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

methodological quality, and descriptions of techniques for critical appraisal are widely available. 21 Inevitably, there is some degree of subjectivity in determining whether particular flaws or a lack of suitable information invalidates an individual study.
    Robustness of findings
    The strength of evidence from a published study would depend not only upon the type and quality of a particular study but also upon the magnitude of any differences and the homogeneity of results. High-quality research may report findings with wide confidence intervals, conflicting results or contradictory findings for different outcome measures or patient subgroups. Conversely, sensitivity analysis within a cost-effectiveness or decision analysis may indicate that uncertainty regarding the exact value of a particular parameter does not detract from the strength of the conclusion.
    Applicability
    Strong evidence in a set of guidelines must be wholly applicable to the situation in which the guidelines are to be used. For example, a finding from high-quality research based upon a hospital population may provide good evidence for guidelines intended for a similar setting but a lower quality of evidence for guidelines intended for primary care.
    Grading system for evidence
    The following is a simple pragmatic grading system for the strength of a statement of evidence, which will be used to grade the evidence in this book (and the other volumes in the Companion series). Details of the definitions are given in Table 1.4 . For practical purposes, only the following three grades are required, which are analogous to the levels of proof required in a court of law:
    Table 1.4
    Grading of evidence and recommendations

    I  ‘Beyond reasonable doubt’. Analogous to the burden of proof required in a criminal court case and may be thought of as corresponding to the usual standard of ‘proof’ within the medical literature (i.e.
P
< 0.05).
    II  ‘On the balance of probabilities’. In many cases, a high-quality review of literature may fail to reach firm conclusions because of conflicting evidence or inconclusive results, trials of poor methodological quality or the lack of evidence in the population to which the guidelines apply. Where such strong evidence does not exist, it may still be possible to make a statement as to the ‘best’ treatment on the ‘balance of probabilities’. This is analogous to the decision in a civil court where all the available evidence will be weighed up and a verdict will depend upon the ‘balance of probabilities’.
    III  ‘Unproven’. Where the above levels of proof do not exist.
    All evidence-based guidelines require regular review because of the constant stream of new information that becomes available. In some areas, there is more rapid development and the emergence of new evidence; in these instances, relevant reference will be made to ongoing trials or systematic reviews in progress.
    Grading of recommendations
    Although recommendations should be based upon the evidence presented, it is necessary to grade the strength of recommendation separately from the evidence. For example, the lack of evidence regarding an expensive new technology may lead to a strong recommendation that it should only be undertaken as part of an adequately regulated clinical trial. Conversely, strong evidence for the effectiveness of a treatment may not lead to a strong recommendation for use if the magnitude of the benefit is small and the treatment very costly.
    The following grades of recommendations are suggested and details of the definition are given in Table 1.4 :
    A A strong recommendation, which should be followed.
    B A recommendation using evidence of effectiveness, but where there may be other factors to take into account in the decision-making process.
    C A recommendation where evidence as to the most effective practice is not adequate, but there may be reasons for making the recommendations in order to minimise cost or