Re: Rehabilitation after stroke: summary of NICE guidance
Drummond and colleagues (20 June 2013) have raised several points on the methods used in NICE clinical guideline development and the assessment of the stroke rehabilitation literature in particular. The standard NICE methodology has been followed in the development of this guideline, with the addition of a formal consensus exercise, and we will attempt to answer their specific concerns, but also highlight the availability of ‘The guidelines manual’ that describes NICE methodology in detail and is freely available via the NICE website.
In response to the specific points made, NICE methodology uses the GRADE (Grading Recommendations Assessment, Development and Evaluation) system for rating quality of evidence reviewed. GRADE rates the quality of the evidence for a particular outcome across all the studies rather than rate the quality of individual studies. The studies included in the guideline have therefore not been labelled as being of low quality, but rather, the quality of the evidence for a specified outcome. This quality assessment represents the Guideline Development Group’s summary of their confidence in the estimate of effect from the body of evidence. The reasons for ‘downgrading’ the evidence are presented within the guideline and vary between outcomes, so it is difficult to provide a single explanation for the quality of the evidence across the guideline.
The selection of studies is guided by the question protocol developed in collaboration with the Guideline Development Group. Studies that match the population, intervention and comparator and outcomes stated in the protocol are selected and included in the systematic review. If a Cochrane review matching the protocol was identified it would be included, such as the Cochrane reviews that were included in the speech and language therapy, the physical fitness and the stroke rehabilitation unit reviews. Other Cochrane reviews were used as a source of references. As studies on a stroke population were available for all the questions addressed by the guideline, the GDG believed it was more helpful to consider these than to include indirect evidence on another population. This is usual practice for NICE guidance unless there is a good reason to consider evidence from another population. One consequence of using more indirect evidence would be to downgrade the quality of the evidence further.
The decision to conduct the modified Delphi survey was made as a direct response to Stakeholders’ comments on the initial consultation draft that the guidance did not present a complete stroke rehabilitation pathway, and further work was needed that addressed the structure and process of stroke rehabilitation. Areas highlighted by stakeholders included multi-disciplinary team working, assessment and care planning, and long term health and social support. From further searches conducted and an evaluation of other national and international stroke rehabilitation guidelines it was apparent that there was an absence of robust evidence about the effectiveness of many of the possible interventions in these areas, and this was highlighted by the number of consensus based recommendations made for these areas within other available guidance. The Delphi survey approach was chosen to try to address the difficulties of ‘opinion’ led guidance, by offering the opportunity to create a more robust base to clinical interpretation and subsequent recommendations. This survey has delivered the largest ‘expert consensus’ view within this field, with good multidisciplinary stroke professional body representation that was reviewed by both the GDG and the RCP Stroke Intercollegiate Working Party to ensure relevant experts were identified. Efforts were made to recruit Delphi panel members from all professions and, in the case of an identified under-representation, organisations were contacted to encourage participation. In the analysis of responses received we were able to filter according to profession, and found no evidence that the responses were systematically different to each other.
Drummond and colleagues have questioned the rigour of the methods employed, however an assessment of the literature around Delphi methodology demonstrates no hard and fast rules in relation to reaching consensus. A protocol was developed in consultation with the Guideline Development Group and NICE detailing the proposed methodology and parameters for agreement. In this case a pragmatic approach was adopted to maximise the opportunity to gain consensus but terminating the survey at the point when it was apparent there would be no added value in continuing due to the diverse multi-disciplinary views and opinion. The survey did not present questions, but rather statements and these were based on the recommendations made in other guidance where available. These statements were refined following comments made by the survey participants in the initial round of the survey, and the example quoted was not subsequently included. Furthermore, Delphi panel members had the opportunity to add free text comments to each statement in each round and a qualitative analysis was conducted to identify themes which featured in the discussion and the drawing up of recommendations.
With regards the wording of the recommendations, this reflects the evidence available and the GDG’s interpretation of this evidence balanced with their clinical knowledge and expertise. Much of the evidence demonstrated that people improved in function and mobility from having an intervention, but there was little evidence to support one type of intervention over another. Therefore the GDG were of the opinion that the patient should have a choice, and it would be appropriate for the health professional to discuss with the patient the therapies that may best enable them to achieve their goals, and a range of therapy options given to aid people in making decisions. Whilst Drummond and colleagues may consider more directive recommendations appropriate, stating which patients should have a specific intervention, this was not reflected by other comments received from Stakeholders who criticised the guideline for not taking a more holistic approach reflecting individual patient’s needs. The authors cite the example of the memory and cognitive function recommendations; however we believe they are aware the clinical picture in stroke is complex, and specific targeting of population and intervention could result in a recommendation that is useless. If the whole of this section is read we state assess and then consider a range of strategies tailored to the patient, thus allowing skilled clinicians to use their judgement.
In relation to formulating research recommendations, these are decided by the consensus of the Guideline Development Group, and are for those areas reviewed by the guideline where the GDG agreed there was a lack of robust evidence, and where further research would benefit stroke rehabilitation practice. The guideline has responded to stakeholders and extended the scope of the stroke rehabilitation pathway. It provides guidance to improve clinical care and inform patients and their carers and reduce variations in practice. The guidance will be updated when new evidence emerges to strengthen the evidence base for these, or other, interventions to improve the outcome from stroke rehabilitation.
Competing interests: No competing interests