Limited search strategies were effective in finding relevant nonrandomized studies

Area of Science:

• Medical Informatics
• Health Services Research
• Systematic Review Methodology

Background:

• Searching electronic databases for nonrandomized studies presents challenges due to diverse designs and inconsistent terminology.
• Standardizing search strategies for comparative nonrandomized studies (cNRSs) is crucial for accurate systematic reviews.

Purpose of the Study:

• To develop and evaluate a method for limiting search strategies by study design for cNRSs.
• To improve the efficiency and accuracy of identifying relevant nonrandomized studies in health care interventions.

Main Methods:

• Selected four Cochrane systematic reviews incorporating nonrandomized studies (cohort, case-control, cross-sectional).
• Devised study-design-limited search strategies in MEDLINE and EMBASE.
• Proposed a progressive method (PM) and a fixed method for selecting optimal search terms for nonrandomized study designs.

Main Results:

• Search strategies limited by study design achieved high sensitivity (90%-100%) in both databases.
• The progressive method (PM) using controlled vocabulary and textwords demonstrated excellent sensitivity.
• A fixed set of controlled vocabulary and textwords also yielded high sensitivity (95%-100%).

Conclusions:

• Implementing search strategies limited by study design for cNRSs in healthcare interventions is feasible and effective.
• This approach enhances the ability to accurately identify relevant nonrandomized studies for systematic reviews.