FormalPara Key Points for Decision Makers

Apixaban, left atrial appendage closure, propranolol, catheter ablation and convergent procedures are potentially efficient to manage atrial fibrillation in high-income countries. Apixaban, warfarin and radiofrequency catheter ablation are potentially efficient to manage atrial fibrillation in middle-income countries.

More population-specific clinical trials, head-to-head trials and other relevant population-specific studies are needed to provide more robust data for economic evaluation, which will in turn support decision making.

The decision to use any treatment strategy should always be individualised and guided by strong objective clinical and economic evidence and sound clinical judgement.

1 Introduction

Among the cardiac arrhythmias, atrial fibrillation (AF) remains the most common from a public health to clinical perspective [1]. In 2019, approximately 57.9 million people worldwide had AF, twice the number in 1990 [2], representing about 1% of the general population aged 18 years and above. Over 30% of hospital admissions for cardiac rhythm problems were due to AF [3]. AF independently confers a significant long-term risk for ischaemic stroke (fivefold), acute coronary events (including myocardial infarction) and the syndrome heart failure (HF) [1]. Independent of these deadly and disabling conditions, AF is associated with impaired quality of life and premature mortality [3, 4]. Within the progressively ageing populations of high-income countries in whom the antecedents for AF remain high (particularly hypertension), the burden AF is progressively increasing [1]. Although incident AF is closely related to older age, in low- and middle-income countries (LMIC) where hypertension remains largely undetected and uncontrolled, the future burden of AF is also likely to rise [5]. The management of this disease is also abysmal in LMIC due to poor access to healthcare [5].

The management of AF is usually individualised and dependent on several factors including patient factors (e.g. risk of a particular therapy for a patient, patient’s overall risk of stroke and other emboli-related problems) and diagnostic factors (e.g. severity of symptoms and the cause or duration of AF). In general, the short-term goals for the management of AF include symptom relief, and prevention of AF-associated complications including acute decompensated heart failure (HF) in those with impaired systolic function [1]. In the longer-term, key goals include the prevention of stroke, symptom relief, heart rate control, rhythm control and aggressive risk factors management [1]. Prevention of (ischaemic) stroke remains the principal goal in the management of most AF patients, although there is increasing focus on the prevention of chronic HF; particularly as HF begets AF and vice versa [6]. To date, randomised controlled trials (RCTs) do not suggest the superiority of rate control over rhythm control to achieve a normal rhythm. However, if the ultimate goal is restoration and maintenance of sinus rhythm, rate control medication is usually continued throughout the follow-up, unless continuous sinus rhythm is present. Managing AF-related abnormal heart rate may also control abnormal rhythm, but in some cases, cardioversion or ablation procedures are required to control the abnormal rhythm. There are different management strategies for AF, so it is important to stratify the risk profile of each patient, as some patients may require earlier and different intervention from others. This led to the establishment of the CHADS2 [congestive heart failure, hypertension, age ≥ 75 years, diabetes, stroke (doubled)] score. Recently, a new and updated version known as CHA2DS2-VASc [congestive heart failure, hypertension, age ≥ 75 years (doubled), diabetes, stroke (doubled), vascular disease, age 65–74 years and sex category (female)] has taken over as the most accurate risk stratification tool, offering more accurate results for low-risk patients [1].

The prevention of AF-related (ischaemic) stroke can be achieved by different intervention strategies. Anti-coagulants including Vitamin K antagonists (VKA) and Factor Xa inhibitors are the most commonly used interventions. Types of VKA that can be administered to prevent AR-related stroke include warfarin, acenocoumarol, phenprocoumon and fluindione. Historically, warfarin is the most used VKA and the mainstay of therapy when anti-coagulation is required. Modern alternatives to warfarin [due to the need for routine international normalized ratio (INR) therapeutic monitoring and high-risk of haemorrhagic strokes/other major bleeding events] are the Factor Xa inhibitors or the direct oral anticoagulants (DOAC). These include apixaban, dabigatran, edoxaban and rivaroxaban. Anti-platelets such as aspirin (a cyclooxygenase inhibitor) and, more latterly, clopidogrel (a selective adenosine diphosphate receptor inhibitor) are also used in AF but often in combination with anti-coagulants. In patients with AF from mechanical heart valves, low molecular weight heparin (LMWH) or unfractionated heparin (UFH) can be used for ‘bridging’ anti-coagulation. Direct thrombin inhibitors such as argatroban and bivalirudin can be used as alternatives for heparin. Studies have shown that DOACs are as effective as the VKAs in preventing AF-related stroke, but in terms of cost effectiveness, VKAs still appears as an attractive strategy in resource-limited settings [7] due to the high cost of DOACs. In high-income countries, however, DOACs seem to be the new conventional strategies [8]. Anti-coagulant treatment is likely unnecessary in males with CHA2DS2-VASc score of 0 and females with a score of 1. Scores ≥ 1 for males and ≥ 2 for females would require anti-coagulants.

Alternatively, strategies to control heart rate associated with AF include beta-blockers (metoprolol, bisoprolol, atenolol, esmolol, propranolol, carvedilol), non-dihydropyridine calcium channel antagonists (verapamil, diltiazem), digitalis glycosides (digoxin and digitoxin) and some class 3 anti-arrhythmic drugs (amiodarone and dronedarone) [1]. The use of any of these strategies for heart rate control depends on the treatment objectives such as acute rate control or long-term rate control, and factors such as left ventricular dysfunction and failed rhythm control. Anti-arrhythmic strategies (if required) include electric cardioversion, pharmacological cardioversion (flecainide, amiodarone, sotalol, ibutilide, propafenone and vernakalant), ablation procedures which could be catheter ablation (e.g. radiofrequency, cryoballoon or hybrid ablation—convergent procedure), or surgical ablation. As mentioned earlier, the use of any strategy for rhythm control depends on several factors including the symptoms, severity of AF (according to duration and class), ventricular dysfunction and haemodynamic stability. Studies have shown that rate and rhythm control strategies can reduce cardiovascular morbidities and mortality [9, 10].

In the long term, the key goal of any AF management strategy is to improve modifiable risk factors associated with the condition including lifestyle modification, monitoring and any potential triggers of paroxysmal AF (e.g. excessive alcohol intake). However, these strategies are often considered as secondary or tertiary and are used in combination with other strategies described above [11].

While some cases of AF are paroxysmal or persistent and self-terminate in less than a year with appropriate management, most cases transit to a life-long/permanent condition. This necessitates lifetime management and monitoring of rate control to reduce the increased risk of tachycardia-induced cardiomyopathy and stroke. Although both are very costly, this raise concerns around the cost effectiveness of applied management strategies. As the cost of managing AF is becoming as important as the health outcomes, economic evaluation of treatment strategies to manage AF has become an indispensable tool to inform decision making. This systematic review answers the following research question: What strategies to manage atrial fibrillation among people aged 18 years and above were cost effective in low-, middle- and high-income countries between 2012 and 2022? The outcome of this study will guide decision making in the management of AF amidst several strategies used in different settings to manage the disease.

2 Methods

2.1 Protocol and Registration

The design of this systematic review was in accordance with the recommendation in the Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) 2020 statement [12]. Details of the PRISMA checklist are provided in Electronic Supplementary Material (ESM) 1. This review has been registered in the International Prospective Register of Systematic Reviews (PROSPERO), CRD42022360590.

To aid the development of the study design, we did a preliminary assessment of literature using PubMed and Google Scholar to identify studies’ characteristics and methodologies. This helped to define the final inclusion and exclusion criteria used for the search.

2.2 Eligibility Criteria

Studies included in this review were original research on the economic evaluation of AF management strategies. The studies presented cost data and health outcome measure(s) for patients aged 18 years and above. There was no restriction on comorbidities or co-treatment strategies. Full-text articles were included. The studies were limited to humans only, in English language and published between January 2012 and November 2022.

Economic evaluations of other arrhythmias (other than atrial fibrillation) or other cardiac diseases, reviews and commentaries, opinion papers, conference abstracts or proceedings and qualitative reports were excluded. Studies whose costs and health outcomes were estimated using proxied data were also excluded. Studies with incomplete information needed for economic evaluation, and studies that evaluated one DOAC for stroke prevention were excluded, except in the case of mixed treatment strategies. The reason for excluding studies that evaluated a single DOAC was because the four novel DOACs (apixaban, dabigatran, edoxaban and rivaroxaban) were under patency in year 2022. Hence, to eliminate analytical bias of the included studies, we included only studies that evaluated the relative efficiency of at least two DOACs in head-to-head comparisons.

2.3 Information Sources

Based on expert librarians’ recommendations [13] and after an assessment of the health economic core library recommendations by the US National Library of Medicine [14], we searched MEDLINE (OvidSp), Embase, Web of Science, Cochrane Library, EconLit (EBSCOhost) and Google Scholar to identify relevant studies.

2.4 Search Strategy

The search was performed in September 2022. Related ‘search terms’ which include related relevant medical subject headings (MeSH), or related text words (title, abstract and keywords) were combined to form a union (concept cluster). For example, if A, B and C represent related search terms, the union was formed as ‘A or B or C’. Several relevant concept clusters were created. The concept clusters were combined to form an intersection cluster. For example, if D, E and F are concept clusters, they were combined as ‘D and E and F’. The results were then reviewed by looking at the MeSH, subheadings, titles and abstract to check if there are terms that could improve our search. For instance, in our preliminary search, we searched for ‘atrial fibrillation’ and ‘arrhythmia’ which present results with all related headings and subheadings and were combined to form a related MeSH ‘search term’. We then searched for text words like ‘arrhythmia’, ‘dysrhythmia’, and ‘abnormal heart rhythm’ appearing in the title, abstract or keywords, which were combined to form related text words, (search term). Both related ‘search terms’ (MeSH and text words) were combined to form a concept cluster. Details of the search strategies are shown in ESM 2. The MEDLINE search strategy was adapted for search in other databases. Auto-alert systems was set up to provide literature updates while the data extraction and analyses are ongoing. The auto-alert systems were stopped 2 months after the data extraction (November 2022).

2.5 Data Management and Selection Process

All searched results from the six databases were exported into a single EndNote library. A union group was created to contain all the articles from the different databases. De-duplication of studies was performed with EndNote. From the union group, different subgroups to represent exclusion criteria were created. Excluded studies were exported to different exclusion ‘group set’ based on the criteria for exclusion. A study that does not meet the inclusion criteria for multiple reasons was exported to the relevant exclusion ‘group set’ in the order of priority: ‘subject area’, ‘originality’, ‘economic evaluation’ and ‘clarity’ completeness, ‘study from 2012 and beyond’, and more than one DOAC evaluated (in the case of stroke prevention) (see ESM 2).

Selection was performed independently by two reviewers (C.O. and C.A.). A third reviewer (J.B.) overviewed the selections by the first two reviewers and resolved any selection disagreement among the first two reviewers. First, titles were screened followed by the abstracts of original research articles that involve the economic evaluation of atrial fibrillation management strategies. Next, an auto search for the full text of the articles whose abstracts were eligible was initiated. The full text of the potential articles was assessed for costs, health outcomes and clarity of reporting.

2.6 Data Collection Process

An electronic data extraction form was used. Two reviewers (C.O. and C.A.) independently extracted and managed the data from the included studies. Disagreement with the extraction results from the two reviewers that cannot be resolved by them was resolved by the third reviewer (J.B.). The data was collected based on the 2022 International Society for Pharmacoeconomics and Outcome Research (ISPOR) Consolidated Health Economic Evaluation Reporting Standards (CHEERS) guideline [15] (see ESM 2).

2.7 Data Items, Outcomes and Prioritisation

Data was extracted based on the following:

  1. (a)

    Publication: title, authors, year, study objectives, sample size, gender, setting/country the study was conducted, etc.

  2. (b)

    Study design: randomised-control trials, cohort studies, case-control studies, cross-sectional studies, type of economic evaluation, time horizon, type of atrial fibrillation, comparators, risk of stroke based on the CHA2DS2-VASc score, etc. A CHA2DS2-VASc score of 0 is ‘low’ risk of stroke, 1 is ‘moderate’ and any score above 1 is a ‘high’ risk.

  3. (c)

    Cost and outcome measures: cost perspective, quality-adjusted life-years (QALY), discount rate, willingness-to-pay threshold, price year and currency, etc.

  4. (d)

    Other relevant information, e.g. study assumptions.

2.8 Risk of Bias Assessment

Risk of bias assessment of the individual studies was done at the outcome level using the Consensus Health Economic Criteria (CHEC) checklist designed for conducting economic evaluation-based systematic reviews [16]. This checklist has 19 reporting standards for economic model characteristics, identification and valuation of costs and outcomes, discussion section, conclusions as well as funding and conflicts of interest statement. As the risk of bias tool did not provide thresholds to include or exclude studies for data synthesis, an internally created conservative classification system was used to ensure that only studies with low risk to moderate risk of bias were included in the synthesis. The risk of bias in this scenario was defined as the inappropriateness or failure of a study to conduct or report the applicable items indicated on the risk of bias tool. The risk of bias was measured as a percentage of the failure or inappropriateness. Studies were classified as ‘low risk’ (0–10%), ‘moderate risk’ (11–30%), or ‘high risk’ (> 30%) based on the percentage of applicable items for each study which were not reported or inappropriately reported. Studies with high risk of bias were excluded from the data synthesis. The checklist was completed in duplicate by two members of the review team (C.O. and C.A.). Differences were resolved with the third author (J.B.). Details of the risk of bias assessment is available in the supplementary file 3 (ESM 3).

2.9 Data Synthesis

A narrative synthesis and summary of answers to the research question was performed due to heterogeneities in the identified studies. Such heterogenies emanated from either difference in country’s economic development or treatment goals for managing AF. Hence, the grou** of studies for narrative synthesis was first based on the income level classification by the World Bank [high-income countries (HIC), middle-income countries (MIC) and low-income countries (LIC)]. Studies were then grouped as per the treatment goals including stroke prevention, rate control and rhythm control. The analysis was performed using Microsoft Excel 365 (Microsoft, Seattle, USA). The ICER for each study was adjusted to 2021 USD value following the guidelines of the Campbell and Cochrane Economics Methods Group (CCEGM) and the Evidence for Policy and Practice Information and Coordinating Centre (EPPI-Centre) [17].

3 Results

3.1 Screening and Selection of Studies

A total of 2019 articles were identified from five databases (Medline, 221; Embase, 1108; Web of Science, 518; Cochrane, 102 and EconLit, 9) and Google Scholar (61 articles). Two potential articles from Medline were identified through the auto alert system but were duplicates. After de-duplication, 1803 articles were available for screening. The studies excluded at the screening phase were moved to exclusion ‘group sets’ based on the reason for exclusion in a hierarchy order described in the methodology. A total of 230 potential articles were available for eligibility check after phase 1 screening. Conference abstracts and studies which had multiple, relevant missing information for economic evaluation were excluded as ‘incomplete information’ studies, while studies with unspecific methodology or analytical approach were excluded as ‘not specific’ studies. At the end of the eligibility assessment, a total of 50 studies met the inclusion criteria. Figure 1 describes the flow diagram of the selection process. The EndNote library (available upon request) summarises the reasons for exclusion of studies.

Fig. 1
figure 1

PRISMA flowchart of the study selection process

Full size image

3.2 Characteristics of the Included Studies

Data were extracted from the included studies as per the data items described in the methodology. All studies were for patients with non-valvular AF. No study was found to evaluate the management strategies of AF in LIC setting. For stroke prevention strategies, 32 studies evaluated the cost effectiveness in HIC [8, 18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,

5 Conclusions

Based on the available evidence, in high-income countries, the review suggests the use of apixaban in patients at low and moderate risk of stroke, and LAAC in patients at high risk. Propranolol should be considered for rate control in patients without contraindication due to cardioselectivity. Catheter ablation and convergent procedure are suggested in paroxysmal and persistent AF, respectively. In middle-income countries, apixaban is suggested for stroke prevention while warfarin should be considered in the case of financial constraint. Radiofrequency catheter ablation is suggested in rhythm control.

Due to limited clinical trials on AF management strategies, the lack of head-to-head trials (especially on the new oral anti-coagulants), the heterogeneity in clinical characteristic of the trials from which the data used in most of the included studies in this review were derived from, and the variations in the modelling methods amongst the studies, the decision to use any treatment strategy should always be individualised and guided by strong objective and subjective clinical and economic evidence in a multi-disciplinary team setting.