Scope, quality, and inclusivity of clinical guidelines produced early in the covid-19 pandemic: rapid review

1In late 2019 a novel coronavirus, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), causing an acute respiratory disease, coronavirus disease 2019 (covid-19), spread from its origins in China to become a pandemic. As of 26 March 2020, 455 770 cases had been identified worldwide, causing 20 740 deaths. No successful therapeutic intervention for covid-19 has yet been established, so supportive care is the most important aspect of clinical management, supporting the patient’s physiology to aid recovery. Optimal provision of supportive care is therefore fundamental both to the wellbeing of individual patients and to securing the confidence of the general population. To enable the provision of best care, clinicians need evidence based recommendations developed using accepted methods. Such clinical guidelines must be readily available, of good quality, and inclusive of vulnerable patient groups.

2Clinical guidelines are defined as “systematically developed statements to assist practitioner and patient decisions about appropriate healthcare for specific clinical circumstances.” Widely agreed, rigorous methods now exist for the production and appraisal of clinical guidelines. The Appraisal of Guidelines for Research and Evaluation (AGREE) II tool is the most widely used guideline appraisal tool, and it has become the international “gold standard” for guideline development.

3During times of crisis, guidelines from the World Health Organization may be the only source of direction available to clinicians globally. They may be adopted internationally with only minor local adaptations. Thus, WHO guidelines must be of the highest possible standard. However, inherent uncertainty exists in the early phase of a pandemic, which, when combined with the considerable pressure to act rapidly, makes the production of gold standard guidelines very challenging. That studies have repeatedly shown WHO guidelines produced in emergencies and in non-emergencies to score poorly in objective appraisals of their methods is therefore not surprising. Often, they do not adhere even to WHO’s internal standard procedures.

4Inclusivity is also vital in a pandemic; covid-19 manifests differently in different patient groups, being most severe among older people and those with comorbidities. Furthermore, the pandemic has now moved to low resource settings, where logistical challenges to a public health emergency are greater. Accordingly, clinical guidelines need to be inclusive of different groups and different resource settings.

5This rapid review aimed to assess the availability, quality, and inclusivity of clinical guidelines produced early in the covid-19 pandemic. To our knowledge, this is the first review of clinical management guidelines produced during a pandemic.

6This study was a rapid review of clinical guidelines for the management of covid-19 produced early in the pandemic. We defined clinical guidelines as systematically developed recommendations produced to direct the management of patients with confirmed or suspected covid-19. To be included, guidelines had to make specific recommendations aimed at the clinical care of patients—for example, concerning fluid resuscitation, oxygen provision, or a therapeutic intervention. We excluded guidelines that exclusively concerned prevention and control of infection or diagnostic studies.

7The study was nested within an extensive systematic review of supportive care in high consequence infectious diseases. That larger study is registered with the PROSPERO international prospective register of systematic reviews (CRD42020167361) and follows Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines on the conduct of systematic reviews (supplementary material). In light of the global covid-19 pandemic, we opted to produce a nested rapid review of guidelines on covid-19 by using a modified protocol for rapid reviews.

8We included guidelines produced by international and national scientific organisations and government and non-governmental organisations relating to global health. We made no exclusions for language. We excluded regional/hospital guidelines to make the search feasible. We included only the earliest version of any guideline.

9We searched the following databases from inception to search date (14 February 2020) for relevant studies: Ovid Medline, Ovid Embase, Ovid Global Health, Scopus, Web of Science Core Collection, and WHO Global Index Medicus. The search strategies applied the CADTH database guidelines search filter to text words and relevant index terms. We applied no limits to the search results. The full search strategies are shown in the supplementary material. We identified further studies through searches constructed using Google Scholar and the PROSPERO database of registered systematic reviews. We augmented this with an extensive grey literature search that continued until 14 March 2020. We requested guidelines from the Ministry of Health of each G20 nation where none was available on their respective websites. We also used the International Severe Acute Respiratory and Emerging Infections Consortium (ISARIC), an international clinical research network for infectious disease.

10To facilitate a more rapid review, one reviewer independently screened the title and abstract of all references. A second reviewer screened 10% of excluded references for quality control. After each reference passed the first screening stage, two reviewers screened the full text independently. Where conflict about inclusion existed, a third reviewer made the final decision.

11We extracted data by using the methodological guide produced by Johnston et al. The team members speak multiple languages; as a last resort when no fluent speaker was available we used Google Translate. We used a standardised form for data extraction. We used Distiller SR (Evidence Partners, Ottawa, Canada) and Microsoft Excel for all screening and data extraction. For each guideline, we extracted data on source, year of production, clinical topics covered, and the patient demographic.

12Two reviewers independently appraised each eligible guideline by using the AGREE II instrument according to the instructions of the AGREE Research Trust. The AGREE II instrument provides an objective framework to assess the quality of clinical guidelines; it consists of six domains and two global rating items. The six domains are scope and purpose, stakeholder involvement, rigour of development, clarity of presentation, applicability, and editorial independence. Each domain is assessed on the basis of several “items,” of which there are 23 in total. The score is completed by at least two independent assessors on a seven point scale. Total scores are scaled to a percentage of the maximum score in each domain; 100% is achieved if each reviewer scores 7 for all items in a domain. The domain would score 0% if each reviewer scored 1 (the minimum value) for all items in the domain.

13There was no public or patient involvement in the course of this project. However, extensive involvement is planned in the wider systematic review of which this review was a part.

14In total, we identified 2996 records through database searching and a further 18 through grey literature searches. We excluded 2731 (96%) studies after de-duplication and title screening and a further 63 (60%) after further screening of the full text. Forty two guidelines proceeded to data extraction and synthesis, of which 18 directly pertained to covid-19 and 24 were guidelines relating to severe acute respiratory syndrome (SARS) or Middle East respiratory syndrome (MERS) by national organisations promoted in the covid-19 response (fig 1).

15We identified 18 national guidelines on covid-19, most of which were published in an upper middle income or high income country (table 1). We did not find a guideline produced in a low income country.

16Often clinical guidelines were embedded within a document that primarily focused on infection control. Generally, the clinical recommendations provided by the guidelines were non-specific and covered a narrow range of topics (table 2). It was evident that most countries relied heavily on WHO guidelines in formulating their own guidelines.

17The format of the supportive care recommendations in the guidelines varied widely, ranging from brief notes or flow diagrams to lengthy, nuanced descriptions of therapeutic options. Emphasis differed among the guidelines, with some being more conservative than others and with variation in specific recommendations such as the choice of antiviral drugs (table 3). Very few guidelines made specific recommendations on the use of treatments for symptom control such as non-steroidal anti-inflammatory drugs. Recommendations on the use of non-invasive ventilation varied widely (table 4).

18Overall quality as assessed by the AGREE II tool was poor (fig 2). The stacked polar chart shows the sum of the total AGREE II scores with sub-bars, representing six domains (100 for each domain), stacked end to end for each country. WHO guidelines were rated as 265.42 (44%) out of 600 in total. Clinical guidelines produced in Spain (260; 43%)) and in Malaysia (248; 41%) scored particularly highly for methodological rigour, whereas the guidelines produced in China (145; 24%) and South Korea (156; 26%) scored particularly poorly. Domains in which all of the guidelines scored poorly were stakeholder involvement, applicability, and editorial independence.

19We observed a lack of clear links between the evidence base and recommendations throughout the guidelines globally—for instance, in the strong discouragement of the use of steroids or the use of antimicrobials (table 5). Antimicrobial recommendations also varied, with several guidelines recommending empirical antimicrobial treatment for all patients with severe acute respiratory symptoms and others recommending it only on the basis of clinical aetiology.

20Globally, very few recommendations were made on prophylaxis for venous thromboemolism (table 6). Some guidelines linked their recommendations to a consideration of the published literature, but many did not. Even where an explicit link was made, no systematic weighting for that evidence (for example, Grading of Recommendations, Assessment, Delivery and Evaluations (GRADE)) was used.

21We found wide variations across individual score domains when comparing guidelines. None of the guidelines scored above 50% for the domains on editorial independence, applicability, or stakeholder involvement (fig 3). The score for rigour of development, a key component for evidence based guidelines, ranged from 10% to 76%. We could find no examples of a systematic review being done, most guidelines did not grade the strength of their recommendations, and little description existed of how these recommendations were made. We found no evidence of a guideline being externally reviewed before release. The guidelines made little provision for vulnerable groups such as pregnant women and children, and few recommendations pertained to the care of older people and immunocompromised patients (table 7).

22We compared the quality and content of WHO guidelines for MERS with the current interim WHO guidelines on covid-19 (table 8). The covid-19 guidelines had significantly lower scores than the MERS guidelines in all AGREE II domains, except the domain of rigour of development. Both guidelines followed similar case definitions.

23WHO produces a handbook for internal guideline development, including details of how it produces interim guidance. WHO states that “although the target audience or other stakeholders may demand that interim guidance be generated quickly, this type of guideline fully complies with all processes and procedures and meets the standards set out in this handbook.”

24However, our evaluation suggests that the WHO MERS guidelines, originally published in 2013 and now in their third version, continue to fail to score highly in the domains of applicability, editorial independence, and stakeholder involvement. This is echoed in the interim guidance for covid-19, which also scored poorly in these domains. The low scores are caused by little discussion of the applicability of the guidelines, inadequate recording of conflicts of interest, a narrow range of included stakeholders, and insufficient planning for updating the document.

25The WHO covid-19 interim guidelines were based on the early MERS guidelines and are very similar in their recommendations. Considerable overlap in recommendations may exist because a betacoronavirus causes both MERS and covid-19, and other guidelines on viral respiratory infections may also have applicable elements. Our search found alternative guidelines for other respiratory infections that may be applicable and of high quality (table 9).

26As the covid-19 pandemic grows, clinical guidelines will be in increasing demand globally. This rapid review contains lessons for both the current pandemic and future pandemics. We found shortcomings in the international body of clinical guidelines covid-19 produced early in the pandemic. Very few organisations constructed their own guidelines independently, meaning that nearly all guidelines incorporated the WHO interim guidance at least partially. Lack of reporting of the process of obtaining evidence and reaching recommendations made assessment of the appropriateness and quality of the recommendations for individual users and organisations difficult.

27Clearly, these guidelines were made under conditions of uncertainty at a time of international crisis. Moreover, elements of AGREE II may be ill suited to the demands of guideline production during the current crisis. Nevertheless, well constructed, evidence based clinical guidelines are crucial to the response to covid-19, to help to guide clinical decision making and improve patients’ outcomes. Clinicians need to be able to rely on the editorial independence of the guidelines they use, but declarations of interest were poorly documented in the early international covid-19 guidelines. In “peacetime,” declaring conflicts of interest is a vital component of both GRADE and WHO-INTEGRATE Evidence to Decision frameworks, so why not during a pandemic? Full disclosure of conflicts of interest is not time consuming and is important when making recommendations on novel or experimental treatment on the basis of limited or no evidence.

28Furthermore, given the complexity of the global health emergency, it seems reasonable that all guideline writers should seek to include as broad a range of stakeholders as possible. Given the resource constraints faced, matters of affordability and availability within health systems should be covered. Finally, almost none of the published guidelines we reviewed reported any mechanism for updates, audit, and monitoring. The covid-19 pandemic is rapidly evolving, and under these circumstances provisions for audit and monitoring of any guideline are crucial.

29The limited level of rigour in constructing the guidelines made accounting for the notable variation in the recommendations difficult. For instance, the Russian guidelines advocated the use of anti-inflammatory drugs whereas most others made no such recommendations. Most guidelines strongly discouraged the use of steroids for covid-19. However, the detail with which this recommendation was made varied widely. The use of steroids in acute respiratory infections such as covid-19 is contested, but the debate is complex and relies on the interpretation of observational studies and surrogate outcomes. Guidelines must be clear but must not obscure the complexity of this debate to guideline users.

30The most marked difference in the content of the guidelines was in the support for antiviral agents, both in terms of whether to use antivirals at all and in the specific antiviral regimen endorsed. Clearly, complex factors are involved in choosing a treatment regimen in an emergency. However, because the guidelines were drafted without clear links between recommendation and underlying evidence, the logic of each regimen was hard to ascertain. Expecting strongly evidenced interventions for a recently emerged disease is unreasonable, and we appreciate that more thorough and a greater number of guidelines will be produced as the pandemic progresses. However, clinicians need guidelines that are evidence based and include a thorough evaluation of the level of evidence on which a recommendation is based, while also conveying which populations and indications the guidance applies to. When no evidence is available, this should also be made clear. Any recommendations made should be directly linked to an evaluation of the supporting evidence. GRADE is a systematic method for making clinical practice recommendations and helps to portray the certainty with which a recommendation is made and could be used. Arguably, the variation seen in the recommendations on the use of supportive care and the lack of recommendations for vulnerable, high risk populations underline the importance of a gold standard framework for guideline construction under conditions of uncertainty.

31All of the covid-19 guidelines found were produced in high income or upper middle income countries, and therefore include assumptions about technology that may not be realistic in low income settings. For instance, avoiding non-invasive ventilation in favour of early intubation and prone positioning might reflect the clinical gold standard in some countries, but it is clearly heavily resource dependent. This must be tackled as the pandemic moves into lower resourced settings.

32The early covid-19 guidelines showed a lack of inclusivity. We sought to explore whether the guidelines incorporated the needs of vulnerable groups, defined broadly to include older people, children, pregnant women, and patients with comorbidities. We found that some groups are only cursorily covered in the guidelines or not mentioned at all. Few guidelines explicitly described care for older or immunocompromised people, who represent vulnerable groups with unique needs.

33This review has some limitations. Firstly, most guidelines were published outside of bibliographic databases. Although we did an extensive search of the grey literature, our searches may have missed guidelines and been biased towards English language literature.

34Secondly, the guidelines were published in a range of languages. We have used native speakers where possible, but we have also had to make extensive use of translation software. This risks losing the finer nuances of a complex topic. We had to exclude Iranian guidelines from the discussion altogether because we could not secure details of their origin and methods with confidence.

35Thirdly, AGREE II assumes a non-emergent process. Although the authors are confident of its applicability to a variety of settings, it was designed for guidelines produced by large teams in non-urgent conditions.

36Finally, this review is limited by its cross sectional nature. We acknowledge and appreciate that more guidelines have emerged since the early pandemic and that some included in this review have been updated. This review can act as a foundation for future research to evaluate temporal changes in the quality of clinical guidelines as the covid-19 pandemic progresses.

37To our knowledge, this is the first rapid review of guidelines produced during a pandemic. Previous work has retrospectively examined the quality of guidelines produced in emergencies and noted serious methodological shortcomings during their production. Our study confirms the need for a rigorous method of producing guidelines during a public health emergency of international concern.

38This review has lessons for clinicians, stakeholders, and governments facing future outbreaks. Guideline development in a pandemic is extremely challenging. A flexible yet robust method of producing guidelines during an emergency is needed, recognising the contingent nature of the evidence while guaranteeing essential methodological rigour and providing a mechanism for regular review.

39We suggest three components that are crucial for the production of emergency guidelines. Firstly, all guidelines produced in an emergency should be considered “living” guidelines and produced with set, transparent timelines for revision and amendment. Secondly, all guidelines produced in an emergency should use a transparent framework for weighing the strength of their recommendations (for example, GRADE or ADAPTE), so that users can understand the mechanism whereby recommendations have been made in uncertainty. Thirdly, all guidelines produced in an emergency should be externally reviewed using a validated tool such as AGREE II, to highlight areas in which they are vulnerable and to allow the authors to remedy these deficiencies in future revisions of their “living” guideline. Ensuring that WHO has the resources to provide the best possible guidelines during an emergency and for these to be updated meticulously is also vital. Our review found comprehensive guidelines already written for related respiratory infections. Building on established guidelines and adapting them could become a key part of any future emergency pandemic response.

40The validity of currently available clinical guidelines for covid-19 will not be known for some time. This review highlights areas for improvements ahead of the next public health emergency of international concern.

41Clinical guidelines produced in previous healthcare emergencies have fallen below gold standards of guideline development

42During the early coronavirus pandemic, a high degree of uncertainty existed about the optimal clinical management of patients with covid-19

43Clinical guidelines written in the early covid-19 pandemic possessed methodological weaknesses, especially in the rigour of their development

44Recommendations for the management of vulnerable groups such as older people were also neglected

45Guidelines produced early in future pandemics should prioritise contingency, adaptability, and methodological rigour