Method of development

Scope

Key areas covered by the guideline include (1) assessment of the child with acute breathing difficulty, (2) management of the children from the point of presentation to the hospital to a decision regarding admission or discharge, (3) discharge, admission and referral criteria and (4) identification of the severity of the problem, appropriate investigations and treatment. The guideline does not cover children known to have chronic respiratory difficulties. Further exclusion criteria have been included in the full guideline, which is available on request as part of a full technical report.

Guideline users

This guideline has primarily been written for use by junior doctors who see children in an acute hospital setting. Senior doctors, nurses or other professionals allied to medicine may wish to refer to its recommendations in order to be aware of current evidence. With the increase in nurse practitioners in the ED or other situations such as walk-in centres or primary care practices, this guideline may also be useful to this group of health professionals.

Guideline development group

In addition to the authors, the development group was composed of eight doctors and nurses who had expertise in either literature appraisal and guideline development or the clinical area being addressed by the guideline. Their role was to overview the development process, in which the Delphi panel had a key role. Input from a primary care doctor and a patient representative was achieved.

Delphi panel

Sixty-five individuals were approached to form the Delphi panel, 59 replied and 50 agreed to participate. The composition of the final panel is provided in table 1. Members were selected from district general hospitals and tertiary referral centres. They included different grades and disciplines of staff. The development group all contributed to the selection process and was not aware of the names forwarded by other members of the group. The panellists were unaware of the identity of the other panel members.

Development process

An initial review of the published literature and of existing guidelines was undertaken by the research fellow (ML) to determine the amount of evidence available and to provide the outline of the decision tree that represents the sequence of events involved in the assessment and management of a child with acute breathing difficulty. The decision tree and clinical questions were all discussed with a respiratory paediatrician to check that all the relevant clinical areas had been addressed. Key clinical questions were then identified, and these were later broken down further into more specific questions, which were addressed initially by a literature search, followed by the Delphi Consensus process. The whole process was based on the Scottish Intercollegiate Guideline Network7 and the Appraisal of guidelines research and evaluation (AGREE) criteria.8

In addition to searching the computerised databases from 1966 using MeSH headings and “text words”, limited to 0–16 years of age, further articles were obtained from colleagues and by hand searching the bibliography of articles. A hand search for the last 5 years of the most relevant journals was performed, and journals not listed on MEDLINE were also searched if thought to be relevant to the subject area. A search for unpublished articles was also performed. If high-quality evidence was not available, a modified Delphi process was used. The literature was then evaluated against standards provided by the Scottish Intercollegiate Guidelines Network.9

The articles were assessed for their relevance and quality and then critically appraised. Data were extracted using standardised data extraction forms.9 After extraction of abstracts and appraising full articles, 77 studies were finally used to provide the evidence for the clinical statements. The statements were graded from 1 to 4 according to the strength of the evidence on which they were based.9 They were then translated into clinical recommendations, graded A to D, grade A being based on evidence from meta-analyses, systematic review or high-quality randomised controlled trial and grade D being based on non-analytical studies or expert opinion.9

All the appraised articles along with the summarised data extracted from them, the clinical questions that they addressed, a summary of the literature discussed and the statements were sent to the Delphi panel. The Delphi method10 was used to provide consensus where evidence was lacking and to help translate the evidence into relevant and unambiguous clinical recommendations. The process also provided an additional opportunity for the accuracy and relevance of the clinical questions to be checked. This method was chosen so that the consensus process could be formalised using an anonymous and iterative process. The panel was asked to provide additional references, question the appraisal of the papers and refine the language of the final recommendations. Consensus of agreement was predefined as 83% of panellists answering between 7 and 9 on the Likert scale and therefore agreeing with the recommendations.11

Individual statements reaching consensus after round one were formulated into recommendations and included in the guideline. Any recommendations not reaching consensus were either rewritten using different language and resubmitted to the second round of the Delphi process, or resubmitted unchanged or removed from the process altogether. All comments were assessed individually even when a recommendation did reach statistical consensus. The reason for this was that some disciplines were represented by only a small number of individuals on the panel, and it was essential to take note of their comments because they may have provided the only insight from their area of expertise.

In the second round, the new questionnaire was sent to the respondents again. They received feedback on the rest of the group’s responses, allowing them to view their judgement in the light of responses from the other respondents. Statistical feedback (median and interquartile range) and qualitative feedback of comments were given to the respondents.

The development group had decided in advance that all recommendations based on evidence should be included in the guideline even if the Delphi panel did not agree with them in line with current opinion on evidence-based medicine; that is, that evidence should always be considered as superior to expert opinion alone. However, the panellists were still invited to comment on the evidence-based statements to assist with clarity and comprehensibility of the statements and to allow for any errors in grading to be amended.

A further review of the literature took place in 2008 to ensure that the recommendations were still current.

A patient leaflet was developed in association with the patient representative and provided to the Delphi panel for comments. After completion of the process, the leaflet was again shown to five sets of parents for their final approval.

Results

The results from both rounds of the Delphi process can be found in table 2. For one of the recommendations, two studies provided conflicting evidence, and the Delphi process was used to identify which study was supported by consensus opinion. All these details are made transparent in the guideline. In summary, 43 statements from round one and 38 statements from round two were used to formulate the 63 clinical recommendations found in the guideline, of which 10 were grade A (evidence from systematic review or meta-analyses), 5 grade B, 17 grade C and 31 grade D (consensus or expert opinion). Key clinical questions not reaching consensus included a “definition of raised respiratory rate for the different age groups”, “use of oral steroids in children with mild wheeze”, “chest x ray in children with first attack of wheeze” and “clinical signs of pneumonia.” At the time of the publication of the guideline, there was insufficient evidence on which to support a firm statement on these issues, which has subsequently led to research to inform these statements.

The guideline

The grades of the final recommendations included in the guideline are provided in table 3 and can be found alongside the corresponding recommendation in the algorithm (Appendix available online).

Discussion

Clinical guidelines have become an integral part of clinical governance and are seen as a tool for promoting best practice. Clinicians need to combine highest quality scientific evidence enriched with clinical experience and tailored to meet, when possible, the opinions of the family.12

Junior doctors have varied experience in managing acute paediatric patients and value guidance. Traditionally, they turn to textbooks as a readily available source for advice, many of which are outdated and their source of evidence is frequently not clear. Few textbooks are based on critically appraised literature. With increasing interest in evidence-based medicine, clinicians seek guidance for their practice based on randomised controlled trials or systematic reviews and meta-analyses. But all too often, clinical questions at the forefront of a clinician’s mind are not currently answered by high-quality research. A guideline based only on the most up-to-date evidence would include large gaps in patient management. In our view, a guideline combining evidence with formal consensus opinion in areas where evidence is not available is useful to junior staff when faced with a patient requiring immediate treatment.

Criticisms regarding the Delphi process10 11 for guideline development include concerns that guidelines based on consensus promote practice that is not based on the highest level of evidence and may in fact not be best practice. Other concerns include consensus guidelines hindering innovation and research in medicine and of consensus opinion over-riding evidence. In our view, the development of guidelines by formal consensus as the next step to take after first identifying and including guidance that is firmly evidence based allows evidence-based recommendations to be linked together in a comprehensive guideline and in which consensus fills the gap until more evidence is available. Thus, they can be applied in clinical practice. The consensus approach itself was formal and transparent. It was not used to undermine available evidence but rather to translate it into a form that could be used by clinicians. In addition, guidelines help to promote, not hinder, research by identifying research deficits and new questions to be addressed by further research as highlighted above.

We decided to make the guideline problem based13 rather than diagnosis based, unlike many other existing guidelines. This was to aid doctors with little paediatric experience to manage a patient at presentation to the hospital and the sequential approach to arriving at a diagnosis. A guideline relating to specific diagnosis is useful only once a correct diagnosis has been made. Our guideline is one of a very few problem-based13 14 guidelines relating to the initial management of a child presenting to the hospital.

The guideline has now been implemented, and results of the implementation process showing changes in practice will be published in a separate paper as were the effects of the other problem-based guidelines developed by the same department.

Conclusion

This guideline is an evidence-based guideline appropriately combined with consensus opinion. It uses formal consensus opinion to make it clinically relevant by linking evidence-based recommendations together when there are intervening steps where evidence is not available. Without this interim process, the development of an algorithm would not be possible. A combination of evidence-based medicine and consensus allows evidence to be used to inform decision making and is transparent, allowing the clinician to be fully informed of the strength of the recommendations made and how they have been derived. Our problem-based guideline was specifically developed for use by junior doctors who are often the first to see patients, who wish to apply basic safe management to their patients who may not have immediate access to senior support and for whom it is more important to provide an appropriate management plan rather than expecting them to reach a diagnosis. This guideline provides them with this help in a logical easily implemented manner and keeps them up-to-date with current evidence and opinions.