The move to a new, stable system, has allowed us to start really improving the quality of Trip. Trip is a hugely valuable tool, but it isn’t perfect and the old system was creaking.
One immediate area for attention has been the way we grab systematic reviews. We have three main ways of adding systematic reviews to Trip:
- A number of publishers are considered producers of systematic reviews and their content is not routinely added to PubMed – so we manually grab those records.
- PubMed – we use a filter to identify systematic reviews
- Others – we try to identify systematic reviews from a small number of third-party sources
The middle one, PubMed filter, is a complex area to navigate given the tension between sensitivity and specificity. Too sensitive (to identify ALL systematic reviews) and you bring in a load of false positives. Too specific (to only identify TRUE systematic reviews) and you miss a load of systematic reviews – false negatives.
So, we’ve been carrying out a lot of tests on PubMed and have plumped for this filter:
(systematic review[sb] OR meta analy*[TI] OR metaanaly*[TI] OR “Meta-Analysis”[PT] OR “Systematic Review”[PT] OR “Systematic Reviews as Topic”[MeSH] OR “systematic review” [TI] OR “health technology assessment” [TI] OR “Technology Assessment, Biomedical”[Mesh])
At the time of writing the above search identifies 372,212 results (click here to try it yourself). We estimate the other sources contribute an additional 80-100,000 systematic reviews. So, we’re on our way to half a million!
The new PubMed filter will also be checked much more regularly than previously and the third option (third-party sources) are next – again improved filter and more regular checking.
Systematic reviews are hugely important in the EBM world and therefore we’re delighted with progress and we hope our users will be too.