Study filters consist of a defined combination of search terms to pre-select references with a specific study design as accurately as possible. Controlled NRS are presented very differently in databases.
Without a special study filter for NRS, searches have therefore so far produced a very large number of hits, from which relevant NRS have to be filtered out manually in a very time-consuming process.
With the help of the new study filters, the number of hits can now be markedly reduced when conducting a systematic search for NRS in the PubMed and Medline (Ovid) databases.
An IQWiG team developed two NRS search filters in collaboration with researchers from the Department of Health Research Methods, Evidence & Impact at McMaster University/Canada.
On the basis of 4544 pre-identified relevant NRS, the two new filters were developed with the help of the McMaster Clinical Hedges Database: The search filter for a comprehensive search identifies controlled NRS in PubMed and Medline with a high sensitivity of 92.17%.
The second search filter is optimized for more focused searches for NRS and yields a specificity of 92.06%. Details on the development and validation of the search filters are now available in a publication.
Source:
Journal reference:
Waffenschmidt, S., et al. (2020) Development and validation of study filters for identifying controlled non‐randomized studies in PubMed and Ovid MEDLINE. Research Synthesis Methods. doi.org/10.1002/jrsm.1425.