Improving the quality of systematic reviews of neurological conditions: an assessment of current practice and the development and validation of six new search strategies.

BACKGROUND AND PURPOSE: Our aim was to study the quality of the literature search strategies used in recent systematic reviews and to develop and assess the diagnostic accuracy of six new search strategies (i.e. hedges).

METHODS: Six neurological conditions were studied: migraine, stroke, dementia, epileptic seizures, Parkinson's disease and multiple sclerosis. Two reviewers independently assessed the quality of the search strategies used in systematic reviews published in 2015-2016. Complex hedges pertaining to the six conditions for use in Ovid MEDLINE were developed. Their diagnostic accuracy was compared to simple, single-term keyword searches.

RESULTS: Almost 60% of quality criteria for the overall literature search strategy used in 182 systematic reviews were not respected. Over 30% of search strategies relied on a single keyword to identify the neurological condition. The sensitivities of our complex hedges amongst 10 311 articles were between 83% and 95%, significantly higher than the simple keyword searches (as low as 48%). The specificities were greater than 97%.

CONCLUSIONS: There is great room for improvement in the search strategies used in systematic reviews of neurological conditions. Complex hedges were developed and validated to improve the accuracy of such searches. It is expected that this will lead to higher quality systematic reviews and meta-analyses.