A clinical practice guideline appraisal for appropriate use of echocardiography in adult infective endocarditis—when and by which mode to perform an Echo?

Abstract


Abstract
Background Echocardiography (Echo) is the primary imaging modality of infective endocarditis (IE). However, the recommendations on timing and mode selection for transesophageal echocardiography (TEE) and transthoracic echocardiography (TTE) are not fully in agreement among different guidelines, which can be confusing for clinical decision makers. In this case, we aim to appraise the quality of recommendations by appraising the quality of guidelines.

Methods
A search of guidelines containing recommendations for the appropriate use of Echo in IE adult patients published in English between 2007 and 2020 has been conducted. APPRAISAL OF GUIDELINES FOR RESEARCH & EVALUATION II (AGREE II) instrument is applied independently by two reviewers to assess the integrated quality of identi ed guidelines. The recommendations of concern are extracted from related chapters.

Results
A total of 9 guidelines meet the criteria with AGREE II score ranging from 36% to 79%, and the domain of "stakeholder involvement" receives the lowest score. The most contentious issue is, whether a follow-up TEE is mandatory in uncomplicated native valve IE with initial positive TTE. And con icting recommendations on it are presented with low evidence level for hardly any related evidence based on.

Conclusions
In general, the recommendations proposed in the 9 identi ed guidelines on the appropriate use of Echo are satisfying. And the guideline quality score could be taken into account by the clinicians when evaluating the recommendations for making clinical decision. More researches are needed in the future, which should be with high evidence level on the most controversial issues of whether a subsequent TEE is mandatory in uncomplicated native valve IE with an initial positive TTE. Background Echocardiography (Echo) is the primary imaging modality for infective endocarditis. A positive Echo de ned as vegetation or abscess, or new dehiscence of prosthetic valve is included as a major modi ed Duke criterion along with positive microorganisms [1]. Besides, echocardiographic examination is also available to show the hemodynamics and mechanism severity [2]. The appropriate use of Echo should be cost-effectiveness, time-e ciency and concern about the potential associated complications [3,4]. In the meantime, it should be available for timely and accurate guidance for the diagnosis and management of IE. As per a single-center study, a considerable number of transesophageal echocardiography (TEE) studies were rated as rarely appropriate tests for infective endocarditis (IE) [5]. Over the past 15 years, there are 9 guidelines that cover IE diagnosis and contain evidence-based recommendations on the appropriate use of Echo (only in English language, and the latest version of updating guidelines).
However, these guidelines differ in the recommendations on TEE and transthoracic echocardiography (TTE) application in certain clinical scenarios, which can be confusing for clinical decision makers.
AGREE is an instrument for comprehensive guideline evaluation from 6 different domains, including scope and purpose, stakeholder involvement, rigour of development, clarity of presentation, applicability and editorial independence. And AGREE II, as the updated version developed in 2013, carries some modi cations on certain items compared with the original version [6]. A systematic review of the appropriate use of Echo based on critical assessment and the quality comparison of different guidelines was presented to allow clinicians to make better decisions in certain confusing clinical scenarios.

Searching process
A literature search of current clinical practice guidelines which contain recommendations on IE imaging examination was conducted on Pubmed, Embase, Web of science and websites of guideline development societies (websites of organizations and societies is shown in Table S1). "adult" "infective endocarditis" "echocardiography" "transesophageal echocardiography" and "transthoracic echocardiography" were searched either singly or in combination (the detailed search strategy of databases is shown in table S2).
The search covered the guidelines published in English from 5 June, 2007 to 5 July, 2019.

Including criteria
An included guideline needs to meet the following criteria: 1. The de nition of a guideline given by The Institution of Medicine [7] should be followed, which was described as "systematically developed statements to assist practitioner and patient decisions on appropriate healthcare for speci c clinical circumstance". Titles and abstracts were reviewed by 2 reviewers independently. After that, the disagreements were discussed before reaching a consensus with the presentation of the third party. At last, the nal selection of guidelines was performed by the 2 reviewers together.
Guideline appraisal and recommendation extraction A comprehensive evaluation was performed to the 9 selected guidelines from 6 domains as per AGREE II instrument, which includes: i) scope and purpose, ii) stakeholder involvement, iii) rigor of development, iv) clarity of presentation, v) applicability, and vi) editorial independence. Two reviewers evaluated 23 items from 6 domains independently in the form of scoring with numbers ranging from 1 to 7, where 1 represents the strong disagreement to which no relevant information was given or the concept was barely reported, and 7 represents the strong agreement of which the reporting quality was exceptional or fully met the criteria set by AGREE II. The nal score of each domain was obtained by the calculation formula given by AGREE II: the scaled domain score = (Obtained score -Minimum possible score) / (Maximum possible score -Minimum possible score). In the case that the difference between the scores given by both reviewers for a certain item is greater than 20% of the lower score, a third reviewer would be introduced for the review and evaluation of the guidelines [8]. The guidelines with the average score higher than 60% was marked with "Strongly recommended", while those with scores ranging from 30% to 60% will be marked with "Recommended with some modi cation", the others with scores lower than 30% will be marked with "Not recommended" [9].
Except for the two guidelines of Chinese Society of Cardiology, version 2015 and Swedish Society of Infectious Diseases, version 2007 (CSC 2015, SSID 2007) without the report of Con icts of Interest (COI) [10,11], we calculated the proportion of panel members with an industrial relationship (RWI) of authors reported in other guidelines, and analyzed the correlation between RWI ratio and AGREE II score using SPSS 25.0. The case with α 0.05 indicates the statistical signi cance [8].
All recommendations on the appropriate use of Echo, including the timing and mode (TTE and TEE) were extracted from the relevant chapters. In an effort to avoid phrasing confusion, the class of recommendation and the level of evidence quoted from SSID 2007 were converted into a uni ed form that is consistent with those used in other guidelines [11]. The grade of recommendation and the level of evidence denoted hereafter were uniformly expressed as I/II/III and A/B/C respectively.

Results
Guidelines meet the criteria A total of 1015 records were searched out during the literature search, from which 1006 records were removed after the review on title, abstract, and the full-text ( Figure 1). As the outcome, 9 guidelines  Table S3. Guideline identi er, host organization, region, average AGREE II score, COI, number of Echo recommendations and RWI are summarized in Table 1.

Guidelines appraisal by AGREE II
The scores of each guideline are presented in the radar charts ( Figure 2) and Table S4. The AGREE II scores of all guidelines are ranging from 36% to 79% with the median of 55%, among them, 4 guidelines ( [10,11,[16][17][18] are marked with "Recommended with some modi cations" with the score from 36% to 55%. No guideline is rated as "Not recommended" with the score lower than 30%. There is no item with the difference between the scores given by the two reviewers greater than 20%. The focus of Domain1 (scope and purpose) includes the items about the overall goal, the speci c health issues contained in the guideline, and its target population. While in contrast with other domains, guidelines' performance in this domain is uneventful, that is, most of them generally summarized the concerning issues without any further elaboration. NHAM 2017, BSAC/BHRS 2014, BSAC 2011[12, 15,17] speci cally expounded on the items following the AGREE II rules, thus receiving a relatively high score.
The concern of Domain2 (stakeholder involvement) includes that whether the guideline was developed by appropriate stakeholders and whether the views of its intended users was considered. This domain received the lowest average score when two-thirds of the guidelines are not higher than 33%. Only NHAM 2017, BSAC/BHRS 2014, and JCS 2017[12, 15,16] speci ed that to which department the clinical guideline should apply for the specialists. In addition, the view of the patients as the target population was only reported by BSAC/BHRS 2014 [15] in the external review, which was not mentioned by the rest of the guidelines of at all. Domain 3 (rigour of development) is comprised of the processes of evidence synthesizing, the formulation of the recommendations and the procedure for updating. CSC 2015 and SSID 2007 [10,11] reported no information in regard to the selection of evidence, while BSAC/BHRS 2014, JCS 2017 and BSAC 2011 [15][16][17] had only a little description on this subject. Besides, no updated statements or detailed information on external expert review are presented by the guidelines other than AHA 2015, ESC 2015 [13,14] and NHAM 2017 [12]. The narrative of methods for formulating the recommendations is relatively clear, therefore, higher scores were given to the items concerned.
Domain 4 (clarity of presentation) involves the clarity of the description and the format of the guideline, of which the average score is the highest of all domains, and the discreteness of scores is the smallest. implication. Yet the monitoring and auditing criteria was set precisely by all guidelines with the average score higher than 80%. Domain 6 (editorial independence) pertains to the transparency declarations, including the funding body and competing interests of guideline development group members. Neither clari cation on the funding body nor its in uence on the content was mentioned by JCS 2017, CSC 2015 and SSID 2007 [10,11,16].
Besides, no records on the disclosure of potential COI were mentioned by SSID 2007[11] and CSC 2015 [10]. As a result, there is no correlation between RWI and AGGEE II score (Pearson's correlation r =-0.081 P =0.863) for the rest of the guidelines.
It was found out that more than half of the guidelines (5 out of 9) had received scores less than 60%. According to the independent-sample t-test, t'-test (editorial independence) and Wilcoxon Rank-Sum test (clarify of presentation), the guidelines marked with "recommended with modi cation" are statistically different from those with "strongly recommended" in the domains of "rigour of development" (P=0.015), "applicability" (P=0.005) and "editorial independence" (P=0.024) (Figure 3). With regards to the speci c items, 8 out of 9 guidelines had received the score of zero on the item of "target population", which indicates that the existing guidelines generally failed to consider the view of a patient during the formulation of recommendations.

Recommendations on appropriate use of Echo
Recommendations for different clinical scenarios were organized and listed in Table S5 (recommendations with controversies) and Table S6 (recommendations without controversies). As shown in the tables, both consensus and controversy are listed, and the consensus outweighs the controversy.
As per the 5 out of 9 guidelines, when a rst-line TTE is proved to be non-diagnostic due to its poor echocardiographic window, a further TEE will be recommended considering its higher sensitivity than TTE (class of recommendation: I, level of evidence: B-C) [1].
For patients with suspected IE under a prosthetic heart valve/intracardiac device, TEE is recommended by 5 out of 9 guidelines (class of recommendation: I, level of evidence: B-C). Furthermore, suggestion for patients with implantable cardiac electronic device (ICED) was given by the guideline of BSAC/BHRS 2014, recommending that Echo should be conducted for patients with implantable cardiac electronic device lead infection (LCED-LI), implantable cardiac electronic device associated native or prosthetic valve endocarditis (LCED-IE) and suspected generator pocked infection concurrent ICED-LI or ICED-IE (level of evidence: B-C) [15]. An intraoperative Echo examination for IE patients requiring surgery is recommended by 3 out of 9 guidelines (class of recommendation: I, level of evidence: B-C), and brought up with no formal recommendation by BASC 2011 [17]. Moreover, BASC/BHRS 2014 mentioned that patients after IECD removal also need follow-up Echo to identify persisting vegetations (level of evidence: C) [15].
At the completion of antibiotic therapy, TEE is recommended by 6 out of 9 guidelines (class of recommendation: I-IIa, level of evidence: C).
The controversial processing steps in different guidelines' algorithms were shown in Figure 4, and it can be seen that 7 out of 9 guidelines include the recommendation on the rst-line modality of suspected IE. In the case that an initial TEE showed a negative result, as per the recommendation by 6 out of 9 guidelines, a subsequent TEE should be conducted within a given time limit stipulated by different guidelines when suspicion exists without diagnosis of IE (class of recommendation: I, level of evidence: B-C). However, the maximum time limit given by different guidelines varied from 5 to 10 days.
Regarding whether a TEE is needed for suspected IE with a positive TTE, guidelines' replies are based on varied target population features. For example, BASC 2011 suggested that a positive TTE should be considered as the indication for a subsequent TEE (no formal recommendation formulated) [17]. While

Discussion
It has been the rst time that the controversies among recommendations on the appropriate use of Echo were evaluated combined with AGREE II score of guidelines.
In the 9 guidelines, NHAM 2017 and BHRS 2014 mentioned in the methodology part that the entire content was compiled based on the AGREE II principle, and both of them received a score higher than 60% [12,15]. However, compared with other guidelines which failed to involve some certain AGREE II items and thus got deducted scores, the superior score of NHAM 2017 largely owes to a more general coverage of items rather than a higher grade of each item. The description referred to Item 19 (facilitators and barriers to the application) and Item 20 (the potential resource) in NHAM 2017 is rather perfunctory without any in-depth content. Though AGREE II items are encouraged to get generally involved during developing guidelines, there is still criticism saying that comprehensive coverage of quantity is just as important as quality.
It should also be noted that, some studies showed that the exposure to information provided directly by pharmaceutical companies had been found to be associated with higher prescribing frequency, higher costs and lower prescribing quality. Therefore, disclosure of Potential COI could be very necessary [19,20]. Yet it is found that there is no correlation between the proportion of RWI and the AGREE II score involved in the study. This could happen for a variety of reasons, i.e., some guidelines were found to have a high RWI thanks to a more thorough and accomplished disclosure process and less underreporting. For example, ESC has a very detailed COI appendix [14]. On the contrary, it is only mentioned in NHAM 2017 that there is no potential con ict of interest to disclose, and no further detailed relevant content to present. In that case, it cannot be ruled out that the actual RWI proportion of its guideline committee members was concealed and underreported.
Based on the comparison of recommendations and viewpoints of different guidelines, it is found that the algorithms of echocardiography usage given by these guidelines are roughly the same in most cases.
Differentiated recommendations are mainly given on more speci c clinical scenarios, which are not con ict to each other, but a complement.
As is shown in Figure 4, there are also 3 issues for con icting views.
The rst issue relates to the rst-line modality of suspected IE. It was found in a cost-effectiveness analysis that the initial use of TEE was the optimal diagnostic strategy for most suspected patients [21].
SSID 2007 recommended TEE as the rst choice for vegetations and complications due to its cost e ciency and higher sensitivity [11], while TTE was recommended by all other guidelines. However, these guidelines recommended TTE for different reasons. AHA 2015 proposed that although TEE was a better choice with higher sensitivity, it was not always available immediately (since a patient did not fast for the preceding 6 hours or the medical institution did not have 24-hour TEE service), therefore a TTE is recommended to be conducted as soon as possible, nonetheless it remains a suboptimal alternative [13]. In addition, there are also other guidelines, like JCS 2017, which recommended TTE for the evaluation of valve dysfunction and hemodynamics based on its value as well as its feature of non-invasiveness and repeatability [16]. Actually, with the technological progress of echocardiography, TTE was proved to have a su cient negative predictive value of NVE for in-patients with low to intermediate risk when strict negative criteria are applied [22,23]. For these low risk patients strati ed based on clinical judgment with negative TTE, although few recommendations was made, guidelines proposed in the context or in the algorithm that a TEE was not required [10-14, 16, 17, 24], and this have been veri ed by the recently published Meta-analyses [25,26]. In this case, the claim of SSID 2007 is outdated, which says that TEE is better and more cost-effective than TTE as the rst-line examination. and SSID was blamed for recommending overuse of TEE as well as citing studies based on obsolete echocardiography methods and lack of hand-on knowledge about echo [27].
For a rst-line negative TTE with undiagnosed but high suspicion of IE, a follow-up TEE is fully agreed. However, the maximum time limits given by different guidelines vary from 5 to 7 days. It is known that the severity of pathology distinguished by echocardiography is helpful for the determination of the following management strategy [28]. As per the research, early (<4 days) de nitive echocardiography is associated with less embolic events than the later one does [29]. Therefore, time delays to diagnostic Echo should be avoided.  [14,16,17]. Therefore, for the patients (with the exception of isolated-right-sided NVE) with no or low risk of complications, which are not uncommon in clinical practice, it is confusing whether TEE examination is necessary. On the one hand, TEE is helpful in evaluating the presence of intracardiac complications, offering prognostic information and helping developing treatment plan. On the other hand, in the case that an initial TTE presents the vegetation clearly, and the probability of complications is low (presented as a small aortic vegetation, mild aortic regurgitation, and normal left ventricular size and function), a subsequent TEE then seems to make no incremental value for the treatment strategy [25], which may lead to the overuse of TEE. However, the recommendation that suggested a subsequent TEE over a positive TTE seems to be less evidence-based and convincing with a recommendation class of IIa which indicates that weight of evidence and opinion is in favor of usefulness and/or effectiveness, and an evidence level of C which refers to only consensus of opinion. Similarly, the evidence strength given by SSID 2007 which explicitly recommended that there was no need to repeat TEE is of C III, which manifests that the evidence to support the recommendation is weak. Therefore, it can be concluded that neither side has su cient evidence for this issue, and more researches with high evidence level are needed in the future. Besides, for the following reasons, the isolated right-side NVE was excluded from the adapted population recommended by NHAM 2017, SSID 2007 and AHA 2015 who requires a subsequent TEE examination, i.e., the right-sided structure is located anteriorly, which is closer to the TTE transducer than transoesophageal transducer, and hence allows TTE to offer more valuable information. And it was also because that a subsequent TEE was found to have no additional value for providing new information in isolated right-sided NVE patients [30].

Conclusion
According to AGREE II and compared with guidelines marked with "recommended with modi cation", those guidelines marked with "strongly recommended" was more rigorously developed, the recommendations involved in these guidelines could be identi ed to have better quality. Clinicians could rely on the guideline quality score when evaluating the recommendations for clinical decision making. Besides, it is also expected that later updating of guidelines will lead to a better performance in the domain of "stakeholder involvement". The main controversy arose from the views of the SSID 2007 with the lowest AGREE II score among all the guidelines may be mainly due to its outdated concepts. For the most contentious issue, i.e., whether a subsequent TEE is mandatory in uncomplicated NVE with initial positive TTE, related studies are rarely seen. Therefore, more researches on this issue, especially those with high evidence level are desired in the future.

Declarations
Funding This study was supported by the National Natural Science Foundation of China (81600206 to ZXD; 81870195 to LXX), and Natural Science Foundation of Guangdong Province (2016A030310140 to ZXD;20160903 to LXX; 20160904 to LDH). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Availability of data and materials
All data generated or analysed during this study are included in this published article and supplementary information les.
Ethics approval and consent to participate Not applicable.

Consent for publication
Not applicable.