Core Topics in General & Emergency Surgery: Companion to Specialist Surgical Practice by Simon Paterson-Brown MBBS MPhil MS FRCS Read Free Book Online

information obtained from investigators?
Data synthesis
How sensitive are the results to changes in the way the review is done?
Interpretation of results
Do the conclusions flow from the evidence that is reviewed? Are recommendations linked to the strength of the evidence?
Are judgments about preferences (values) explicit?
If there is ‘no evidence of effect’, is care taken not to interpret this as ‘evidence of no effect’?
Are subgroup analyses interpreted cautiously?
    From Oxman A. Checklists for review articles. Br Med J 1994; 309:648–51. With permission from the BMJ Publishing Group Ltd.
    It is hoped that the results of a systematic review will be precise, valid and statistically powerful in order to provide the highest quality information on which to base clinical decisions or to produce clinical guidelines. The strength of the evidence provided by a study also needs to be assessed before making any clinical recommendations. A grading system is required to specify the levels of evidence, and several have previously been reported (e.g. those of the Antithrombotic Therapy Consensus Conference 50 or that shown in Table 1.3 ).
    Table 1.3
    Agency for Health Care Policy and Research grading system for evidence and recommendations

    From Hadorn DC, Baker D, Hodges JS et al. Rating the quality of evidence for clinical practice guidelines. J Clin Epidemiol 1996 49:749–54. With permission from Elsevier.
    The grading of evidence and recommendations within textbooks, clinical guidelines or ICPs should allow users easily to identify those elements of evidence that may be subject to interpretation or modification in the light of new published data or local information. It should identify those aspects of recommendations that are less securely based upon evidence and therefore may appropriately be modified in the light of patient preferences or local circumstances. This raises different issues to the grading of evidence for critical appraisal and for systematic reviews.
    In 1979, the Canadian Task Force on the Periodic Health Examination was one of the first groups to propose grading the strength of recommendations. 51 Since then there have been several published systems for rating the quality of evidence, although most were not designed specifically to be translated into guideline development. The Agency for Health Care Policy and Research has published such a system, although this body considered that its level of classification may be too complex to allow clinical practice guideline development. 52 Nevertheless, the Agency advocated evidence-linked guideline development, requiring the explicit linkage of recommendations to the quality of the supporting evidence. The Centre for Evidence-based Medicine has developed a more comprehensive grading system, which incorporates dimensions such as prognosis, diagnosis and economic analysis. 53
    These systems are complex; for textbooks, care pathways and guidelines, such grading systems need to be clear and easily understood by the relevant audience as well as taking into account all the different forms of evidence that may be appropriate to such documents.
    Determining strength of evidence
    There are three main factors that need to be taken into account in determining the strength of evidence:
    • the type and quality of the reported study;
    • the robustness of the findings;
    • the applicability of the study to the population or subgroup to which the guidelines are directed.
    Type and quality of study
    Meta-analyses, systematic reviews and RCTs are generally considered to be the highest quality evidence that is available. However, in some situations these may not be appropriate or feasible. Recommendations may depend upon evidence from other kinds of study, such as observational studies of epidemiology or natural history, or synthesised evidence, such as decision analyses and cost-effectiveness modelling.
    For each type of evidence, there are sets of criteria as to the