Instructions for reviewers
The literature search strategy and overall methodology will be laid out in a document for the group.
Briefly we carried out a global search on the subject area angina and then the references generated were sifted first for relevance (by looking at the titles and abstracts) and then for methodology. The references that I shall be sending to you to review should be both relevant to the area of primary care management and meet broad methodological criteria.
The methodological sift
The grids attached to the front of the references indicate the main questions asked.
Randomised controlled trials
These provide the strongest evidence. Appropriately double blinded studies should be given more weight than single blind studies. Open studies that should have been blinded will be rejected on methodological grounds of potential bias.
Most of the references that are rejected at this point fail to meet the group size or follow-up criteria.
Cohort and case control studies
These must have a matched control group - uncontrolled studies should have been be rejected at the methodological sift stage. Group size and follow-up also remove some from this group of papers.
Reviews
Only meta-analyses and systematic reviews with clear method sections should pass the methodological sift.
Those papers which do not have a method section will also be sent to you, but with a note attached to ask you to check the reference section only - while these articles are not acceptable for evidence they provide a source of references for us. In these cases please mark those which look relevant and return the paper to me so that I can collect the articles.
Reviewing the papers
Weight the evidence of the papers according to study design:
Double blind randomised controlled trial > single blind randomised controlled trial > cohort study > case control study.
Good meta-analyses provide a good way of overviewing the area and structuring your summaries.
Check that the method is OK
Was the population appropriate?
Was the follow-up period reasonable?
Was the sample size large enough ?- this is often difficult to judge.
If the study is open in design then check that it could not have been blinded - if it could have been then reject.
Look at the results
What can you conclude?
How relevant are the outcomes measured?
Were the differences in outcome clinically important?
Were confounding factors and biases accounted for ?(important in case control and cohort studies)
Summary sheet on front of each article
Please briefly fill in the summary grid - and answer the 3 questions:
Is the methodology OK? if no reject it and do not include in overall summary.
Is it clinically relevant? if no reject it and do not include in overall summary.
Is it a key reference ?- should the whole group see a copy?
Please also circle those words at the bottom of the page that would be appropriate keywords.
Overall summary
Summarise the main findings of the each paper in one or two sentences.
Return this information in good time for the meeting so that it can be circulated to the group members. If hand written please print - if you prefer we can provide and transcribe micro-tapes for you. (Please ensure that you refer to the references by their reference number which is held on a database).