I found the following paper yesterday and wonder if others on this list
would also find it interesting?
Evidence for health policy. Peter Littlejohns and
Kalipso Chalkidou. Journal of Research in Nursing 2006; 11(2): 110-117.
If I understand the authors correctly, NICE are no longer grading their
recommendations according to the level of evidence, and are considering
abandoning the practice of ranking evidence according to a hierarchy of
methodological quality. The reasons for this change in approach to
developing evidence-based guidelines include:
1) grades of recommendation have misled users (e.g. into thinking that a
grade A recommendation is necessarily more important than a grade B
recommendation)
2) hierarchies of evidence have misled guideline developers into
ignoring useful evidence
3) recommendations need to be based on 'fit for purpose' evidence; i.e.
the evidence from research methods that best address the question(s)
4) recommendations should be based on the totality of (critically
appraised) evidence; for example evidence on benefits, evidence on
harms, evidence on the burden of the intervention, and evidence from
economic studies
5) recommendations are based on values as well as evidence
6) recommendations are based on judgements: scientific judgements and
value judgements
7) the reasoning that leads from evidence and values to judgements and
recommendations should be explicit and transparent
8) recommendations should address external validity; in other words
guideline developers should be explicit about the confidence with which
a recommendation can be applied to a hypothetical typical patient; and
they should help healthcare professionals assess the confidence with
which they can apply recommendations to patients who are certainly not
hypothetical and probably not typical
The authors do not discuss the likely consequences of their new approach
to guideline development. However, I guess that it will:
a) entail more work for guideline developers: it is not easy to be
explicit and transparent about reasoning
b) lead to more productive debate about controversial recommendations:
an argument is more likely to be focused on the nub of the matter
c) make it more difficult for users to ignore complexity and jump to
inappropriate conclusions (managers might not appreciate this)
Finally, two questions:
(i) Does this paper presage the end of absolutism in EBM?
(ii) Would progress in the methodology of guideline development have
been faster if the EBM community had been aware of Stephen Toulmin's
work?
(Stephen Toulmin became famous (or infamous) for his analysis of
rational argument - The Uses of Argument. Stephen Toulmin. Cambridge
University Press. 1958 - He uses different language, but the principles
of reasoning that he described differ only in unimportant details from
those proposed in the above paper.)
