Searching for the Best Evidence

Searching for the Best Evidence in Clinical Journals

This section is about journal searching and we start with a reminder that there are other resources that already summarise the results of such searches, incorporating quality filters and critical appraisals. Examples are the Cochrane Library and the ACP Journal Club/EBM.

These handy hints for maximising the efficacy of your searching were developed for use at the 3rd UK Workshop in Teaching Evidence-Based Medicine by Robin Snowball of the Cairns Library and Carol Lefebvre of the UK Cochrance Centre. These hints are aimed at people using Medline to search for information.

How to get the right stuff and avoid getting the wrong stuff
Methodologic Filters for getting more right stuff, specifically for:

How to Get the Right Stuff and Avoid Getting the Wrong Stuff!

aka Sensitivity and Specificity

In searching, these two terms have an analogous meaning to the way they are employed in diagnosis. Specificity refers to the proportion of documents which you retrieve which are relevant, while sensitivity refers to the proportion of all the documents wh ich are relevant that your search manages to retrieve. (These terms are also sometimes called precision and recall respectively).

Put another way:

Sensitivity is the likelihood of retrieving relevant items;
Specificity is the likelihood of excluding irrelevant items.

If you're doing a search and it yields an unmanageably large number of hits, you probably need to increase the specificity of your search; conversely, if you get too small a number of hits, you probably need to increase sensitivity.

Increase specificity by:

narrowing your question!
using more specific terms in Free Text search
using a Thesaurus/Subject search rather than Free Text/Textword search
using more specific Thesaurus/Subject headings
selecting specific subheadings with Thesaurus/Subject MeSH headings
adding in terms (using AND) to represent other aspects of the question
limiting to language of article, to human or animal subjects, to publication types (e.g. randomised controlled trials, reviews, etc.), to country or year of publication.

Increase sensitivity by:

broadening your question!
finding more search terms from relevant records
trying different cobinations of terms
using truncation (* or $) or wildcard (?) in Free Text/Textword OR Thesaurus/Subject
adding in and combining terms of related meaning using OR
using NEAR to retrieve items in the same sentence in either order
using a combination of Free Text/Textword and Thesaurus/Subject searches
using explosion feature of Thesaurus searches
selecting All Subheadings with Thesaurus/Subject MeSh terms
searching further back in time

Methodologic Filters

These searching techniques are separated into different tips for Ovid and for SilverPlatter (inc WinSpirs) users. The Filters (developed by Anne McKibbon) are search strategies designed to retrieve high quality evide nce from published studies appropriate to decision-making (e.g. clinical trials for diagnosis, cohort studies for prognosis, etc.).

It is important to realise that these are strategies to be tried in combination with a comprehensive (i.e. sensitive) subject search strategy which has first been used to retrieve as much information as possible releva nt to the subject.

Randomised Controlled Trials and Systematic Reviews

WINSPIRS version:

Enter your subject terms first and combine where appropriate. You can then limit your results to randomised controlled trials (RCTs) and controlled studies using the following strategy:
1. RANDOMIZED-CONTROLLED-TRIAL in PT (use LIMIT)
2. META-ANALYSIS in PT (use LIMIT)
3. CONTROLLED-CLINICAL-TRIAL in PT (use LIMIT)
4. CLINICAL-TRIAL in PT (use LIMIT)
5. random* in ti,ab,mesh
6. (meta?anal* or meta analy*) in ti,ab,mesh
7. (doubl* or singl*) and blind* in ti,ab,mesh
8. explode CLINICAL-TRIALS (use THESAURUS)
9. crossover in ti,ab,mesh
10. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9
  To increase the sensitivity of the above search you may consider adding the following:
11. (clin* trial*) in ti,ab,mesh
12. control* near (trial* or stud*) in ti,ab,mesh
13. (singl* or doubl* or tripl* or trebl*) near (blind* or mask*) in ti,ab,mesh
14. placebo* in ti,ab,mesh
15. RESEARCH-DESIGN (use THESAURUS)
16. TG=COMPARATIVE STUDY (use LIMIT Checktags)
17. 10 or 11 or 12 or 13 or 14 or 15 or 16
You may limit your results to Systematic Reviews using the following strategy:
1. REVIEW-ACADEMIC in PT (use LIMIT)
2. REVIEW-TUTORIAL in PT (use LIMIT)
3. systematic* near (review* or overview*) in ti,ab,mesh
4. (meta?analy* or meta analy*) in ti,ab,mesh,p
5. 1 or 2 or 3 or 4 or 5

Ovid Version

Enter your subject terms first and combine where appropriate. You can then limit your results to randomised controlled trials (RCTs) and controlled studies using the following strategy:
1. RANDOMIZED-CONTROLLED-TRIAL in PT (use LIMIT)
2. META-ANALYSIS in PT (use LIMIT)
3. CONTROLLED-CLINICAL-TRIAL in PT (use LIMIT)
4. CLINICAL-TRIAL in PT (use LIMIT)
5. random$ (Textword)
6. (meta-anal$ or metaanaly$ or meta analy$) (Textword)
7. (doubl$ or singl$) and blind$ (Textword)
8. explode CLINICAL-TRIALS (use SUBJECT)
9. crossover in ti,ab,mesh (Textword)
10. 1 or 2 or 3 or 4 or 5 or 6 or 7 or 8 or 9 (Combine)
  To increase the sensitivity of the above search you may consider adding the following:
11. (clin$ trial$) (Textword)
12. control$ and (trial$ or stud$) (Textword)
13. (singl$ or doubl$ or tripl$ or trebl$) and (blind$ or mask$) (Textword)
14. placebo$ (Textword)
15. RESEARCH-DESIGN (use SUBJECT)
16. COMPARATIVE STUDY (use SUBJECT)
17. 10 or 11 or 12 or 13 or 14 or 15 or 16 (Combine)
You may limit your results to Systematic Reviews using the following strategy:
1. REVIEW-ACADEMIC (use LIMIT)
2. REVIEW-TUTORIAL (use LIMIT)
3. systematic$ and (review$ or overview$) (Textword)
4. (meta?analy* or meta analy*) (Textword)
5. 1 or 2 or 3 or 4 or 5 (Combine)

Diagnosis

Type in each line of the search strategy separately; combine the separate lines with AND or OR in the usual way at the end of the strategy.

WINSPIRS version:
Hyphenated terms, or terms with 'exp': Thesaurus search;
Other terms: type as given in Free Text

Best one-term strategy:
- sensitivity in ti,ab,mesh
Maximum sensitivity strategy:
- sensitivity-and-specificity or
- sensitivity in ti,ab,mesh or
- diagnosis in mesh or
- radionuclide imaging in mesh or
- diagnostic use in mesh or
- specificity in ti,ab,mesh
Maximum specificity strategy:
- exp sensitivity-and-specificity or
- (predictive and value*) in ti,ab,mesh

Ovid Version

(T) = Textword search; (S) = Subject (Thesaurus) search; (L) = Limit;
.fs. = &QUOTfloating subheading&QUOT: a subheading attached to any MeSH term in the record; subheadings are abbreviated: e.g. du = diagnostic use; type as given, in Subject ; (see WinSpires strategy above for subheading in full form).

Best one-term strategy:
- sensitivity (T)
Maximum sensitivity strategy:
- sensitivity-and-specificity (S) or
- sensitivity (T) or
- di.fs or
- ri.fs. or
- du.fs. or
- specificity (T)
Maximum specificity strategy:
- exp sensitivity-and-specificity (S) or
- (predictive and value$) (T)

Therapy

Type in each line of the search strategy separately; combine the separate lines with AND or OR in the usual way at the end of the strategy.

WINSPIRS version:
Hyphenated terms, or terms with 'exp': Thesaurus search;
Other terms: type as given in Free Text

Best one-term strategy:
- clinical-trial in pt
Maximum sensitivity strategy:
- randomized-controlled-trial in pt
- drug therapy in mesh or
- therapeutic use in mesh or
- random* in ti,ab,mesh
Maximum specificity strategy:
- (double and blind*) in ti,ab,mesh or
- placebo* in ti,ab,mesh

Ovid Version

Best one-term strategy:
- clinical-trial (L)
Maximum sensitivity strategy:
- randomized-controlled-trial (L) or
- dt.fs or
- tu.fs. or
- random$ (T)
Maximum specificity strategy:
- double and blind$ (T) or
- placebo$ (T)

Aetiology, Causation or Harm

Type in each line of the search strategy separately; combine the separate lines with AND or OR in the usual way at the end of the strategy.

WINSPIRS version:
Hyphenated terms, or terms with 'exp': Thesaurus search;
Other terms: type as given in Free Text

Best one-term strategy:
- risk in ti,ab,mesh
Maximum sensitivity strategy:
- exp cohort-studies
- exp risk or
- (odds and ratio*) in ti,ab,mesh or
- (relative and risk) in ti,ab,mesh or
- (case and control*) in ti,ab,mesh
Maximum specificity strategy:
- case-control-studies or
- cohort-studies

Ovid Version

Best one-term strategy:
- risk (T)
Maximum sensitivity strategy:
- exp cohort-studies (S) or
- exp risk (S) or
- odds and ratio$ (T) or
- relative and risk (T) or
- case and control$ (T)
Maximum specificity strategy:
- case-control-studies (S) or
- cohort-studies (S)

Prognosis, Natural History

Type in each line of the search strategy separately; combine the separate lines with AND or OR in the usual way at the end of the strategy.

WINSPIRS version:
Hyphenated terms, or terms with 'exp': Thesaurus search;
Other terms: type as given in Free Text

Best one-term strategy:
- exp cohort-studies
Maximum sensitivity strategy:
- incidence in mesh or
- exp mortality or
- follow-up studies or
- mortality in mesh or
- prognos* in ti,ab,mesh
- predict* in ti,ab,mesh
- course in ti,ab,mesh
Maximum specificity strategy:
- prognosis in mesh
- survival-analysis

Ovid Version

Best one-term strategy:
- exp cohort-studies (S)
Maximum sensitivity strategy:
- incidence (T) or
- exp mortality (S) or
- follo-up-studies (S) or
- mo.fs. or
- prognos$ (T) or
- predict$ (T) or
- course (T)
Maximum specificity strategy:
- prognosis (T) or
- survival-analysis (S)