Using
MEDLINE to find Evidence Based Medicine
MAJ John Heflin, MC, USA
(edited for web page by CDR DL Hufford, MC,USN, 03 JUL 99)
Faculty Development Fellowship
Madigan Army Medical Center
Goal: To be able to utilize MEDLINE to retrieve high quality evidence based articles which are appropriate for clinical decision making
The Basics Terms of MEDLINE:
MEDLINE – A computerized database maintained by the National Library of Medicine(NLM). All search engines (Internet Grateful Med, HealthGate, OVID, Paperchase, etc.) use either the NLM MEDLINE database or a private copy. The NLM updates the MEDLINE database weekly but not all vendors update weekly.
Medical Subject Heading (MeSH) – A catalog of subject headings which are used by the NLM to index articles within MEDLINE and Index Medicus. Every article is reviewed by the NLM librarians and assigned MeSH headings which represent the topics covered in the article. Each MeSH heading is also indexed as to whether it is a major focus of the article (Major MeSH) or a minor focus (Minor MeSH).
Textword (Free Text) – Literally searches the title and/or abstract of the articles within MEDLINE for an exact match to the words listed. Some articles will only be found with a Textword search because the medical topic does not have an official MeSH heading or the article wasn’t correctly cataloged for MeSH.
Boolean Operations – Used to combine search terms
AND – the articles selected need to have both terms
OR – the articles selected need to have either of the terms
NOT – the articles selected should not have this term
Sensitivity – The proportion of all of the documents pertaining to the subject that the search manages to retrieve. "Does it retrieve all of the documents which are relevant to the subject?"
Specificity – The proportion of the retrieved documents which are relevant to the subject.
"Does it retrieve only documents which are relevant to the subject?"
Choice of MEDLINE search engine:
Differences pertain to the cost, ease of use, interface style, available options, efficacy of the search engine and the comprehensiveness of the MEDLINE database. No one service consistently outperforms all others.
I recommend HealthGate (http://www.healthgate.com) and Internet Grateful Med (http://igm.nlm.nih.gov/) as they are free, easy to use, very effective, have numerous options available and both access a comprehensive MEDLINE database.
- OVID (http://www.ovid.com/), BioMedNet (http://biomednet.com/), and Knowledge Finder (http://www.kfinder.com/) offer a different interface style but an annual subscription fee is required with OVID and Knowledge Finder.
Sensitivity vs. Specificity:
High sensitivity and specificity is desired but they are often inversely related - as one rises the other falls
If you get too many hits, increase the specificity (narrow the question)
- Use more specific terms for the Textwords
- Use MeSH terms rather than Textwords
- Use more specific MeSH terms
- Add additional search terms with the AND function
- Limit the search via the publication language (English, German, etc.), gender, age groups, study groups (human vs. animals), publication years, article/publication type (randomized controlled trials, reviews, etc.), and journal type (all, medical, nursing, or dental) tools
If you get too few hits, increase the sensitivity (broaden the question)
- Use more search terms
- Use the truncation or wildcard ($, *) operators with Textword searches
- Search with both Textword and MeSH terms
- Add additional search terms with the OR function
- Use the Explode function (if available) to find articles in the MeSH tree which are at levels below the current term
- Select ALL Subheadings for the MeSH terms
- Increase the number of years searched
Methodological Filters: Search strategies designed to retrieve high quality evidence based articles. These strategies are added to the basic search.
Clinical Trials:
- Enter the search terms (Textword or MeSH as appropriate) in the usual fashion and combine (AND, OR, NOT) as appropriate.
- Use the article type (publication type) drop-down list to select a type of clinical trial which will limit the search results to clinical trials.
Choices
Available in HealthGate Choices
Available in IGM
Clinical Trial Clinical Trial
Clinical Trial, Phases I - IV Meta-Analysis
Controlled Clinical Trial Randomized Controlled Trial
Randomized Controlled Trial
Multicenter Study
Meta-Analysis
Note:
o Clinical Trial retrieves all articles contained within the headings: Controlled
Clinical Trial; Randomized Controlled Trial; and Clinical Trial, Phase I – IV.
Multicenter Study is an independent heading
Meta-Analysis is an independent heading
Systematic Reviews:
Enter the search terms (Textword or MeSH as appropriate) in the usual fashion and combine (AND, OR, NOT) as appropriate.
- Use the article type (publication type) drop-down list to select a type of review article which will limit the search results to review articles.
Choices
Available in HealthGate Choices
Available in IGM
Review Review
Review Literature Meta-Analysis
Review, Academic
Review, Tutorial
Review of Reported Cases
Consensus Development Conference
Meta-Analysis
Note:
o Review retrieves all articles contained within the headings: Review Literature;
Consensus Development Conference; Review, Academic; Review, Tutorial; and Review of Reported Cases
- Review Literature will retrieve all articles in the headings: Review,
Academic; Review, Tutorial; and Review of Reported Cases.
Meta-Analysis is an independent heading
Diagnosis:
- Enter the search terms (Textword or MeSH as appropriate) in the usual fashion.
- Use the AND function to add any of the following search terms: Sensitivity, Sensitivity and Specificity, Diagnosis, Diagnostic Use, Diagnostic Imaging, Radionuclide Imaging, Predictive Value of Tests
- For a quick search, "Sensitivity and Specificity" is the best single term to use
- Maximum sensitivity is obtained by simultaneously using and combining with OR the following terms: Sensitivity, Sensitivity and Specificity, Diagnosis, Diagnostic Use, Diagnostic Imaging, Radionuclide Imaging
- High specificity is obtained by using the term Sensitivity and Specificity
Therapy:
- Enter the search terms (Textword or MeSH as appropriate) in the usual fashion.
- Strongly consider use of the AND function to add any of the following search terms: Therapy, Drug Therapy, Therapeutic Use
- Maximum sensitivity when Therapy is combined via OR with Therapeutic Use
- Use the article type (publication type) drop-down list to select a type of clinical trial which will limit the search results to clinical trials.
Note:
Clinical Trial retrieves all articles contained within the headings: Controlled
Clinical Trial; Randomized Controlled Trial; and Clinical Trial, Phase I – IV.
Multicenter Study and Meta-Analysis are independent headings
Etiology, Causation, Harm:
- Enter the search terms (Textword or MeSH as appropriate) in the usual fashion.
- Use the AND function to add any of the following search terms: Risk, Cohort Studies, Case-Control Studies, Odds Ratio, Causality
- For a quick search, "Risk" is the best single term to use
- Maximum sensitivity is obtained by simultaneously using all of the terms and combining with OR
- High specificity when Case-Control Studies is combined via OR with Cohort Studies
Prognosis:
- Enter the search terms (Textword or MeSH as appropriate) in the usual fashion.
- Use the AND function to add any of the following search terms: Cohort Studies, Prognosis, Incidence, Mortality, Survival Analysis
- For a quick search, "Cohort Studies" is the best single term to use
- Maximum sensitivity is obtained by simultaneously using and combining with OR the following terms: Incidence, Mortality, Prognosis, Course
- High specificity when Prognosis is combined via OR with Survival Analysis
