IQWiG study filters for identifying controlled non-randomized studies (PubMed/Ovid MEDLINE)
These are two search filters developed by the German Institute for Quality and Efficiency in Health Care (IQWiG) to help find controlled non-randomized studies in PubMed and Ovid MEDLINE. One filter is designed to catch as many relevant studies as possible, while the other focuses on returning only the most relevant ones.
At a glance
Use when
Conducting systematic reviews requiring comprehensive identification of controlled non-randomized studies, especially when randomized trials are limited or unavailable
Avoid when
Searching for uncontrolled observational studies or when using databases other than PubMed or Ovid MEDLINE
Inputs
Search queries in PubMed or Ovid MEDLINE requiring identification of controlled non-randomized studies
Outputs
Filtered search results containing controlled non-randomized studies with either high recall or high precision
How it works
Two study filters were developed and validated to identify controlled non-randomized studies (NRS) in PubMed and Ovid MEDLINE. The reference set included 2110 relevant publications from 271 Cochrane reviews and 4333 irrelevant references. The sensitivity-maximizing filter achieves 92.42% sensitivity and 79.67% specificity, while the specificity-maximizing filter achieves 92.06% specificity and 80.94% sensitivity. Both filters support comprehensive or focused literature searches for controlled NRS, enabling effect size estimation in evidence synthesis.
- HTA domains
- Clinical Effectiveness, Organisational aspects
- Assumptions
- Controlled non-randomized studies are more reliably identifiable than uncontrolled ones due to more explicit reporting patterns and search syntax; inclusion of control groups enables effect size calculation
- Strengths
- One filter achieves sensitivity above 92%, meeting the threshold for comprehensive retrieval; both filters are validated against a large reference set from Cochrane reviews; supports evidence synthesis where RCTs are lacking
- Limitations
- Filters are specific to controlled NRS and do not capture uncontrolled observational studies; performance may vary outside the validation set; not designed for databases other than PubMed and Ovid MEDLINE
- Also known as
- IQWiG controlled NRS filters, IQWiG study filters for non-randomized studies, IQWiG PubMed/Ovid filters for controlled NRS
Questions this answers
- › How can controlled non-randomized studies be efficiently identified in PubMed and Ovid MEDLINE?
- › What search filters achieve high sensitivity for controlled NRS?
- › What search filters offer high specificity when precision is prioritized?
- › Which filter should be used for comprehensive systematic reviews?
- › When should a more focused search strategy be applied for non-randomized evidence?
- › How can effect size estimation be supported through better identification of controlled observational studies?
References & sources
Similar by meaning
Beta record. Based on the original catalogue summary; primary-source enrichment pending.