Skip to Main Content

Systematic Reviews and Other Expert Reviews

#systematicreviews #sysrev

Developing Search Strategies in Multiple Databases

  • Use the controlled vocabulary for each database; controlled vocabularies include MeSH in PubMed and Emtree in Embase. 
  • Adapt search syntax (field tags, phrase searching, etc.) for each database.
  • Check whether proximity operators are available and will help with keyword searching.
  • Use validated search filters as appropriate for each database. See the "Using Search Filters" box below for more information.

Using Search Filters

What are Search Filters?
  • Validated filters, such as those listed under the adjacent tabs, are most commonly used to locate specific study types (e.g., RCTs) or to exclude animal studies.
  • However, rarely will a review of observational studies include a study filter.  
  • The PubMed RCT filters developed by the Cochrane Collaboration (provided under the adjacent tab) are considered the gold standards for identifying randomized controlled trials.
Filters vs. Limits
  • Do not limit by language, since this may introduce bias.
  • Limit by date only if this is dictated by the content of the review (e.g., the date a drug, product, or medical test was first introduced).
  • It is best to treat a limit (e.g., age) as a concept in the search strategy or as a part of your inclusion/exclusion criteria.

The following filters were created and validated by the Cochrane Collaboration and are listed on the Cochrane Work Review Group website:

  • Sensitivity-Maximizing Version (2008 revision) - Maximizes recall

(randomized controlled trial[pt] OR controlled clinical trial[pt] OR randomized[tiab] OR placebo[tiab] OR drug therapy[sh] OR randomly[tiab] OR trial[tiab] OR groups[tiab] NOT (animals [mh] NOT humans [mh]))

Link to this filter in PubMed.

  • Sensitivity- and Precision-Maximizing Version (2008 revision) - Provides a balance of recall and precision

(randomized controlled trial[pt] OR controlled clinical trial[pt] OR randomized[tiab] OR placebo[tiab] OR clinical trials as topic[mesh:noexp] OR randomly[tiab] OR trial[ti] NOT (animals[mh] NOT humans [mh]))

Link to this filter in PubMed.

A validated Embase RCT filter is in development.  Until a validated filter is available, the Cochrane Handbook (Part 2, Section - Search filters for identifying randomized trials in EMBASE), recommends two sources for filters. Please consult each source to decide which filter is most appropriate for your topic. 

  • Lefebvre et al. Filter

'crossover procedure':de OR 'double-blind procedure':de OR 'randomized controlled trial':de OR  'single-blind procedure':de OR (random* OR  factorial* OR crossover* OR cross NEXT/1 over* OR placebo* OR doubl* NEAR/1 blind* OR singl* NEAR/1 blind* OR assign* OR allocat* OR volunteer*):de,ab,ti

Source: Lefebvre C, Eisinga A, McDonald S, Paul N. Enhancing access to reports of randomized trials published world-wide - the contribution of EMBASE records to the Cochrane Central Register of Controlled Trials (CENTRAL) in The Cochrane Library. Emerging Themes in Epidemiology 2008; 5:13.

  • Wong et al. Filters

Sensitivity Maximizing Strategy
random*:ab,ti OR (clinical NEXT/1 trial*):de,ab,ti OR  'health care quality'/exp

Specificity Maximizing Strategy
(double NEXT/1 blind*):de,ab,ti OR placebo*:ab,ti OR blind*:ab,ti

Best Optimization of Sensitivity and Specificity
random*:ab,ti OR placebo*:de,ab,ti OR (double NEXT/1 blind*):ab,ti

Source: Wong SS, Wilczynski NL, Haynes RB. Developing optimal search strategies for detecting clinically sound treatment studies in EMBASE. Journal of the Medical Library Association 2006; 94(1): 41-47.

Note that the filters provided in Lefebvre et al.and Wong et al. are for use in Embase on the Ovid platform. The Cochrane Work Group has translated these filters for use in the platform, which is the platform you have access to at Hopkins.

Rarely, a review of observational studies will include a study filter. In these cases, check for validated filters before building your own.

Several organizations maintain lists of filters.

Note:  If you choose to use one of the Qualitative Study filters for PubMed, you will need to update the MeSH term. As of December 2015, “questionnaires”[mesh] is now “surveys and questionnaires”[mesh].

Evaluating the Quality of Search Strategies

  • Test to see if your search strategies are capturing gold-standard articles in the field.
  • Use the PRESS Checklist:
  1. Is the search question translated well into search concepts?
  2. Are there any mistakes in the use of Boolean or proximity operators?
  3. Are any important subject headings (i.e. controlled vocabulary terms) missing or have any irrelevant ones been included?
  4. Are any natural language terms or spelling variants missing, or have any irrelevant ones been included? Is truncation used optimally?
  5. Does the search strategy have any spelling mistakes, system syntax errors, or wrong line numbers?
  6. Do any of the limits used seem unwarranted or are any potentially helpful limits missing?
  7. Has the search strategy been adapted for each database to be searched?

Adapted from: McGowan J, Sampson M, Salzwedel DM, Cogo E, Foerster V, Lefebvre C. PRESS Peer Review of Electronic Search Strategies: 2015 guideline statement. J Clin Epidemiol. 2016 Jul;75:40-6.