Critical appraisal of guidelines for coronary artery disease on dual antiplatelet therapy: More consensus than controversies

Abstract Background Dual antiplatelet therapy (DAPT) in the form of aspirin plus a P2Y12 inhibitor, when indicated, is one of the key treatments in coronary artery disease (CAD). Many recommendations on DAPT in patients with CAD based on current guidelines are largely inconsistent. In our current study, we aimed at systematically reviewing DAPT‐relevant clinical practice guidelines, and highlighting their commonalities and differences for better informed decision‐making. Methods Contemporary guidelines in English were searched in MEDLINE, Embase and websites of guideline organizations and professional societies. Guidelines with recommendations on DAPT for CAD patients were included. Guideline quality was appraised with the 6‐domain Appraisal of Guidelines for Research and Evaluation II (AGREE II) instrument. The reporting of conflicts of interest (COI) was assessed individually with supplementary items from the RIGHT (Reporting Item for Practice Guidelines in Healthcare) checklist. Meanwhile, extraction of recommendations was performed. Results A total of 18 guidelines fulfilled our inclusion criteria. Most of them were graded with relatively good scores averaging from 42% to 74%. Domains for lower scores were in “stakeholder involvement” and “application.” The reporting of COI was satisfactory. For the recommendations on DAPT, most guidelines with high AGREE II scores included consistent recommendations on the timing and P2Y12 inhibitor selection. Nonetheless, conflicts still exist on the duration of DAPT. Conclusions Quality of guidelines for DAPT in CAD was relatively high, though defects existed in “Applicability” and “Stakeholder Involvement.” As these guidelines developed, DAPT recommendations gradually converged on a consensus. Clinical decision should be made on an individual basis.


| INTRODUCTION
Coronary artery disease (CAD) is the leading cause of morbidity and mortality globally. The high burden of disease and the costs seriously affect individuals, families, and society at large. 1 Dual antiplatelet therapy (DAPT) in the form of aspirin plus a P 2 Y 12 inhibitor, when indicated, is one of the key treatments in CAD. For example, it is the secondary prevention for patients after acute coronary syndrome (ACS) and prevents stent thrombosis in patients with recent stent implantation. 2 In the meantime, clinicians usually rely on evidencebased clinical guidelines to make decision. As a result, several international organizations have published clinical guidelines (CPGs) with recommendations of DAPT for patients with CAD.
There were more than 10 DAPT-relevant CPGs published in the past two decades. Though most of them claimed to be evidencebased, we found a considerable amount of differences in their recommendations, which may be confusing for clinicians. Since the quality of these guidelines remained unclear, critical appraisal of these guidelines is important. Several methodologies have been developed for the assessment of CPGs. 3 Among these, the Appraisal of Guidelines for Research and Evaluation (AGREE) instrument was widely accepted and also endorsed by the WHO, 4 which was then updated to AGREE II. 5 Systematic review of CPGs using AGREE (AGREE II) was documented to be of tremendous value. 6 2 | METHODS

| Data sources and searches
A systematic literature search for existing guidelines was performed in MEDLINE and Embase from 1 January 1999 to 1 September 2019, using keywords of "coronary artery disease," "stable coronary artery disease," "angina," "acute coronary syndrome" and "guideline." Additional guidelines were obtained by reviewing websites of guideline organizations and professional societies (see Table S1).

| Quality appraisal of guidelines
We assessed the quality of each included guideline using the standardized Appraisal of Guidelines Research and Evaluation instruction II (AGREE II), 5 which consists of 23 items in six domains: (a) scope and purpose; (b) stakeholder involvement; (c) rigor of development; (d) clarity of presentation; (e) applicability; (f) editorial independence.
Each domain was independently rated by two reviewers (Z.S.Z and Z.H.M) who were blinded to each other's ratings. Each item within an individual domain was rated from 1 (strongly disagree) to 7 (strongly agree). For each reviewer, AGREE II scores of each domain were calculated as a percentage using the sum of all items and the maximum possible score. Agreement between reviewers on AGREE II scores was assessed using the intraclass correlation coefficient by two-way random model with the type of absolute agreement. The final scores of each domain was calculated by combining the scores of each reviewer based on the recommended formula: (S ob − S minp )/ (S maxp − S minp ), where S ob is the obtained score, S minp is the minimum possible score, and S maxp is the maximum possible score. A guideline was "strongly recommended for use in practice" if most domains (four or more) scored above 60%. A guideline was "recommended for use with some modification" if most domains scored between 30% and 60%. "Not recommended for use in practice" implied that most of the domains of the guideline were scored as approximately or below 30%. 8,9 In addition, we assessed the reporting of conflicts of interest (COI). Except for the two items in domain 6 of AGREE II, another four items from the RIGHT (Reporting Item for Practice Guidelines in Healthcare) checklist 10,11 were also appraised by one reviewer (Z.S. Besides, the proportion of guideline panel member-industry relationships was also calculated. 6 Discrepancies were resolved by discussion. Recommendations were categorized into strategies for patients with percutaneous coronary intervention (PCI), medical therapy or fibrinolytic therapy. Each recommendation was categorized into DAPT initial timing, P 2 Y 12 inhibitor selection and DAPT duration.

| Data extraction and analysis
We examined the correlation between the proportion of guideline panel members who reported relationships with industry and the AGREE II score with guideline. The guidelines (2012 JCS and 2018 TSC) did not have explicit statement on COI of panel members was excluded from the analyses. All analyses were performed using SPSS 21.0 and P < .05 was considered statistically significant.

| Selected guidelines
A total of 18 guidelines from seven organizations were identified ( Figure 1). Table 1  one from Australia and New Zealand, 25 two from Japan, 26,27 and one from Taiwan. 28

| Guideline appraisal by AGREE II
The final scores of each domain for the included guidelines were shown in Figure 2 (specific scores could be found in Table S2). This described as "strongly recommended" with average AGREE II scores ranged from 62% to 74%, while others were described as "recommended with modification." No guideline was described as "not recommended for use in practice." Reproducibility of the two reviewers' average AGREE II was good, with an intraclass correlation coefficient of 0.79.
For domain 1 (scope and purpose), most guidelines scored as high as around 80% and 2013 NICE1 even scored 94% for clearly describing their objective, health question and target population according to the principles of AGREE II. 5 However, 2012 JCS did not define these items well and thus received a relatively low score of 47%.
In domain 2 (stakeholder involvement), most guidelines scored poorly (below 60%), for none of them had sought the views and pref- Most guidelines clearly presented different options for management.
Moreover, most of the key recommendations could be easily identified.
Domain 5 (applicability) got the poorest score averaging 38%, for these guidelines rarely described facilitators and barriers to the application. In addition, most of them did not provide advice or tools to help clinicians to put recommendations into practice, except for guidelines from European Society of Cardiology (ESC), which all provided educational tools and implementation programmers. 1,[12][13][14][15] Few clearly presented the monitoring and auditing criteria.

| Recommendations for DAPT
Recommendations about DAPT in CAD were summarized in Table 2  did not mention about. In addition, recommendations for P 2 Y 12 inhibitor selection or DAPT duration were similar to that for NSTE-ACS (

| DISCUSSION
In general, we identified 18 guidelines with recommendations on DAPT for patients with CAD. To the best of our knowledge, it was the first comprehensive evaluation of the international DAPT guidelines. Based on the rigorous systematic assessment recommended by the AGREE II appraisal instrument, we found the general quality of these guidelines to be relatively high, though minor deficiencies still existed. And for the recommendations, there were still inconsistencies primarily with regards to DAPT duration.
Several recently published articles have assessed the quality of clinical guidelines using AGREE II tool, including various topics from different specialties. 6,29,30 Low scoring domains in these studies were found mostly in domain 2 "stakeholder involvement," domain 3 "rigor of development" and domain 5 "applicability," which reflected the widespread problem of the existing guidelines in different specialties.
In our study, it is somewhat reassuring to note that, most of these DAPT-relevant guidelines, especially more recent guidelines of ESC, AHA/ACC and NICE got relatively good score in domain "rigor of development." This showed that most DAPT guidelines were strictly followed by evidence-based principles when they were developed. It was significant as it ensured the DAPT recommendations in these guidelines were based on evidence and were credible. However, similar to other studies, we found most DAPT guidelines got a low score in domain "stakeholder involvement" and "applicability." In contrast to the rigor of the guidelines, most guidelines did not attach importance to the representativeness of the guidelines' stakeholder and the application of the recommendations. In addition, none of the guidelines presented the views and preferences of the target population, which would limit the judicial use of the guidelines in various scenarios.
Improving this item would help make the clinical guidelines more accessible to a wider audience. 31 Also, many guidelines failed to pre- It is generally believed that appropriate use of high-quality guidelines can improve patient outcomes. 3 In the last 23 years, various lengths of appropriate duration were proposed. 38 Although these studies provided significant amount of data, the results of the main studies were apparently inconsistent because of the differences in study designs, type of coronary devices, and targeted populations. 2 As more and more evidence accumulated, major controversy revolved around DAPT prolongation or shortening of the standard duration. 14,38 Premature discontinuation of DAPT before standard duration may lead to stent thrombotic events, [39][40][41] while shortened duration is associated with lower bleeding rate and non-inferiority cloud be found in several studies. 42 There are several possible limitations in our study. First, only guidelines published in English were included. Hence, many meaningful guidelines from non-English regions were not included in our study. Besides, one of the guidelines in our study from Japan, the 2012 JCS, was only a digest version published in English and did not provide the detailed information of its own development. This may account for the low AGREE II scores of it. In addition, the only three guidelines from Asia in our study could not represent other Asian countries. Second, the AGREE II tool considers the reported information only, whereas the real quality of the development of the guidelines may or may not be provided. As a result, a guideline developed well may also get a low score. Third, the AGREE II tool provides the overall scores of the developmental process of guidelines. However, a poorly developed guideline may also provide a reliable recommendation. A recommendation of low quality may also come from a high score guideline. Fourth, considering the simplicity and readability of the recommendation table, we did not include the level of evidence for each recommendation in it, though it is important in the comparison of recommendations from different guidelines. Finally, it is difficult to comprehensively evaluate the potential conflict of interest simply by reviewing the guidelines.

| CONCLUSION
Quality of guidelines for DAPT in CAD was relatively high, though defects existed in "Applicability" and "Stakeholder Involvement." As these guidelines developed, DAPT recommendations gradually converged on a consensus. Clinical decision should be made on an individual basis.

CONFLICT OF INTEREST
The authors declare no potential conflict of interests.

AUTHOR CONTRIBUTIONS
Zhimin Du, Xinxue Liao and Shaozhao Zhang devised the study.