Researchers have recently evaluated a wide array of search filters designed to retrieve systematic reviews from databases like MEDLINE and Embase. The goal was to determine how well these filters perform in identifying relevant studies while excluding irrelevant ones, a measure known as sensitivity and specificity.

The findings suggest that while many filters demonstrate acceptable sensitivity and specificity, no single filter can be universally recommended. This is largely because many of these filters were developed using older sets of reviews, which may not accurately reflect current reporting standards.

The Challenge of Finding Reliable systematic Reviews

“Search filters combine words and phrases to retrieve records with a common feature… and are typically evaluated in terms of their sensitivity and precision.”

Systematic reviews are crucial for healthcare professionals as thay summarize and synthesize scientific evidence. Databases offer access to these reviews, and search filters are intended to streamline the retrieval process. However, the effectiveness of these filters varies.

The study aimed to identify existing search filters for systematic reviews, assess their quality, and gather data on their sensitivity, specificity, and precision. The researchers conducted searches for studies that developed, evaluated, or compared search filters for retrieving systematic reviews in MEDLINE and Embase.

The research team identified nine studies that developed filters for MEDLINE and three studies for Embase. The analysis revealed that for MEDLINE, all filters exhibited similar sensitivity and precision, with one filter showing higher specificity. In Embase, the filters showed variable sensitivity and precision, with limited study reports perhaps affecting the accuracy of assessments.

Limitations and Future Considerations

One meaningful limitation is that some filters were developed for specific topics, such as public health. Additionally, most filters were created using older studies, which may not reflect current reporting practices for systematic reviews. Furthermore, these filters may struggle to differentiate between high- and low-quality reviews.

The evidence considered in this analysis is up-to-date as of January 2023.