in both groups, the Cox proportional-hazard regression method was used to fit univariable and multivariable survival models, the results of which are reported as hazard ratios (HRs) with 95% CIs. Among the factors, those with a P value of less than 0.05 were selected and included in the multivariable Cox regression analysis.
To adjust for potential confounding due to baseline imbalances in study covariates while preserving sample size, we used propensity score matching.6 With this method, the propensity score (district hospital treatment was set as the reference) was used to generate patient-specific stabilized weights that control for covariate imbalances. Covariate balance between the weighted cohorts was assessed using standardized mean differences. A standardized difference of 0.05 or less indicates a negligible difference between groups. The distributions of propensity scores and stabilized weights were inspected for outliers. Analyses presented in Figures 1 and 2 as well as in Supplementary material, Tables S2 and S5 were based on propensity score matching–adjusted cohorts. Weighted Cox proportional hazards regression with robust estimation was used to estimate time-to-primary end point event in district compared with academic (reference) cohorts. Variables used for propensity score calculations included: age, sex, AF type, heart failure, hypertension, coronary artery disease, diabetes mellitus, ischemic stroke / transient ischemic attack / thromboembolism, previous major bleeding, chronic obstructive pulmonary disease, smoking, renal function based on estimated glomerular filtration rate (calculated using the Cockcroft–Gault formula). Weighted Kaplan–Meier analysis was used to establish the relation of type of hospital (academic vs district) to MAE and its components, and differences in adverse events were analyzed using the log-rank test. Subgroup-specific adjusted HRs with 95% CIs analyses were performed for all outcomes in categories defined by age, sex, heart failure, hypertension, coronary artery disease, diabetes, chronic kidney disease, and the HAS-BLED score. A 2-tailed P value of 0.05 was considered statistically significant. For database management and statistical analysis, we used SAS, version 14.1 (SAS Institute Inc, Cary, North Carolina, United States).
Figure 1. Kaplan–Meier analysis of time to: major adverse events (A), all-cause death (B), thromboembolic events (C), and hemorrhagic events (D) in patients hospitalized in academic and district hospitals, after propensity-score matching
Figure 2. Subgroup-specific hazard ratios with 95% CIs for major adverse events (A) and all-cause death (B) and thromboembolic (C) and hemorrhagic events (D) in patients treated in academic (reference) vs district hospitals, after propensity-score matching
Abbreviations: CAD, coronary artery disease; DM, diabetes mellitus; HF, heart failure; HR, hazard ratio; HT, hypertension
Results
Study population
Out of the entire cohort of 3528 patients with AF included in the CRAFT study, follow-up data were available for 3307 individuals. Out of them, 2983 patients had indications for long-term oral anticoagulation (OAC) treatment in accordance with the European Society of Cardiology guidelines4 and were included in the current analysis. The rest of the cohort had temporary indications to OAC, for example, before and after cardioversion / ablation.
Baseline characteristics
Patients were divided into 2 groups: 2271 patients (76.1%) hospitalized in an academic hospital and 712 patients (23.9%) admitted to a district hospital. Differences between patients recruited in the district vs academic hospital with regard to baseline characteristics were reported elsewhere.7 Briefly, in district hospitals, patients were older, more likely female, more frequently had permanent AF and were more likely to have comorbidities and a higher thromboembolic and bleeding risk as compared with those hospitalized in academic hospitals (Table 1). After propensity score matching, the cohorts were well balanced across all covariates (Supplementary material, Table S2).
Variable | All patients (n = 2983) | Academic hospitals (n = 2271) | District hospitals (n = 712) | P value |
|---|---|---|---|---|
Demographics | ||||
Age, y | 70 (62–79) | 68 (61–78) | 76 (68–82) | <0.001 |
Female sex | 1224 (41) | 872 (38) | 352 (49) | <0.001 |
Atrial fibrillation typea | ||||
Paroxysmal | 1497 (52); (n = 2868)b | 1201 (56); (n = 2156) | 296 (42) | <0.001 |
Long-standing persistent | 99 (3.5); (n = 2868) | 99 (4.6); (n = 2156) | 0 | <0.001 |
Persistent | 380 (13); (n = 2868) | 276 (13); (n = 2156) | 104 (15) | 0.06 |
Permanent | 868 (30); (n = 2868) | 556 (26); (n = 2156) | 312 (44) | <0.001 |
Comorbidities | ||||
Heart failure | 1185 (40); (n = 2978) | 748 (33) | 437 (62) (n = 707) | <0.001 |
Hypertension | 2363 (79); (n = 2979) | 1829 (81); (n = 2269) | 5334 (75); (n = 710) | <0.001 |
CAD | 1366 (46) | 916 (40) | 450 (63) | <0.001 |
DM | 854 (29); (n = 2970) | 591 (26); (n = 2267) | 263 (37) (n = 703) | <0.001 |
History of TEs | 423 (14); (n = 2975) | 294 (13); (n = 2270) | 129 (18); (n = 705) | <0.001 |
History of bleeding | 241 (8.1); (n = 2981) | 67 (3) | 174 (30); (n = 710) | <0.001 |
COPD | 284 (10); (n = 2978) | 161 (7.1); (n = 2270) | 123 (17); (n = 708) | <0.001 |
Smoking | 165 (5.5); (n = 2973) | 94 (4.1); (n = 2268) | 71 (10); (n = 705) | <0.001 |
Laboratory parameters | ||||
Hemoglobin, g/dl | 14 (13–15); (n = 2258) | 14 (13–15); (n = 2258) | NA | NA |
Platelet count, 103/mm3 | 204 (168–239); (n = 2262) | 204 (168–239); (n = 2262) | NA | NA |
eGFR ≥50 ml/min/1.73 m2 | 1718 (73); (n = 2356) | 1195 (73); (n = 1646) | 523 (74) (n = 710) | 0.61 |
Thromboembolic and bleeding scores | ||||
CHA2DS2-VASc score | 4 (2–5) | 3 (2–5) | 5 (3–6) | <0.001 |
HAS-BLED score | 2 (1–2) | 2 (1–2) | 2 (2–3) | <0.001 |
Antithrombotic treatment | ||||
VKA | 1696 (57) | 1410 (62) | 276 (39) | <0.001 |
NOAC | 1297 (43) | 861 (38) | 436 (61) | <0.001 |
Dabigatran | 406 (14) | 285 (13) | 121 (17) | 0.003 |
Rivaroxaban | 891 (30) | 576 (25) | 315 (44) | <0.001 |
Reduced doses | 556 (43); (n = 1282) | 318 (38); (n = 848) | 238 (55); (n = 434) | <0.001 |
Antiplatelet drugs | 435 (15) | 351 (16) | 84 (12) | 0.02 |
Other medications | ||||
β-Blockers | 1907 (84); (n = 2270) | 1907 (84); (n = 2270) | NA | NA |
Calcium channel blockers | 524 (23); (n = 2270) | 524 (23); (n = 2270) | NA | NA |
Antiarrhythmic drugs | 511 (23); (n = 2980) | 392 (17); (n = 2269) | 119 (17); (n = 711) | 0.78 |
RAS inhibitors | 1848 (81); (n = 2271) | 1848 (81) | NA | NA |
Statins | 1539 (68); (n = 2271) | 1539 (68) | NA | NA |
Long-term outcomes | ||||
MAEs | 1228 (41) | 849 (37) | 379 (53) | <0.001 |
All-cause death | 828 (28) | 541 (24) | 287 (40) | <0.001 |
TEs | 273 (9.2) | 178 (7.8) | 95 (13) | <0.001 |
HEs | 445 (15) | 339 (15) | 106 (15) | 1.00 |
Data are presented as number (percentage) or median (interquartile range). a The Bonferroni correction was applied to address the multiple comparison issue. b Numbers provided in parentheses indicate the total number of patients available for that variable. Abbreviations: COPD, chronic obstructive pulmonary disease; CRT, cardiac resynchronization therapy; eGFR, estimated glomerular filtration rate; HE, hemorrhagic event; ICD, implantable cardioverter-defibrillator; MAE, major adverse event; NA, not applicable; NOAC, non–vitamin K antagonist oral anticoagulant; PM, pacemaker; RAS, renin–angiotensin system; TE, thromboembolic event; VKA, vitamin K antagonist; others, see Figure 2 | ||||
Follow-up outcomes
During the median follow-up of 4 years, 1228 patients (41%) experienced MAEs: 828 (28%) died, 273 (9.2%) reported thromboembolic events, and 445 (15%) hemorrhagic events. Overall, patients treated in district hospitals more often experienced MAEs (53% vs 37%; P <0.001), all-cause death (40% vs 24%; P <0.001), thromboembolic events (13% vs 7.8%; P <0.001) with similar risk of hemorrhagic events (15% vs 15%; P = 1.00) as compared with those hospitalized in academic hospitals (Figure 1). District (vs academic) conditions were associated with higher risk of MAEs and all-cause death in men and those with low risk of bleeding, and with higher incidence of thromboembolic events in women, elderly patients, and those with high risk of bleeding and diabetes (Figure 2).
In multivariable logistic regression, in academic hospitals, female sex, coronary artery disease, smoking, and antiplatelet drug therapy were significantly associated with a greater likelihood of thromboembolic events. Heart failure, renal failure, and the use of a vitamin K antagonist (VKA) were significantly associated with a greater likelihood of hemorrhagic events in academic hospitals. On the other hand, only coronary artery disease was significantly associated with hemorrhagic events in patients admitted to district hospitals (Supplementary material, Table S3).
Oral anticoagulant treatment and long-term outcomes
In academic settings, 576 (25%), 285 (13%), and 1410 (62%) patients were treated with rivaroxaban, dabigatran, and a VKA, respectively, whereas in district hospital—315 (44%) with rivaroxaban, 121 (17%) with dabigatran, and 276 (39%) with a VKA. In both types of hospitals, the incidence of MAEs and all-cause death was the highest within VKA-treated patients, and the lowest within those who received dabigatran. The highest and the lowest risk of hemorrhagic events was observed only in VKA and dabigatran groups respectively, only in patients hospitalized in academic hospitals (Supplementary material, Table S4). There were no differences between patients who were prescribed with a VKA and those on non-VKA OAC (NOAC) according to MAEs incidence and its components with a single exception. There were higher rates of hemorrhagic events among patients who received a VKA (vs dabigatran) and higher rate of hemorrhagic events among patients treated with rivaroxaban (vs a VKA) hospitalized in district hospitals (Supplementary material, Table S5).
Discussion
The current study presented a unique and direct comparison between anticoagulation treatment of patients with AF treated in academic and district conditions with a special focus on long-term outcomes. Despite the introduction of standardized guidelines that are regularly updated to best adapt the treatment to the current needs of patients, the management of AF still poses a challenge.8 The overall profile of patients with AF is diversified and may be affected by regional population demographics and risk factors. At the same time, AF management varies between primary care settings in relation to resources, fund allocation, and equipment. However, there is a paucity of data regarding possible differences in the effectiveness of implementing AF treatment strategies in patients between academic and district hospitals. Most registries collect data mainly from large academic centers or highly specialized hospitals with poor representation of district hospitals. In the German Atrial Fibrillation NETwork (AFNET) study,9 two-thirds of 9577 patients were treated by office-based cardiologists and tertiary care centers, followed by district hospitals and general practitioners or internists. The authors found that the center type affected the decision on the stroke prevention strategy and heart rhythm and / or the rate control therapy.9 Although the risk of stroke was similar between patients from particular centers, only tertiary care centers (68.8%) and office-based cardiologists (73.6%) more frequently administered adequate antithrombotic therapy than district hospitals (55.1%) or general practitioners (52%). This study also showed that, taking university hospital–based cardiologists as the reference type of center, office-based cardiologists (odds ratio [OR], 1.40) prescribed significantly more frequently guideline-recommended anticoagulation for secondary stroke prevention than regional hospitals (OR, 0.47) and general practitioners / internists (OR, 0.40). This physician-dependent under-utilization of OACs in AF cohorts has previously been described for office-based general practitioners as compared with cardiologists.10 Adherence to anticoagulation guidelines is associated with better survival in patients with AF at a high risk of stroke by reducing morbidity, all-cause mortality, and stroke.11-13 Interestingly, higher rates of NOAC prescription in district hospitals were observed. It might be related to the available guidelines at the moment of hospitalization, in which NOACs had a lower class of recommendation. It is likely that some patients had their treatment changed after the publication of new evidence. The current study assessed drugs that were prescribed at discharge from the hospital. Unfortunately, the dosage regimen was insufficiently reported to be analyzed. However, in our study, despite OAC prescription in all patients, those hospitalized in the district hospital more often experienced MAEs including all-cause death and thromboembolic events. To explain this observation, one should bear in mind that patients from academic hospitals more frequently had undergone pulmonary vein isolation and electrical cardioversion than patients from district hospitals.7 For organizational reasons, most elective cardioversions in district hospitals are performed in the emergency department and, simultaneously, no AF ablations are performed, which can explain the different distribution of AF types and stroke risk in both types of centers.7 Moreover, based on our previous study,14 most Polish patients with AF, even those with the lowest risk of stroke, are often overtreated with NOACs and as those results came from academic centers, the rate of inappropriate dosage of NOACs could be even higher in district hospitals.
Literature search revealed that NOACs have similar efficacy as VKAs in the prevention of thromboembolic events in AF patients, but significantly reduce the risk of major bleeding.15,16 It was also observed in our study. There were no differences between patients on VKAs and NOACs with regard to thromboembolic events; however, a higher rate of hemorrhagic events was observed in the VKA group (vs dabigatran) and the rivaroxaban group (vs VKA) in patients hospitalized in district hospitals.
Limitations of the study
This retrospective study has several limitations. Firstly, it was not a nationwide registry with a truly representative cohort of patients with AF. Secondly, only inpatients were included in the registry. Moreover, our registry is limited by the fact that it depends on the data obtained from cardiology departments only. Finally, due to the small number of patients on apixaban, it was not included into the analysis.
Conclusions
Long-term outcomes of patients with AF depend not only on patient characteristics, but also the type of healthcare system. Further research is needed to investigate the differences in the management of AF and long-term outcomes between academic hospitals (participating in global registries that are the source of AF guidelines), and district hospitals (where these guidelines are implemented in real-life clinical practice).
Monika Gawałko, MD, 1st Department of Cardiology, Medical University of Warsaw, ul. Banacha 1a, 02-097 Warszawa, Poland, phone: +48 22 599 29 58, email: mongawalko@gmail.com
May 16, 2021.
June 27, 2021.
July 2, 2021.
The authors thank Katarzyna Żukowska, Martyna Zaleska, Katarzyna Szepietowska, Kacper Maciejewski, Anna Praska-Oginska, and Inna Zaboyska for data collection.
PL, MG, and PB were responsible for the concept and design of the study. MG, LK, AŚ, CM, BK, AT, and KO were involved in data collection. PL and MG analyzed the data. MG was responsible for statistical analysis. PL and MG wrote the draft of the manuscript. All authors edited and approved the final version of the manuscript.
PL, GO, and PB received speaker honoraria from Bayer, Boehringer Ingelheim, and Pfizer. AT received speaker honoraria from Novartis and Boehringer Ingelheim. KO received speaker honoraria from Novartis, Boehringer Ingelheim, and Orion Pharma. Other authors declare no conflict of interest.
Lodziński P, Gawałko M, Kraj L, et al. District versus academic hospitals: clinical outcomes of patients with atrial fibrillation. The Multicenter Experience in Atrial Fibrillation Patients Treated with Oral Anticoagulants (CRAFT) study. Pol Arch Intern Med. 2021; 131: 16053. doi:10.20452/pamw.16053
- 1.
- Henriksson KM, Farahmand B, Asberg S, et al. Comparison of cardiovascular risk factors and survival in patients with ischemic or hemorrhagic stroke. Int J Stroke. 2012; 7: 276-281.Crossref
- 2.
- Kishore A, Vail A, Majid A, et al. Detection of atrial fibrillation after ischemic stroke or transient ischemic attack: a systematic review and meta-analysis. Stroke. 2014; 45: 520-526.Crossref
- 3.
- Hindricks G, Potpara T, Dagres N, et al. 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association of Cardio-Thoracic Surgery (EACTS). Eur Heart J. 2021; 42: 373-498.
- 4.
- Kirchhof P, Benussi S, Kotecha D, et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Europace. 2016; 18: 1609-1678.
- 5.
- Balsam P, Ozieranski K, Tyminska A, et al. Comparison of clinical characteristics of real-life atrial fibrillation patients treated with vitamin K antagonists, dabigatran, and rivaroxaban: results from the CRAFT study. Kardiol Pol. 2018; 76: 889-898.Crossref
- 6.
- Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011; 46: 399-424.Crossref
- 7.
- Bednarski J, Cieszewska E, Strzelecki A, Filipiak KJ. Anticoagulant and antiplatelet therapy for stroke prevention in atrial fibrillation patients in the clinical practice of a single district hospital in Poland. Kardiol Pol. 2013; 71: 1260-1265.Crossref
- 8.
- Farkowski MM, Karliński MA, Sterliński M, et al. Educational needs among physicians treating patients with atrial fibrillation: lessons for Poland from the European Society of Cardiology international educational needs assessment study. Pol Arch Intern Med. 2019; 129: 586-591.Crossref
- 9.
- Kirchhof P, Nabauer M, Gerth A, et al. Impact of the type of centre on management of AF patients: surprising evidence for differences in antithrombotic therapy decisions. Thromb Haemost. 2011; 105: 1010-1023.Crossref
- 10.
- Turakhia MP, Hoang DD, Xu X, et al. Differences and trends in stroke prevention anticoagulation in primary care vs cardiology specialty management of new atrial fibrillation: The Retrospective Evaluation and Assessment of Therapies in AF (TREAT-AF) study. Am Heart J. 2013; 165: 93-101.Crossref
- 11.
- Hylek EM, Go AS, Chang Y, et al. Effect of intensity of oral anticoagulation on stroke severity and mortality in atrial fibrillation. N Engl J Med. 2003; 349: 1019-1026.Crossref
- 12.
- Nieuwlaat R, Olsson SB, Lip GY, et al. Guideline-adherent antithrombotic treatment is associated with improved outcomes compared with undertreatment in high-risk patients with atrial fibrillation. The Euro Heart Survey on Atrial Fibrillation. Am Heart J. 2007; 153: 1006-1012.Crossref
- 13.
- Camm AJ, Kirchhof P, Lip GY, et al. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Europace. 2010; 12: 1360-1420.
- 14.
- Lodziński P, Gawałko M, Budnik M, et al. Trends in antithrombotic management of patients with atrial fibrillation. A report from the Polish part of the EURObservational Research Programme – Atrial Fibrillation General Long-Term Registry. Pol Arch Intern Med. 2020; 130: 196-205.Crossref
- 15.
- Acanfora D, Ciccone MM, Scicchitano P, et al. Efficacy and safety of direct oral anticoagulants in patients with atrial fibrillation and high thromboembolic risk. A systematic review. Front Pharmacol. 2019; 10: 1048.Crossref
- 16.
- Undas A, Drabik L, Potpara T. Bleeding in anticoagulated patients with atrial fibrillation: practical considerations. Pol Arch Intern Med. 2020; 130: 47-58.Crossref