EFFECTIVE MEDICAL

LITERATURE
SEARCHING

Sandra L. Martin, M.L.S., EdD

Knowledge Management and Eskind Biomedical Library
Objectives

1. Formulate your question

2. Understand basic database structure
3. Use of Boolean Logic
4. Use Field Searching
5. Use of Controlled Vocabulary
6. Specialty techniques (truncation, etc.)
7. Building your search strategy
 Reasons for Searching the Medical Literature

 To answer a specific patient case-related question

(clinical practice)
 To learn more about a medical topic (education)
 To determine current best practice (guideline)

To give the best possible care to patients using

evidence-based medicine
Information Overload
PubMed (Medline) Year # new
contains over 18 citations
million journal added*
citations going 2005 606,000
back to ~1950
2006 623,089

2007 670,943

2008 671,904

*statistics from http://www.nlm.nih.gov/bsd/bsd_key.html

Asking the Question: PICO
 Patient, Population or Problem
 What are the characteristics of the patient or population?
 What is the condition or disease you are interested in?
 Intervention or exposure
 What do you want to do with this patient (e.g. treat, diagnose,
observe)?
 Comparison
 What is the alternative to the intervention (e.g. placebo, different
drug, surgery)?
 Outcome
 What are the relevant outcomes (e.g. morbidity, death,
complications)?

http://healthlinks.washington.edu/ebp/pico.html
 Parsing the Question
 What are the main concepts in your question?
Sample question:
Does nutrition therapy improve decubitus (pressure)
ulcer healing in an elderly patient?
Concepts:

Nutrition therapy Decubitus/pressure ulcers

Treatment
Ulcer healing Elderly patients efficacy
A database is an organized
collection of data.

 Bibliographic databases are searched through many

of the same techniques as any other database
 Focus will be on PubMed, but these techniques
apply to ANY literature database
Examples of Records and Fields
Record # Author Title Publisher Date of
Publication

1 Jones KM Infectious Wiley 2001

Diseases

2 Smith BR Medicine Cambridge 2004

Univ. Press

3 Johnson Cancer Elsevier 2005

AS

4 Bradley PL Asthma Synergy 2003

Examples of Records and Fields
Record # Author Title Publisher Date of
Publication

1 Jones KM Infectious Wiley 2001

Diseases

2 Smith BR Medicine Cambridge 2004

Univ. Press

3 Johnson Cancer Elsevier 2005

AS

4 Bradley PL Asthma Synergy 2003

RED=RECORD BLUE=FIELD PURPLE = ONE PIECE OF DATA

 Searching a Database
Different search interfaces do the same things in slightly different ways

Good search interfaces should provide

 Ability to search for a specific item
 Ability to search for related items to a known item
 Ability to search in a specific field or fields
 Ability to combine search terms using Boolean Logic
 Ability to retrieve search results in a useful way
Boolean Logic
A British mathematician named George Boole (1815-
1864) developed an algebraic system of logic that is
now widely used in computer and electronic systems
including database searching.

While Boole’s algebraic system can be complex, a very

simple form of Boolean Logic is used for searching
most bibliographic databases.
Boolean Operators
Standard Boolean Logic for database searching uses
3 relationships among search terms.
AND
OR
NOT

It is both simple and powerful.

AND

BOTH terms included in any

results.

If a record has only one of the

two terms, it will not be
retrieved.
If the record has neither term, it
will not be retrieved.

What does this do to the amount

of records retrieved?
OR
Only one (NOT both) of
the terms are in the results

‘OR’ will retrieve the record

if both are included.

What does OR do to the

amount of records
retrieved?
NOT
Excludes any results
containing the term

Records containing both

will not be retrieved.

What does NOT do to

the amount of records
retrieved?
Using OR
“OR” groupings contain terms for the same
idea/concept and are usually put in parenthesis

(term OR term OR term)

where all terms are difference ways of representing the
same concept

(faculty OR teachers OR professors)

(students OR learners OR pupils)

Using AND
“AND” groupings contain terms for different
ideas/concepts and can combine OR groupings
Term AND (Term OR Term) where each represents a
different concept

heart attack AND smoking

Diabetes AND exercise

Cancer AND (treatment OR therapy)

Using NOT
“NOT” statements are usually put last and can
contain an “OR” grouping; they are often used to
get rid of a common subgroup

Students NOT dental

Diabetes NOT juvenile

Putting Them Together
1. Identify the concepts (Parse the question)
2. List specific terms for each concept
3. Put the terms for each concept in an OR
statements within parentheses
4. Combine OR statements with AND
5. Add any NOT statements to the end
Creating a Boolean Search

QUESTION: Is Vitamin C helpful in treating the flu?

1. Identify concepts and list terms

Concept 1 Concept 2 Concept 3 Concept 4

Influenza Vitamin C Treatment helpfulness
Influenza Vitamin C Treatment Outcome

Flu Ascorbic acid Therapy Recovery

Orange Juice Management Success

Step 2
2. Make your OR statements, one per concept
 (influenza OR flu OR orthomyxovirus)
 (vitamin C OR ascorbic acid OR ascorbate)
 (treatment OR therapy OR management)
 (outcome OR recovery OR success)
Steps 3 and 4
3. Put “AND” between each of the OR statements
(influenza OR flu) AND (vitamin C OR ascorbic
acid OR orange juice) AND (treatment OR
therapy OR management) AND (outcome OR
recovery OR success)

4. Consider any NOT statements you might want to

add.

Note: NOT isn’t used very often

Parsing a Boolean Search
(emergency OR acute OR critical) AND
(treatment OR therapy OR management OR
care) AND (motor vehicle accident OR car
crash) NOT (pedestrian OR walking)

What are the four concepts?

What terms are used for each concept?
Which three concepts must be included in all records found?
Which concept must not be included in any record found?
Beyond Basic Boolean
 Field Searching
 Controlled Vocabulary
 Subject vs. Keyword Searching
 Specialty Features
 Truncation
 Phrase searching
Field Searching
Almost all databases will provide you with some ability
to search a specific field or fields.
 Allows faster searching

 Allows more accurate searching

Not all databases may make all fields searchable.

Each search system will require a specific format.
 All Field vs. Specific Field Searches
I would like to find articles by John Smith.

 Search all fields: John Smith

 Search Author Field only: John Smith

I would like to find an article published in 1997.

 Search all fields: 1997

 Search Publication Date Field: 1997

Why waste time searching for a date in the author field or an

author in the volume field?
Formats for Field Searching
Different databases provide different formats for
specifying fields.

Most use field names or nicknames

 Field ‘tags’ OR ‘labels’

which may follow a period or be placed in brackets or

parentheses.

Some databases offer forms or drop-down menus.

PubMed
Field tags go in [] and follow term
Field tags can be used within Boolean queries
PubMed Field Tags
http://www.ncbi.nlm.nih.gov/books/bv.fcgi?rid=helpp
ubmed.section.pubmedhelp.Search_Field_Descrip

[au] = author [dp] = date of publication

[ti] = title [la] = language
[tw] = textword [gr] = grant number
[tiab] = title and abstract [ta] = journal name
[mh] = medical subject [ad] = affiliation
heading
OVID databases require field tags to follow the term
separated by a period.

Meharry.in and 2005.yr

 Some databases, such as the ISI’s Web of Science
(Science Citation Index Expanded) provide forms to
fill out.

Select Field
from drop-
Select down menu
Boolean
Operator
 Combining Field Searches
Multiple field searches can be combined using Boolean
logic.

Find a 2005 article by an author named Hubble about

ankle fractures.

Combine with AND:

 2005 in date/year field

 Hubble in author field

 Ankle fractures in title field

2005[dp] AND Hubble[au] AND ankle fractures[ti]

Final Notes on Fields
 Each database provides its own specific fields
 Each database requires a specific format to
designate field searching
 When searching a new database, take a moment to
read the help documentation; most will provide a
list of fields and how to search them.
 Controlled Vocabulary
A controlled vocabulary is a set of established terms
where
 every term represents a single concept

 only one term is used for that concept

Another example
 How many words could you think of for the idea of
“cancer”?
 Cancer, tumor, malignancy, neoplasm, sarcoma…

 Articles in a database
 Article one: “Breast tumors in young women”
 Article two: “Surgery for prostrate cancer.”
 Article three: “Diagnosing Melanoma.”

All three articles are about types of cancer but different

terms are used in titles.
In a controlled vocabulary ONE word (i.e., cancer) is chosen and
placed in a special field, usually called a subject field.

For all three articles

 Article one: “Breast tumors in young women”
 Article two: “Surgery for prostrate cancer.”
 Article three: “Diagnosing Melanoma.”
The subject term (concept term) “cancer” is placed in the subject
field by database indexers.

Now, Searching the database for cancer in the subject field will
identify all records about the concept of cancer even if a
different word for cancer is used.

 Search ‘cancer-in-subject-field’ finds all three articles

 How many articles would the search ‘cancer-in-title-field’
find?
 Structure of Controlled Vocabulary

Broader
Concepts

Narrower
Concepts
More on Controlled Vocabulary
 “Expanding” = Search includes all narrower terms
beneath the searched term
 Some databases do it automatically, others don’t
 “Focusing” or “Majoring” = For a given item, some
subject terms are considered the major focus; you can
select to return only those articles.
 Hip fracture[majr] = only give me articles where hip
fracture is an important concept
 Subject vs. Keyword Searching
Controlled vocabulary Free-text (keyword) searching
searching
 Some concepts have many
 Matches terms against a synonyms. A free-text search
specific field in the record. statement would mean "OR"ing
 You need to consult a all those terms together
thesaurus (paper or online) to
find out what the controlled
 Matches terms against words
vocabulary term is for each anywhere in record (abstract,
concept. title, etc.).
Advantages to Controlled
Vocabularies
 Using the controlled vocabulary can make your search more
precise and easier.
 Increases the relevancy of results (fewer false drops)
 The indexers have already done much of the work for you.
 Searchable tree structures of terms can help you find new
terms to use.
 Problems with Controlled
Vocabularies
 NOT all databases use a controlled vocabulary
 New concepts take time to be added
 There is often a lag phase during which the newest
articles aren’t indexed
 Controlled vocabularies can contain some very
strange things and some concepts may not be handled
well
 The controlled vocabulary must be easily searchable

Trying to understand what is and isn’t in a particular

controlled vocabulary can give you a big headache!
 Combining Subject and Keyword
Searching
To be comprehensive, it is often helpful to combine
subject and keyword searching

(diabetes mellitus[mh] OR diabetes[tw])

(sickle cell anemia[mh] OR sickle cell anaemia[ti])

Don’t forget…
 Boolean logic to combine terms
 Use of other search fields in combination with
subject terms

A Complex Search:
(head[mh] OR head[tw]) AND (wound and
injuries[mh] OR trauma[ti] OR injury[ti]) AND
2005[dp] AND English[la]
Some Specialty Features
 Truncation
 Phrase searching
 Neighboring and other rarer Boolean operators
 Truncation
What about including the singular and plural versions of words as
well as other word variations?
For example: therapy, therapies, therapeutics,

You could combine them all in an OR relationship:

(therapy OR therapies OR therapeutics OR therapeutic)

But an easier way is by the use of truncation.

therap*

Each database handles truncation in a unique way.

The ‘*’ and ‘$’ are the most common wildcard symbols.
 More on Truncation
Some examples:
Bacter$
Proc*
Vir?
Staph?

Be cautious when truncating!

If the word stem is too short, there may be too many possible
variations and you might pick up unrelated terms.

For example, using proc* for finding procaine-like drugs will

also include words like proceedings and process.
Phrase Searching
 Sometimes you want to force the database to search
for a set of words in exact order
“fever of unknown origin”

Most databases will accept a phrase in quotes.

BUT…some do not handle phrases well and will
automatically break them up – usually ‘AND’-ing the
terms
Check how the database handles phrase searching
before doing it!
Limits Options
 Many databases provide “limits” pages that make it
easier for you to select common options such as
language, article type, publication dates, human or
animal, gender, age groups, etc.
 Each database’s limits options are unique
 Most limits can be done ‘by hand’ using field tags,
but sometimes limit pages save time
PubMed Limits Page
Step-By-Step Search Construction
1. State the question
2. Identify the concepts in the question
3. For each concept, determine keywords and subject terms
4. Specify field tags after terms if needed
5. Combine terms for the same concept with “OR” in
parenthesis
6. Combine “OR” statements with AND
7. Put any NOT terms at the end

Keep track of your searches, how many articles were found total,
and how many you selected as relevant
Example
 Question: What is the appropriate ED medical
management of adult patients with intracranial
hemorrhage (either trauma or spontaneous)?