Friday, April 29, 2005

MeSH: Alphabetical List

The Medical Subject Headings (MeSH) are the basis of the MEDLINE database...and they are what allow the searcher to quickly gather a set of citations that are clearly about specific concepts in your research or clinical question.

MeSH is organized in two ways - alphabetically and hierarchically. This post examines the alphabetical list.

Even though most of us never see the printed alphabetical list because we use the electronic versions of MEDLINE (Ovid MEDLINE or PubMed), it's helpful to take a look at the printed list. All of sudden some of those choices you're faced with as you're searching make more sense. It's a building block...

Here's the entry for VIOLENCE:

In the Alphabetical List:
  • Descriptors (main headings): characterize the subject matter or content
  • Entry terms (see references): synonyms or closely related terms appearing that are cross-references to descriptors
  • See Related: indicate the presence of other descriptors related to the topic conceptually
  • See: indicate that information related to one term will be found under a different term.

Wednesday, April 27, 2005

MeSH: Medical Subject Headings

Some people think that MeSH is old hat...that using free text searching is much more effective than utilizing this controlled vocabulary. But that's not true...using MeSH is the most effective and efficient way to search MEDLINE.

Some reasons why:
  • The problem of synonyms is taken care when you use the appropriate MeSH term (see Controlled Vocabularies...hmmm and Controlled Vocabularies: why should I care?)
  • Its heirarchical organization allows instantaneous selection of group concepts (more about this later)
  • Utilizing the 'Focus' allows the searcher to designate which concept in their search question has the highest priority (more later)
  • By coordinating subheadings with a MeSH term, a searcher can pinpoint the specific aspect(s) of the concept that is most important (more later)
  • By judiciously selecting MeSH terms, applying appropriate subheadings and combining sets with Boolean operators, you'll quickly identify a group of relevant citations.
With a bit of a time investment to understand MeSH and some practice, I'll bet you can improve the quality of your searches.

I'll bet you didn't know that MeSH has its own homepage! Yup, it makes this librarian smile...

Tuesday, April 26, 2005

Database Highlight: Natural Standard database

The Natural Standard database, founded by clinicians and researchers, provides high quality, evidence-based information about complementary and alternative therapies to clinicians, patients, and healthcare institutions. The Natural Standard database provides information on herbs & supplements, conditions, alternative modalities and includes a dictionary. Comprehensive, evidence-based monographs are available for each therapy containing information on scientific evidence, theory, dosing, safety and interactions.

The international research team utilizes a comprehensive methodology and grading scales to create the evidence-based, consensus-based, and peer-reviewed monographs. They regularly monitor scientific literature and industry warnings. When clinically relevant new data emerge, best efforts are made to update content immediately. In addition, regular updates with renewed searches occur every 3-18 months, variable by topic.

You can find this database under 'All Major Resources A-Z' from the Medical Library's home page.

Monday, April 25, 2005

Proximity Operators Correction

I’m making it correction to last week’s post “Proximity Operators” (April 15), specifically about the Ovid ‘adjacency operator’.

Even as I wrote about the adjacency n (adjn) operator, I was fairly certain that it was wrong and against my better judgment, I went ahead with the post.

As I write each entry, I check and double check to make certain that what I say is correct. And I discovered a discrepancy between the information published in Ovid’s Online MEDLINE Field Guide and how I have been teaching and utilizing the adjacency operator for years. I checked Ovid’s Advanced Searching printed brochure and it corroborated what I believed to be true. What to do? I felt that because the online Field Guide was more recent information that I would believe that and I posted the information on the ‘adjacency operator’. At the same time, I wrote to Ovid to say that I thought the Online Field Guide was incorrect and asked for a definitive answer on the adjacency n operator. I heard from Ovid Support on Friday, April 22. And, yes, the Online Field Guide is incorrect.

Here’s what I originally reported about the ‘adjn’ operator:

blood adj2 clot$ - retrieves records that contain both terms, in order, with two words between

Now, here’s the correct information:

blood adj2 clot$ - retrieves records that contain both terms, in any order, with two words between

And here’s why it’s so important – because when you’re looking for a phrase or for words, you want to be able to compensate for the order in which the phrase or words were used. Keep in mind that the closer the words are together the tighter the relationship between them; the further they are apart, the looser the relationship. So when using adjn, try to keep the n somewhere between 2 and 7 – the higher it is, the greater the likelihood of retrieving irrelevant results.

I’ve made the correction to the original post.

Thursday, April 21, 2005

Database Highlight: GIDEON

GIDEON: The Global Infectious Disease and Epidemiology Network

GIDEON is a global infectious disease knowledge management tool and is made of four modules: Diagnosis, Epidemiology, Therapy and Microbiology. The database includes 337 disease, 224 countries, 1,147 microbial taxa and 306 antibacterial (-fungal, -parasitic, -viral) agents and vaccines.

The tool provides:
  • a reliable diagnosis or simulation of clinical problems based on patient presentation and epidemiology
  • a ranked differential diagnosis that considers: patient symptoms, travel and exposure history, incubation period, laboratory test results, immune status, bio-terrorism
  • simultaneous comparison of clinical and laboratory features for two or more diseases
  • an explanation of why a certain disease is not included in the differential diagnosis
  • up-to-date epidemiological information including agent, vector, vehicle, and resevoir

Check it out!

You can find this database under 'All Major Resources A-Z' from the Medical Library's home page.
Updated weekly

Wednesday, April 20, 2005

Wildcards, #$*!?...

Wildcards are slightly different than truncation symbols -- although many people use the terms interchangeably. Here's my definition...

A truncation symbol allows you to truncate a word at any point after the first three characters. This is what you use to ensure that you are retrieving variations of word endings. For example, retin$ retrieves retina, retinas, retinitis, etc.

Wildcards usually replace one, two, or no characters. For example, wom#n retrieves both women or woman.

To find out what wildcards are supported in your favorite database or search engine, check the HELP!

For Ovid
  • # - retrieves records that contain the search term with substituted character(s) [wom#n retrieves women or woman]
  • ? - retrieves records that contain the search term with either no characters substituted, or with substituted character(s) [colo?r retrieves color or colour]
For Web of Science
  • * - used in the middle of a word searches for terms with alternate spellings [sul*ur retrieves sulfur or sulphur]
  • ? - used for one character [Barthold? retrieves Bartholdi or Bartholdy]
  • $ - used for one or no charachers [Barthold? retrieves Barthold, Barholdi, or Bartholdy]; also use the $ to find British or American spellings of the same word [vapo$r retrieves vapor or vapour]
For PubMed
  • Unfortunately, there are no wildcards

Tuesday, April 19, 2005

New in the ISI Web of Knowledge

ISI announced these new enhancements to the ISI Web of Knowledge:

  • Links to full text within results summary lists
  • "Search within results," available on search summary pages; allows you to modify an original search without having to start from scratch
  • One-click search from the full record; you will be able to start another search by clicking on the hyperlinked terms within certain fields
  • Access to the full title of the cited article when users roll over the "View Record" link in cited reference lists within Web of Science
  • Ability to reference a set number in a query
  • Additional field - Country/Territory - within the Analyze tool provides another way for you to refine and analyze search results (available within Web of Science)

Monday, April 18, 2005

Proximity Operators [Amended April 24, 2005]

Many search interfaces support proximity operators as well as the Boolean operators AND, OR and NOT. Proximity operators allow you to search for words occurring in the same part of the record or no more than a certain distance apart or with a certain frequency.

Each database will support different proximity operators and more importantly, will define them differently so before you use any make sure you know what you're doing.

Examples of proximity operators are: NEAR, ADJACENCY, SAME, WITH, FREQUENCY
And here's a few definitions:

Ovid: ADJ (adjacency) and FREQ (frequency)
  • ADJ retrieves records that contain both terms, in order, and adjacent in the same sentence. [blood clot - adjacency is the default operator; blood adj2 clot$ - retrieves records that contain both terms, in any order, with two words between]
  • FREQ retrieves records that contain n occurrences of the term in the specified field. [ blood.ab./freq=5 - retrieves records containing the word "blood," at least five times in the Abstract (ab) field].
Web of Science : SAME
  • SAME operator and the definition says that all terms separated by SAME must appear in the same subfield. [laser* SAME gas finds records containing laser or lasers and gas in the title, the same sentence in the abstract, or the same keyword phrase.]
Where do you find this information? In the HELP, of course!

Ovid supports AND, OR, NOT, ADJ, FREQ
Web of Knowledge supports AND, OR, NOT, SAME
Silverplatter (WebSPIRS) supports AND, OR, NOT, ADJ, NEAR, WITH
PubMed supports AND, OR, NOT

Friday, April 15, 2005

More about NOT

This Venn diagram illustrates the use of the Boolean operator NOT -- the colored area represents the result of the Boolean operation.

In general, NOT finds records that contain one term but not the other. My example is using mixed Boolean operators. I'm searching for AIDS (the disease concept) but I want to make sure that I don't pick up records that have to do with either 'visual aids' or 'hearing aids'.

And here's how the search strategy would look using Ovid MEDLINE:

1. aids
2. hearing
3. visual
4. 1 not (2 or 3)

  • In this search, I did not use MeSH terms; instead, I did textword searches.
  • Caution! Be careful how you use NOT. You could potentially exclude relevant articles.

Thursday, April 14, 2005

More about OR

This Venn diagram illustrates the use of the Boolean operator OR -- the colored area represents the result of the Boolean operation.

In this case, two concepts were combined using the Boolean OR and the result is a large set containing one or both of the terms.

And here's how the search strategy would look using Ovid MEDLINE:

1. coffee (4770 records)
2. caffeine (14310 records)
3. 1 or 2 (17991 records)

Multiple OR operators can be used to combine more than two terms.
OR is most frequently used to combine synonyms.

Wednesday, April 13, 2005

More about AND

This Venn diagram illustrates the use of the Boolean operator AND -- the colored area represents the result of the Boolean operation.

In this case multiple AND operators were used to combine more than one concept or term. Only records containing all the terms will be retrieved.

And here's how the search strategy would look using Ovid MEDLINE:

1. Diabetes Mellitus, Type II (33995 records)
2. obesity (59850 records)
3. child (1035163)
4. 1 and 2 and 3 (315 records)

Note: I used MeSH terms in this example.

Tuesday, April 12, 2005

ANDs and ORs and NOTs, oh my!

Boolean Operators and Venn Diagrams

Many bibliographic databases use Boolean operators as the building blocks for the construction of search strategies, from the simplest to the most complex. Boolean operators should bring back memories of your high school algebra days where together with Venn diagrams, the world of the algebraic equation was illuminated.

AND, OR, and NOT are the Boolean operators we’re concerned with and I’m going to use a Venn diagram to illustrate what is meant.

Boolean operators are used to combine sets. In our case, the sets are made up of search terms. Using Boolean operators, we can construct complex equations.

My Venn diagram shows all the possible relationships for two sets – Set 1 and Set 2:

A – the records found exclusively in Set 1
B – the records found exclusively in Set 2
C – the records found in both Set 1 and Set 2
D – the records not found in either set

As a searcher, you can use these operators to indicate how you want the sets of subject terms you’ve created to be combined.

AND retrieves records containing all the terms
OR retrieves records containing one or more of the terms
NOT retrieves records containing one term but not the other

Monday, April 11, 2005

Database Highlight: EMBASE

This comprehensive pharmacological and biomedical database is renowned for extensive indexing of drug information from over 3500 biomedical journals from 70 countries. EMBASE is one of the databases delivered to us on the Ovid platform and this version covers 1980 to the present, with approximately 350,000 records added annually. These records reflect all current developments in biomedical and drug-related fields. Because of this, it is critically important to search EMBASE, especially when you need to be comprehensive.


• Drug Research, Pharmacology, Pharmacy, Pharmacoeconomics, Pharmaceutics and Toxicology
• Human Medicine (Clinical and Experimental)
• Basic Biological Research
• Health Policy and Management
• Public, Occupational and Environmental Health
• Substance Dependence and Abuse
• Psychiatry
• Forensic Science
• Biomedical Engineering and Instrumentation

If you're having difficulty searching this database, try asking your friendly librarian. Chances are pretty good she can provide some in-depth help.

Just the Facts
Available from Ovid
Updated weekly
1980 to present
Indexing and abstracting service
Produced by Excerpta Medica, Elsevier
EMBASE (Ovid Field Guide)

Friday, April 08, 2005

Trying a new database?

Here are a few things that I always check before I start searching a new database or using a new interface or search engine:

1. Is the database based on a controlled vocabulary?
This is important because it immediately determines what my approach is going to be. If there is a controlled vocabulary, then I'll want to take advantage of that by letting the system map whatever I type to the appropriate vocabulary word for that particular database. For example, several years ago I was looking for information on surrogate mothers in a particular database and wasn't retrieving as much as I expected. I allowed the system to map my phrase 'surrogate mothers' to the vocabulary term commonly used in this discipline and 'host mothers' came up. I continued to search on the appropriate term and a world of information was revealed.

If the system isn't based on a controlled vocabulary, then I need to do alot of up front preparation.

2. What's the default operator?
This is especially important in search engines on the web. I want to know exactly how the system is going to treat the phrase that I type on the query line. Is it going to assume adjacency, put an AND between the words, or put an OR between the words?

3. What's the truncation symbol? What are the wildcards?
Critical! When I start to free text search, I need to be able to account for variations in words and word endings...if I don't know what these symbols are, I'm doomed.

4. Where's the HELP?
Well, good golly, how can you not want to know where the HELP is? This is where you find out why you keep getting those inexplicable results. Hints and tips galore are usually a click away. But, my colleagues tell me that I'm not normal when it comes to HELP. What can I say...I'm the one who reads the manual cover to cover, too.

Wednesday, April 06, 2005

STOP words

Stop words are common words that databases or search engines simply ignore while running a search. These words have little intrinsic meaning and usually modify other words - such as adverbs, prepositions and conjunctions. Some examples of stop words are: almost, with, approximately, it, be, then, these.

Some databases provide ways to force the system to search for the word or the word in a particular phrase. Check the HELP to identify the appropriate syntax in your favorite database or search engine.

PubMed Stop Words
Ovid MEDLINE Stop Words
Google Stop Words
Web of Science Stop Words

Tuesday, April 05, 2005

Controlled Vocabularies - why should I care?

Look at the following journal article titles and their associated MeSH terms that I've selected from MEDLINE.

Notice the authors have used different ways to describe kidney calculi -- kidney stone, kidney stones, renal stone, renal stones. The appropriate MeSH term is Kidney Calculi which in each case is part of the list of terms in the MeSH field.

Why should I care? If you use the MeSH term Kidney Calculi, you don't need to search all the synonyms!

TI: Calcium may cut kidney stone risk in younger women
MeSH: Calcium, Dietary/administration & dosage; Kidney Calculi/ prevention & control; Phytic Acid/ administration & dosage

TI: A 44-year-old woman with kidney stones.
MeSH: Insurance, Health; Kidney Calculi/diagnosis, epidemiology, physiopathology, therapy; Kidney Calculi; Obesity; Risk Factors

TI: Urine examination for calculogenic crystals in renal stone patients--a newer approach using refrigeration
MeSH: Calcium Oxalate/urine; Case-Control Studies; Kidney Calculi/urine; Predictive Value of Tests; Urine/chemistry

TI: Crystals, Randall's plaques and renal stones: do bone and atherosclerosis teach us something?
MeSH: Arteriosclerosis/ physiopathology; Bone and Bones/metabolism; Calcium Oxalate/metabolism; Crystallization; Kidney Calculi/etiology

For this example, only selected MeSH terms are displayed from the MeSH field for each article.

Monday, April 04, 2005

Controlled Vocabularies...hmmm

What is a 'controlled vocabulary'?

As we all know, language is imprecise. Even in science and medicine where you'd think language would be extremely precise, concepts are represented by a variety of words and phrases. For example, look at all the different words you can use to describe 'adolescence':

teenagers, teenager, teens, teen, adolescent, adolescents, adolescence, young adult, young adults, youngster, youngsters, 13 year old, freshman, freshmen, 9th grader, etc.

A controlled vocabulary is a standardized collection of terms -- each term represents a concept or a subject. The term is usually defined, examples and other pertinent information are provided all in an effort to reduce ambiguity.

Most controlled vocabularies or thesauri will show the relationships between words through hierarchical relationships (broader terms, narrower terms), see related cross references (other related terms which may be of interest) and see cross references (link to an appropriate term from common synonyms or from an unused more specific term).

In MEDLINE, the Medical Subject Headings (MeSH) is the controlled vocabulary. The indexers who read and analyze the journal articles assign MeSH terms to describe the subject content of the article. Usually an article will have approximately 10-12 controlled vocabulary terms assigned.

Examples of Controlled Vocabularies:
Thesaurus of Psychological Terms[PsycINFO]
Medical Subject Headings (MeSH)[MEDLINE, CINAHL]
Library of Congress Subject Headings [Online catalogs]
Phone Book Yellow Pages

Friday, April 01, 2005

Field qualification in Web of Science

Searching for information in specific fields in the Web of Science is best done by using the Advanced Search screen simply because the field tags are presented clearly in a table right there on the screen.

However, as you become more adept, you can enter the strings in the Topic query box on the General Search screen.

In the Web of Science the syntax is slightly different than both Ovid and PubMed. Here are some examples:

TS=medline AND AU=glover j* [TS is topic; AU is author]
SO=JAMA* AND AD=Yale [SO is source; AD is address]

So, to try this out make sure you click on the Advanced Search button located on the top button bar of the General Search screen. Check out the context specific HELP as well as the HELP Contents and HELP Index.

Extremely beneficial!