Search process
Summary
Step 1: search databases
- Search for systematic reviews.
- Search for randomised controlled trials.
- Search for observational studies if required / studies on adverse effects if required.
Step 2: assess and select results based on predetermined selection criteria (outline below)
- Initial assessment of abstracts by information specialists.
- Selected results sent to BMJ Clinical Evidence contributor for further consideration using full-text of studies.
If you have any comments on our processes, please contact us - we would love to hear from you.
Step 1: search databases
Search for systematic reviews
Main sources:
- Cochrane Database of Systematic Reviews (on CD-ROM)
- Medline [see search strategy]
- Embase [see search strategy]
- Other databases (e.g. PsycInfo) as appropriate
Additional sources:
- Centre for Reviews and Dissemination (CRD) website
- Database of Abstracts of Reviews of Effects (DARE) online database
- Health Technology Assessment (HTA) online database
- National Institute for Health and Clinical Excellence (NICE) website
- TRIP online database
Search for randomised controlled trials
Sources:
- Medline [see search strategy]
- Embase [see search strategy]
- Cochrane Central Register of Controlled Trials
- Other databases (e.g. PsycInfo) as appropriate
For some very large reviews, we only search for randomised controlled trials published after the search date of systematic reviews found.
We use validated search strategies to search Medline and Embase for systematic reviews and randomised controlled trials. Our search strategies are heavily based on strategies developed by Brian Haynes and Anne McKibbon at McMaster University in Canada,[1] and strategies developed by Carol Lefebvre et al at the UK Cochrane Centre[2].
Search for observational studies / studies on adverse effects if required
Observational studies
In areas where there are few or no good systematic reviews or randomised controlled trials, we may search for observational studies. Again we search the full Medline and Embase databases. The observational study search strategies have been designed in-house by BMJ Knowledge information specialists to retrieve studies most likely to meet our methodological criteria.
- See cohort studies search strategies
- See cohort and case-control studies search strategies
- See cohort, case-control and case series search strategies
Studies on adverse effects
We may also perform specific searches on Medline and Embase to retrieve studies on adverse effects. The adverse effects search strategies have been designed in-house by BMJ Knowledge information specialists.
See adverse effects search strategies
Subject search strategies
All above search strategies (SR, RCT, observational and adverse effects) are combined with subject search strategies – population / interventions – to retrieve relevant studies. These are designed in-house by BMJ Knowledge Information Specialists. For adverse effects searches, only specific harms are searched for – these are also incorporated into the strategy. For details of subject search strategies, please contact us at BMJ Clinical Evidence.
Step 2: assess and select results based on predetermined selection criteria (outline below)
Before the search is carried out, the selection criteria for each review is determined through a review planning process. At the assessment of abstracts stage, we apply this selection criteria to the search results, in order to exclude references which are definitely irrelevant. Remaining references are then sent to the BMJ Clinical Evidence contributors for further consideration. Below is the default selection criteria we use. This criteria may differ for some reviews, based on discussions with review editors and contributors.
To select systematic reviews for further consideration:
Abstracts are selected which indicate the review:
- used comprehensive search methods
- fits the pre-determined inclusion criteria for the review - population, intervention(s), comparison(s) and outcome(s)
To select randomised controlled trials for further consideration:
Abstracts are selected which indicate the study:
- is randomised
- has concealed allocation of interventions
- is appropriately blinded
- has at least 20 patients in total
- has at least 80% follow-up
- fits the pre-determined inclusion criteria for the review - population, intervention(s), comparison(s) and outcome(s)
If an abstract indicates the study definitely does not match one of the above criteria it is excluded. If we are unable to definitely exclude a study using the information in the title / abstract, the reference is included in the selected set of references to be sent to the BMJ Clinical Evidence contributor for further consideration.