Quality assessment of clinical guidelines for the treatment of obesity in adults : application of the AGREE II instrument

There are various guidelines for the treatment of obesity, and thus the quality of these clinical guidelines has become a matter of concern. The objective was to describe and assess the quality of clinical guidelines for treatment of obesity in adults. We collected several studies, dated from 1998 to 2016, produced by different countries. The literature search included the National Guideline Clearinghouse (NGC), Guidelines International Network (GIN), PubMed (MEDLINE), Scopus, Web of Science, webpages of health institutions from different countries, and search sites, with the criterion: “clinical guidelines for treatment of obesity in adults and published until the 2016”. The guidelines were assessed with the Appraisal of Guidelines for Research & Evaluation (AGREE II), according to the domains of the instrument. The search identified 21 guidelines: nine from Europe, six from North America, three from Latin America, and one each from Asia and Oceania and a transnational association. The Australian guideline had the best assessment. Of the six guidelines with the highest scores, five had been elaborated by the government sector responsible for the country’s health. The domains “scope and purpose” and “clarity of presentation” had the highest score. Except for the Canadian guideline, the three guidelines drafted before the elaboration of AGREE II had the worst quality. In the domain “stakeholder involvement”, only four guidelines (Australia, Scotland, France, and England) mentioned patient participation. Guideline development and quality enhancement are ongoing processes requiring systematic appraisal of the guideline production process and existing guidelines. Obesity; Practice Guidelines as Topic; Health Technology Assessment Correspondence E. C. Reis Escola Nacional de Saúde Pública Sergio Arouca, Fundação Oswaldo Cruz. Rua Leopoldo Bulhões 1480, Rio de Janeiro, RJ 21041-210, Brasil. erika.careis@gmail.com 1 Escola Nacional de Saúde Pública Sergio Arouca, Fundação Oswaldo Cruz, Rio de Janeiro, Brasil. 2 Instituto Nacional de Infectologia Evandro Chagas, Fundação Oswaldo Cruz, Rio de Janeiro, Brasil. doi: 10.1590/0102-311X00050517 Cad. Saúde Pública 2018; 34(6):e00050517 ARTIGO ARTICLE This article is published in Open Access under the Creative Commons Attribution license, which allows use, distribution, and reproduction in any medium, without restrictions, as long as the original work is correctly cited.


Introduction
Obesity is a chronic, multifactorial disease, that has been increasing at alarming rates worldwide.A recent analysis on trends in adult body mass index (BMI), based on data from 200 countries, showed that between 1975 and 2014 the age-standardized global mean BMI in men increased from 21.7 to 24.2kg/m 2 .In the same time interval, age-standardized prevalence of obesity (BMI ≥ 30kg/m 2 ) increased from 3.2% to 10.8% in men and from 6.4% to 14.9% in women 1 .This rapidly growing prevalence in recent decades has turned obesity into a serious public health problem in both developed and developing countries, and currently obesity is considered a global epidemic.
The impact of obesity on quality of life has been widely discussed.Obesity is also an important risk factor for other chronic non-communicable diseases (NCDs) such as diabetes, arterial hypertension, musculoskeletal disorders (especially osteoarthritis), and some types of cancer (mainly endometrial, breast, and colorectal) 2 .
According to the Brazilian National Health Survey, 36.1% of Brazil's adult population are overweight (BMI defined between 25kg/m 2 and 30kg/m 2 ), of which 20.8% are obese (BMI greater than 30kg/m 2 ) 3 .Given this scenario, the Brazilian Unified National Health System (SUS) has endeavored to organize care pathways for obesity since 2013 4 .Care pathways are organizational arrangements of health interventions and health services that seek to rationally incorporate the best available evidence on treatment protocols, guidelines and multidisciplinary clinical practice 5 .
Literature emphasizes the need for up-to-date, concise, and accessible information produced by reliable sources to orient health practices 6 .Clinical guidelines are defined by the Institute of Medicine as structured recommendations developed to orient health professionals on appropriate health care in specific clinical circumstances 7 .They include treatment indications and contraindications, as well the expected benefits and risks from the use of health technologies (for example: procedures, diagnostic tests, medicines) for specific patient groups.
Clinical guidelines based on the best available scientific evidence aim to optimize health professionals' decisions, and play a key role in health systems management and regulation 7,8,9,10 .Health professionals, administrators, and funders view guidelines as tools that can reduce the gap between professionals' routine clinical work and orientation from scientific evidence on such practice 10,11 .
Taba et al. 12 found that physicians do not consider the use of clinical guidelines as a hindrance to their professional autonomy, which theoretically facilitates their adherence to (and use of) such guidelines.Good-quality guidelines can provide less uncertainty in clinical decision-making and support health professionals in their health education process and in dealing with questions on their clinical practice 13 .
The increase in the production of clinical guidelines in recent years raises the concern that many clinical guidelines may not have the requirements to be classified as such, due to their low quality.This creates the need to internationally define acknowledged criteria for their assessment 14 .This has led to standardized instruments for assessing the development and validity of clinical guidelines 15,16,17,18 , including the Appraisal of Guidelines for Research & Evaluation (AGREE II) instrument 16 .
The AGREE II 19 has been gradually incorporated in Brazil in the last two years, having been applied to guidelines produced by the Brazilian Ministry of Health 20 , in the assessment of a set of Brazilian clinical guidelines for NCDs 21 , and supporting the production of clinical guidelines by medical experts 22,23 .
Given the diversity of published guidelines and the uncertainty surrounding their quality in a scenario of growing demand for the implementation of care pathways for obesity in the SUS, this study aims the assessment of the quality of clinical guidelines for obesity in adults, produced worldwide from 1998 to 2016.

Methods
This is a review of the quality of clinical guidelines for the treatment of obesity according to the AGREE II criteria 19 .A search was conducted for clinical guidelines for the treatment of obesity in adults, published from 1998 to 2016, in the National Guideline Clearinghouse (NGC), Guidelines Cad.Saúde Pública 2018; 34(6):e00050517 International Network (GIN), PubMed (MEDLINE), Scopus, Web of Science, and websites of health institutions and medical specialty societies.
To identify the clinical guidelines in the database indexes (PubMed, Scopus, and Web of Science), we used the following search strategies: (1) for PubMed -title (practice guideline) and MeSH terms (obesity); publication type (practice guideline) and MeSH terms (obesity); (2) for Scopus -title (practice guideline) and title (obesity); title (clinical practice guideline) and title (obesity); (3) for Web of Science -title (practice guideline) and title (obesity); title (clinical practice guideline) and title (obesity).The term "obesity" was used in the GIN website.In the NGC website, the term "obesity" was used and the target population characteristics/age selected were: adult (19-44 years) and middle age (45-64  years).
Inclusion criteria were: (1) clinical guidelines for obesity; (2) treatment recommendations for adults; (3) elaboration by a government agency and/or state-level societies of health professionals, because this allows greater comparability between guidelines in terms of scope and resources; and (4) studies published from 1998 to 2016.Exclusion criteria were: (1) clinical guidelines dealing exclusively with obesity associated with another health condition; (2) clinical guidelines dealing exclusively with obesity during life phases other than adulthood; and (3) non-inclusion of treatment recommendations for adults.There was no limitation on any language and the Google Translator tool was used when necessary to translate a guideline into English or Portuguese.
A total of 746 records were identified, of which 725 were excluded for not fitting into the inclusion criteria (Figure 1).Twenty-one clinical guidelines were included and assessed by a pair of appraisers/authors: a nutritionist (E.C.R.) and one of two physicians (S.R.L.P. or M.A.B.S.).The quality of the guidelines was assessed with the translated version of the AGREE II instrument, cross-culturally adapted to Brazilian Portuguese 24 .
AGREE II aims to analyze the methodological rigor of the development of clinical guidelines and orientation on information they are expected to contain 19 .This instrument has 23 items, covering six domains of health care quality, consisting of items scored on a Likert scale from one to seven, in which one is "strongly disagree" and seven is "strongly agree".The user's manual 19 provides guidance on how to score each item using the rating scale.Each appraiser scores the items of a domain independently, according to the 7-point scale (1 -strongly disagree to 7 -strongly agree), and the scores are not subjected to consensus.Scores of a domain are calculated by summing up the scores of the individual items in this domain and the total is calculated as a percentage of the maximum possible score for that domain.Thus, the maximum possible score for each domain is rated on a 7-point scale using the scores X number of items in the domain X number of appraisers.Domain (1), "scope and purpose", assesses whether the guideline is clear in its overall objective, in the specific health issues to which it refers, and its target population (patients and public).It has three items (1-3) and maximum score per appraiser is 21.Domain (2), "stakeholder involvement", assesses whether the guideline clearly defines the target population and participation of the stakeholders in its development, including different health professionals and patient groups in its elaboration.It also assesses whether the guideline reflects the views of the expected users and target populations.This domain has three items (4-6) and maximum score per appraiser is 21.Domain (3), "rigor of development" assesses the process of collection and synthesis of the evidence and the methods used to formulate the recommendations.It requires a clear description of the strengths and limitations of the body of evidence, consideration of the benefits, side effects, health risks in the formulation of the recommendations, the procedure for updating the guidelines, and external expert review.This domain has eight items (7-14) and maximum score per appraiser is 56.Domain (4), "clarity of presentation", assesses whether the different options for approaching the health condition or problem are clearly presented, and more specifically whether the recommendations are clear, unambiguous, and easily identifiable.This domain has three items (15-17) and maximum score per appraiser is 21.Domain (5), "applicability", discusses facilitators and barriers for implementation of recommendations, strategies to improve application, and the necessary resources for such.This domain has four items (18-21) and maximum score per appraiser is 32.
Cad. Saúde Pública 2018; 34(6):e00050517 Domain (6), "editorial independence", assesses potential conflicts of interest of professionals drafting the guideline, which might lead to biases in the recommendations.This domain has two items (22-23) and maximum score per appraiser is 14.
AGREE II sets no standard scores for the domains to distinguish between high-and low-quality guidelines, which would help to define their recommendation or non-recommendation.This decision should be made by the evaluators and oriented by the context in which AGREE II is being used 19 .According to the AGREE II User's Manual, the overall assessment of a guideline requires the appraiser to judge its quality.
Therefore, our cutoff for recommending a guideline was established by the appraisers.It considered the overall score attributed by appraisers to each guideline and a cutoff of recommendation closer to the highest possible score than to the lowest possible score, as follows: recommended (6.0-7.0),recommended with changes (4.0-5.9) and not recommended (≤ 3.9).

Table 1
Main treatment recommendations in the best scoring-guidelines for adult obesity.

Dietary interventions
Reduce caloric intake (deficit of 500Kcal-1,000Kcal) Reduce caloric intake by restricting food consumption (caloric intake and type of food not specified)

Physical activity
Up to 150 minutes of physical exercise/week - Gastric and pancreatic inhibitors-Orlistat associated with lifestyle interventions In cases of treatment failure with other therapeutic strategies (diet, exercise, psychotherapy) In cases of treatment failure with other therapeutic strategies (diet, exercise, psychotherapy, pharmacotherapy) Discuss potential risks and benefits with patient BMI: body mass index; ICSI: Institute for Clinical Systems Improvement.
In addition to assessing the quality of guidelines according to AGREE II, the analysis was complemented by providing a brief description of the main therapeutic recommendations in the best scoring guidelines.Using the eight best scoring guidelines, the treatment recommendations found in 4 or more of them were selected and summarized.The main recommendations concerning diet, physical activity, psychological support, pharmacotherapy and surgery are shown in Table 1.

Results
Twenty-one clinical guidelines were identified from: Europe (Belgium 25 , Finland 26 , France 27 , England 28 , Germany 29 , Italy 30 , Netherlands 31 , Norway 32 , Scotland 33 ), North America (Canada 34 and USA: National Heart, Lung and Blood Institute -NHLBI 35 , Defense Department 36 , Institute for Clinical Systems Improvement -ICSI 37 , American College of Cardiology 38 and American Association of Clinical Endocrinologists -AACE 39 ), Latin America (Argentina 40 , Brazil 41 , Mexico 42 ), Asia (Malaysia 43 ), Oceania (Australia 44 ) and one from the World Gastroenterology Organisation 45 .No African clinical guideline was found (Table 2).One guideline was included although it represented a global organization and was not related specifically to one country.The database that yielded the most was the GIN.
Table 3 describes the overall assessment and scores of guidelines as shares of maximum total score per individual domain of the AGREE II.

Stakeholder involvement
Its median score was 72 % and the IQR was from 49.3% to 97%.Seven guidelines included no representatives of the users' group 26,29,30,38,39,41,43 in the elaboration of guidelines or review team.In addition, drafting teams for other guidelines consisted mainly of physicians 38,39,41 .

Rigor of development
The overall score for this domain was relatively high: the median was 78% and the IQR was from 48% to 96.3%.The selection and classification of scientific evidence that provided the basis for each recommendation of guidelines varied considerably.The process of evidence selection and appraisal was often unclear.Some guidelines specifically mentioned the use of questions in the PICO format (population, intervention, comparator, outcome) 36,40,44 to orient the search for evidence.Others used reviews of clinical guidelines, produced by other countries or organizations, as the preliminary stage in drafting their own guidelines 29,40,44 or as part of the evidence compiled in the guideline 42 .
Differences were observed between classification systems for the quality of evidence and the strategies for generating and scoring the strength of recommendation.The Australian guideline, for example, showed the criteria separately for level of evidence (quality of studies), effect size of treatment (relevance of the clinical benefits in the range of the estimate set by the confidence intervals), and estimation of certainty.Based on this, the elements were summarized as a matrix for the body of evidence to support the strength of recommendation 44 .
Meanwhile, some guidelines did not describe the connection between the recommendations and their supporting evidence, showing a summary of recommendations without citing degrees of evidence or the strength of recommendation, or being clear on the process used by guideline developers to make decisions 28,32 .The English guideline provided attachments with details on the method for search and selection of evidence, and had high scores in the assessment using AGREE II.Meanwhile, it showed the interventions for obesity as "recommended" and "not recommended", without specifying the degrees of recommendation or levels of evidence 28 .Domain 3 also involved simple issues such as a clear description of search terms and strategies and procedures for updating the guideline, which, curiously, were not made explicit in most guidelines analyzed here.Such procedures and updating timeframes were only described by the German, Argentine, Australian, American (U.S.Department of Defense), French and Scottish guidelines 27,29,33,36,40,44 .
Key recommendations tended to be easily identified in most guidelines.Some guidelines showed their recommendations in a summary of easy access for health professionals 28,33,35,38,40,44 .
However, clinical guidelines showed no clear pattern in their content and format.In terms of content, the highest-quality guidelines for obesity took greater care with the contextualization, characterization, epidemiology, diagnostic approaches, and classification of obesity.Nearly all the guidelines had the recommended interventions divided into nutritional/dietetic, behavioral or cognitive, physical activity-related, pharmacological, and surgical.
The formats for presentation also differ.Some guidelines were more explicit in their inclusion of issues that had oriented the recent focus on the dissemination of health information, providing recommendations in various formats to adjust to the reality of distinct audiences.Algorithms with recommendations for obesity management were a key characteristic of the guidelines of studies from USA, Italy, England, Mexico and Norway 28,30,32,35,36,37,38,39,42 , but there were also guidelines Cad.Saúde Pública 2018; 34(6):e00050517 that provided charts with "practical tips" 33,40,42 or detailed attachments with information on the search for evidence 28,29,38,40,42,44 , for audiences with more available time and the need for more complete information.

Applicability
This was the domain with the widest range of scores (0% to 100%), the lowest median (52%) and IQR (27.5% to 69%).The Australian guidelines scored the highest 44 , while Malaysian, Finnish, Brazilian and Italian guidelines scored the lowest 26,30,41,43 .The overall score of this domain reveals an insufficient perception of the importance of including recommendations for the implementation of guidelines.Only the Australian guidelines reached 100% of the score as assessed by AGREE II, followed by the English one, with 96% 28,44 .
Most guidelines failed to suggest implementation tools for recommendations.One exception was the guideline from the ICSI 37 , in which implementation strategies were an explicit objective.This guideline had special attention to the implementation issue, including the recommendation of indicators and measures for monitoring obesity in the population.
The Argentine guideline was one of the few that proposed a strategic planning tool to increase the opportunities for its use, such as the Guideline Implementation Appraisal (GLIA) 40 .
Another important issue that received little attention in this domain considers the resources required to apply the recommendations.Lack of resources can pose an obvious barrier to the applicability of recommendations.Only the guidelines from Australia 44 , England 28 , and Germany 29 explicitly addressed this issue.A single guideline, from the World Gastroenterology Organisation 45 , linked the recommended interventions to the availability of resources according to the country's income level (high, medium, and low income), although this link was schematic.The main impact of this resource gradient was seen in the recommendation for pharmacological and bariatric treatment.For high-income countries, the guidelines recommended bariatric surgery for patients with BMI from 30 to 35kg/m 2 , compared to the recommendation in low-income countries, where the cutoff is BMI > 40kg/m 2 .

Editorial independence
Domain 6 also showed wide variation between the guidelines analyzed, IQR from 35.8% to 100% and median of 88%.The guidelines that scored highest in this domain were the Argentine, Australian, American (ICSI), Belgian, Mexican and French 25,27,37,40,40,44 .We highlight that the guidelines that failed to report the existence or absence of conflicts of interest and possible influences from funding agencies or industries also had the lowest score in this domain 30,32,36,41,43 .
The mere description of presence or absence of conflict of interest does not automatically increase the score of the guideline when its drafting committee includes members affiliated with industry or who have an economic interest in the sale of products or medicines for obesity.This item assesses whether the drafting committee made any attempt to minimize the influence of conflicts of interest in the drafting process of the guideline or in the formulation of recommendations.

Therapeutic recommendations in the best scoring guidelines
A synthesis of the main treatment approaches recommended in the best ranked guidelines 27,28,29,37,40,42,44 is shown in Table 1.Dietary recommendations emphasize reducing intake of food and fat by 500-1,000 calories/day.In addition, physical activity, varying from a minimum of 150 minutes a week to 300 minutes/week, is widely recommended.
The best scoring guidelines uniformly recommended psychological follow-up based on cognitivebehavioral therapy.Recommendations concerning pharmacotherapy in most guidelines restricted its use to patients with BMI exceeding 30kg/m 2 and to the drug Orlistat, a gastric and pancreatic lipase inhibitor.Bariatric surgery is recommended for BMI > 40kg/m 2 for patients in whom other treatment approaches failed.France provided no recommendations related to these last two types of interventions.

Discussion
Most countries included in this study are in the Organization for Economic Co-operation and Development (OECD) where average prevalence of self-reported obesity was 18.4% (2013), with a 27% increase from 2000 to 2013.Italy (10%), Norway (10.3%),Netherlands (11.1%),Belgium (13.7%) and France (14.5%) showed rates below the OECD average.The French growth rate of obesity of 61% in 13 years is an alarming matter of concern.USA (35.3%) and Mexico (32.4%) showed the highest levels of self-reported obesity, whereas Germany, Finland, England and Canada had obesity rates ranging from 23.6% to 25.8% 46 .Australia outstood due to its high rate of self-report obesity at 28.3% in 2013 and rapid growth of this condition (43% from 1995 to 2011), as is the case of many countries of the Asia-Pacific region.For Malaysia, however, obesity rates registered in 2014 were 10.6% (males) and 16% (females) 47 .
Curiously, countries with lower prevalence of obesity (Italy, Norway and Malaysia) had the worst ranks and older guidelines, pointing to a possible lower prioritization of the condition compared to other countries selected in this study.Additionally, for USA, with the highest prevalence of obesity, five guidelines were found.
As observed in previously published analyses of guidelines 17,18,48 , between one fourth and one third failed to achieve the minimum quality standards.In our study, the guideline with the best assessment was the Australian, which received the highest score in all the domains.Except for the guideline from Canada (country where the AGREE instrument was developed), the three guidelines drafted before the AGREE tended to show worse quality than those drafted since the instrument was published 19 , suggesting that it had been used in the development of guidelines.The explicit mention to the use of the AGREE instrument for guideline development was only reported by Argentina, USA (ICSI) and England (National Institute for Health and Clinical Excellence -NICE) 28,40,49 .
Some authors highlighted that guidelines developed by professional associations generally have lower scores than those produced by government agencies 17,48,50 .Of the seven guidelines with the highest scores, five (Australia, England, Argentina, France, and Mexico) were produced by the government agency responsible for health in that country.The German guideline, which had a high score, had a rare characteristic for guidelines, namely the fact that it was drafted jointly by 11 associations and specialty societies, ranging from family physicians to nutritionists, bariatric surgeons, and diabetes and obesity societies.The quality of guidelines may thus reflect the diversity of professional categories involved in the guideline drafting process, emphasizing the fact that obesity is a multifactorial disease that requires a multidisciplinary and multicomponent approach.
Although many of the obesity guidelines were drafted by government agencies, the focus was primarily for interventions in individual health, with limited attention to recommendations linked to public policies in most of the documents.England outstood with an integrated guideline for both clinical practice and public health.It specified that obesity cannot be addressed only at the individual level and highlighted the importance of the environment and public policies.It featured a long list of advices that precede the clinical recommendations, related to the need for involvement by government authorities and "budget holders", training of professionals for the recommended levels of care, and public awareness-raising (including teachers) on healthy lifestyles 28 .
The elaboration of guidelines and the evidence-based medicine movement 51 obviously converge in the incorporation of questions in the PICO format (population, intervention, control, and outcome).This format is used as the basis for gathering evidence in some guidelines and takes into consideration the patients' involvement in the guideline drafting process.The overall score in the "stakeholder involvement" domain suggests difficulty in integrating patients in the discussion of treatment approaches, unlike evidence-based medicine, in which the patient's preferences have played an increasingly central role 51 .
In the overall assessment of guidelines, the domains (1) "scope and purpose" (median of 100%) and (4) "clarity of presentation" (median of 92%) had the highest score.The domains with the lowest scores were (5) "applicability" (median of 52%) and "stakeholder involvement" (median of 72%).
Moreover, the German guideline conducted a review of ten guidelines for obesity 29 , based on the German Instrument for the Methodological Appraisal of Guidelines (DELBI), which is a quality assessment instrument, quite similar to the AGREE II (DELBI incorporates six domains from AGREE II and Cad.Saúde Pública 2018; 34(6):e00050517 includes two additional domains).The study had already shown the potential for improvement in the domains "stakeholder involvement" (DELBI domain 2) and "general applicability" (DELBI domain 5), as well as in the domain "methodological rigor of development" (DELBI domain 3).Although 15 of the guidelines assessed in our study were drafted after those assessed in the German review, there are still some inadequacies in these domains.
Problems related to the lack of adequacy in the domain "methodological rigor of development" confirm findings from studies prior to the creation of AGREE 17,18,52 .A study by Grilli et al. 17 analyzed 431 guidelines developed by professional societies from 1988 to 1998 and found that only 12% provided information on the levels of evidence and grading of recommendations.Some recent studies 53 , a few conducted in countries such as Korea 13 and Brazil 21 , also identified shortcomings in the guideline development process as one of the aspects that most contributed to the low quality of guidelines.
"Applicability" was the domain that had the lowest score among those assessed by AGREE II.By failing to describe tools for their application and possible barriers, many guidelines gave little attention to mechanisms for facilitating their implementation in clinical practice.According to Locatelli et al. 54 , the implementation of clinical guidelines is the only way to ensure that health outcomes will be improved.In Brazil, a study to assess the use of clinical guidelines for arterial hypertension in primary care showed low adherence to guidelines by professionals in the Family Health Program and reiterated the importance of guaranteeing the necessary resources and strategies to implement changes in routine health services 55 .
In addition to quality differences, the guidelines analyzed in this study differed in their treatment recommendations.Treatment recommendations are not assessed by the AGREE II instrument, but the best-ranked guidelines showed relative consistency in this aspect.Considering low-quality guidelines, the recommendations may differ significantly in content and in the direction of strength of evidence, when compared to higher-quality guidelines.
Although recommendations for obesity treatment seemed simple and consensual in the best scoring guidelines, most of them required changes in individual behavioral patterns and lifestyle contexts, which may be very difficult to achieve.Implementation barriers were evident, inasmuch as "applicability" remained the domain with lowest scores.Additionally, despite adverse events related to them, surgical and pharmaceutical interventions emerged as rapid solutions from the patient's perspective, potentially leading to overuse of these approaches.
The potential limitations of the AGREE II instrument and the subjectivity of the score of domains are worth mentioning, even though we identified this issue and conducted the assessment carefully.Also, we did not assess the cutoff points that allowed recommending the guideline ("I recommend this guideline"), which could lead to different choices in terms of recommending or not recommending the guidelines.
Developers of the Argentine obesity guidelines 40 had previously assessed, as part of their guideline development process, 25 obesity guidelines based on the AGREE II.They opted not to establish cutoffs for recommending a guideline, but instead they ranked a group of guidelines as "highly recommended"."Highly recommended" guidelines were those which received a minimum of 60% of highest scores in four domains, necessarily including "rigor in development", and at least 30% in the remaining ones.They thus selected six guidelines, all of which were also included in our study 28,33,34,36,39,40 .The sequence established on ranking guidelines of countries by quality scores was very similar to our results.The consistency between these two independent scoring processes points to the reliability of the AGREE II as a quality assessment instrument.

Final considerations
According to Locatelli et al. 54 , although guidelines are an important instrument for promoting the quality of patient care, they can also be disadvantageous if they neither defend the best treatment options for patients, nor specify the areas in which there are still scientific uncertainties.
Obesity is a global health problem, and various countries have produced clinical guidelines for interventions to address this problem.The quality of the guidelines, as assessed by AGREE II, is Cad.Saúde Pública 2018; 34(6):e00050517 extremely heterogeneous.The Reporting Items for Practice Guidelines in Healthcare (RIGHT) checklist is the most recent addition to instruments dealing with quality problems in practice guidelines 56 .This checklist from 2017 addresses a number of problems pointed out in our study concerning obesity guidelines.Among others, we highlight the requirement for greater emphasis on: a detailed description of the selection process of guideline developers; the statement of the key question in a PICO format; and a clearer description of the rationale for recommendations, especially of approaches used by guideline developers to appraise evidence, make decisions and formulate the recommendations.
Guideline development and quality enhancement are ongoing processes requiring systematic appraisal of the guidelines under production process and existing guidelines.We recommend a wider use of the instruments for guideline development.

Table 2
Clinical guidelines for treatment of obesity included in the study and assessed with the Appraisal of Guidelines for Research & Evaluation II(AGREE II) instrument.

Table 3
Overall assessment of countries' obesity guidelines.Scores are given as percentage (%) of the maximum total score per individual Appraisal of Guidelines for Research & Evaluation II (AGREE II) domain.