Publications > Searching the Medical Literature Using CD-ROMs:

Basic Information and Useful Tips With Special Emphasis on MEDLINE

Aida A. Farha I, Hilda T. Nassar II

The rapid advancement in information technology obliges all interested seekers, specially in the medical field, to update themselves with this progress. Trends in information technology include the evolution of online searching from a remote place, CD-ROM (Compact Disk Read Only Memory) searching, Internet, etc. Several bibliographic indexes that were published only in printed versions are now also produced on CD-ROM, and the number is rapidly increasing. This paper presents basic information and useful tips on MEDLINE CD-ROM searching, as it is the most widely used computerized bibliographic database for searching the medical literature; many physicians depend on it to keep themselves abreast of the latest developments, to check information about rare cases and to use it as a "continuing medical education" tool.

CD-ROM bibliographic searching is a relatively new technology for many Lebanese searchers, due mainly to the strife period that delayed its application. However, once a searcher is introduced to this exciting technology, he/she is preferring it to manually searching the printed indexes. An exception exists, the older the searcher &/or the more years spent searching manual indexes, results in preference of the "traditional" searching over the computerized one.

To execute a computer search, a user has to first decide which database is best suited for a specific topic. This is achieved after having some knowledge of the various databases available, the subject scope of each, the years of coverage etc... Here is a brief description of the most important CD-ROM bibliographic databases in the medical field:

MEDLINE: the computerized version of Index Medicus published by the National Library of Medicine (NLM in Bethesda, MD, USA), plus citations taken from the International Nursing Index and the Index to Dental Literature. It is the major bibliographic database for the biomedical literature, containing references to articles taken from more than 3500 serials published world wide (the Lebanese Medical Journal included). The program has been available in a computerized version since 1966 and is updated monthly on CD-ROMs.

EMBASE: the electronic version of Excerpta Medica produced by Excerpta Medica of Amsterdam, covers all aspects of human medicine including chemistry, sociology, management and economics. It concentrates in particular on European sources and contains a large section (about 40%) on drug-related literature.

BIOSIS: the electronic version of Biological Abstracts and Biological Abstracts/ Reports Reviews Meetings (BA and BA/RRM) is published by the Bioscience Information Service in Philadelphia, USA. It abstracts more than 8000 international serial titles covering the biological and medical literature from over 100 countries. In addition to journal articles, books, proceedings of meetings, conferences, reports and notes are included in BIOSIS database. Over 80% of the data is medically oriented.

ClinPSYC: a clinical psychology / health psychology subset of PsycINFO database produced by the American Psychological Association.

CINAHL: the computerized version of the Cumulative Index to Nursing and Allied Health Literature, indexes more than 500 journal, plus books and dissertations in nursing and allied health disciplines.

POPLINE: a population database giving information on population, family planning and related health issues. The material is indexed from articles, books, reports and published or unpublished documents. Any document in this database, is supplied free of charge if request comes from a developing country.

HealthSTAR: a bibliographic file giving informative abstracts covering the non-clinical aspects of health care delivery, including all aspects of administration and planning of health care facilities, financial management etc... The data is supplied by the U.S. National Library of Medicine and the American Hospital Association.

Most of the above CD-ROM bibliographic databases are marketed by a number of commercial vendors, providing the user with a wide choice of collection. Basically the information is the same, but the way it is searched or displayed may vary from one vendor to another.

Before dealing with MEDLINE CD-ROM searching, it is worth comparing it to its printed version, Index Medicus. Table I gives a brief comparison between MEDLINE and Index Medicus.

Table I: Comparison between MEDLINE and Index Medicus
Coverage Index Medicus & International
Nursing Index & Index to Dental Literature
Index Medicus
Goes back to 1966 1879
1 - 2 months up-to-date coverage 1 - 2 months behind
Access points Author, subject, journal, publication year, title, abstract, address of author, etc.... Only Author or Subject
Textword search available not available
Indexing depth Profound : major and minor MeSH Less profound : only major MeSH
Searching speed Quick Slow specially for complicated searches
Abstract available Yes (for about 70% of articles) No
Author's address available not available
Controlled vocabulary MeSH MeSH

Once the database to be searched is selected, the next step is to design the search strategy, i.e. to translate a search query (after breaking it into as many concepts as is necessary) into a language acceptable to a particular computer software. To perform a good search strategy, the searcher should have some knowledge about the database indexing plus some logical and analytical skills. This step is affected to a great extent by the personal approach, for two searchers may design different search strategies for the same topic and hence retrieve different records.

One main advantage to searching MEDLINE CD-ROM is the possibility to search two or more concepts together using Boolean operators ("AND", "OR", "NOT").

retrieves all records having concept A and concept B in the same record.

retrieves records containing either concept A or concept B


retrieves records containing concept A but not concept B.

This option should be used with caution as it might eliminate some relevant records. When more than two concepts are searched together, the "AND" is processed before the "OR". To override this logic, nesting or grouping using parenthesis should be utilized.

ex. A AND B OR C is equivalent to (A AND B) OR C the computer will first search A AND B, then will OR the result with C.


(A AND B) OR C                             A AND (B OR C)

Whereas A AND (B OR C) the computer searches for B OR C, and the result is "AND"ed with A. These two examples retrieve different records.

In addition to Boolean operators, MEDLINE allows the use of proximity operators that allow the searcher to specify that two concepts appear within the same field (like title or abstract) or more precisely appear in the same sentence (in contrast to Boolean "AND" where the two concepts may appear in different fields in the same record). Hence proximity operators allow more specific searching and retrieve a fewer number of hits than Boolean operators. (See Table II).

Another useful option is truncation, which allows the searcher to search for all occurrences of words that start with a specific root. Truncating the word "mutagen" retrieves all records containing words like mutagen, mutagens, mutagenic, mutagenicity etc...(See Table II).

Table II: Basic commands for searching MEDLINE on Silver Platter:
Search Example
Retrieves Records
AND influenza AND cough containing both influenza and
cough in the same record
OR liver OR hepatic containing either liver or hepatic
NOT twins NOT identical containing twins but not identical
in the same record
With heart WITH surgery containing heart and surgery in
the same field
NEAR heart NEAR surgery containing heart and surgery in
the same sentence
IN cancer in ti with cancer appearing in title
= La = English published in English
  PY = 1995 published in 1995
  TG = human about human beings
  PT = review that are reviews of the literature
> PY > 1995 published after 1995

To efficiently search MEDLINE, a user should acquire some knowledge on its indexing principles. All records are indexed using controlled vocabulary or thesaurus terms equivalent to the Medical Subject Headings (MeSH), familiar to users of Index Medicus. Records are indexed by selecting the most appropriate or specific MeSH terms that describe the subject content of a record; these are the same MeSH terms that should be used to retrieve those records. Furthermore, subheadings (see Table III) are assigned to almost all MeSH terms, so that a user may design a specific search strategy (coupling a MeSH term to a particular subheading(s)), to retrieve a high percentage of relevant records.

Table III: List of the 82 subheadings available for MEDLINE users:
abnormalities embryology poisoning
administration & dosage enzymology prevention & control
adverse effects epidemiology psychology
agonists ethnology radiation effects
analogs & derivatives etiology radiography
analysis genetics radionuclide imaging
anatomy & histology growth & development radiotherapy
antagonists & inhibitors history rehabilitation
biosynthesis immunology secondary
blood injuries secretion
blood supply innervation standards
cerebrospinal fluids instrumentation statistics & numerical data
chemical synthesis isolation & purification supply & distribution
chemically induced legislation & jurisprudence surgery
chemistry manpower therapeutic use
classification metabolism therapy
complications methods toxicity
congenital microbiology transmission
contraindications mortality transplantation
cytology nursing trends
deficiency organization/administration ultrasonography
diagnosis parasitology ultrastructure
diagnostic use pathogenicity urine
diet therapy pathology utilization
drug effects pharmacokinetics veterinary
drug therapy pharmacology virology
economics physiology  
education physiopathology  

Another useful feature is the fact that MeSH terms are grouped in a hierarchical arrangement. As a result a broad MeSH term will have all its narrower MeSH terms indented under it in a tree-structure form. Hence a searcher on a computer has the option to either search a broad MeSH term alone (to retrieve the general aspect of that topic only) or explode that broad MeSH term to also include all of its indented narrower MeSH terms, grouped together in an "OR" Boolean operation (see Figure I).

Figure I: Tree structure for mouth neoplasms taken from MeSH 1997:[6]

    Mouth Neoplasms
      Gingival Neoplasms
      Leukoplakia, Oral
        Leukoplakia, Hairy
      Lip Neoplasms
      Palatal Neoplasms
      Salivary Gland Neoplasms
        Parotid Neoplasms
        Sublingual Gland Neoplasms
        Submandibular Gland Neoplasms
      Tongue Neoplasms

Exploding the MeSH term "mouth neoplasms", retrieves all articles discussing any of the MeSH terms included in Figure I combined in an "OR" relationship. Whereas exploding the MeSH term "salivary gland neoplasms" retrieves all articles on any of its indented MeSH terms i.e. "parotid neoplasms", or "sublingual gland neoplasms", or "submandibular gland neoplasms" in addition to articles under the MeSH term "salivary gland neoplasms".

To search MEDLINE CD-ROM for a specific subject, one may search:

  • Thesaurus or MeSH terms.
  • Textword or free-text search in which the computer retrieves all occurrences of the word in any searchable field like title, abstract...
  • A combination of the above two methods.

Thesaurus searching is usually choice number one, as it retrieves records discussing that particular subject. Textword searching does not necessarily guarantee this and as a result many irrelevant records are retrieved while many relevant ones are missed. To search for "Acquired Immunodeficiency Syndrome", if the user searches for AIDS as a textword, irrelevant records such as "Technical aids to patients with spinal cord injuries" etc, are retrieved. Also records about AIDS syndrome that do not mention AIDS in its abbreviated form, are missed. Moreover records having the abbreviation AIDS appearing for example in the abstract would be retrieved even though the article is not necessarily discussing AIDS, but merely mentioning it in the abstract. So the number of records retrieved from textword searching might overwhelm the relevant ones or may produce a large number of records that the user might not have the time or patience to browse through.

Textword searching becomes a good choice when there is no specific MeSH term for the subject of concern. This might happen if the subject is new or is very rarely documented in the literature. To do such a search, a broader MeSH term may be "AND"ed with that specific term searched as a textword. When using textwords, "OR" all possible synonyms, variant spellings, singular, plural, adjectival, British, American, and abbreviated forms where applicable.

Furthermore, searching the MeSH terms, a user may specify a subheading(s) for that term, to retrieve relevant records on a specific topic. The user should understand the subject scope of each subheading to correctly choose the appropriate one(s). The most frequent misconception about subheadings arises when searching for the treatment of a disease, the user wrongly searches only the subheading "therapy" thinking that this would retrieve all aspects of therapy. To have a comprehensive search about treatment of a disease, the user has to search "diet therapy", "drug therapy", "prevention & control" "radiotherapy", "rehabilitation", "surgery" in addition to "therapy". Similarly when searching for "analysis" the user should also consider the possible inclusion of the subheadings "blood", "cerebrospinal fluid", and "urine". Also when searching for "adverse effects" one should consider also "poisoning" and "toxicity" [3].

MEDLINE CD-ROM searching, being an interactive process between the searcher and the computer, allows the searcher to easily modify the search strategy as the search goes on, depending on the number of hits retrieved and on their relevance. This modification may be done in the following two ways:

  • Broadening the search when the number of retrieved records is zero or very small:
  • use "OR" Boolean operator to combine related or synonymous terms.
  • search broader MeSH terms.
  • search all subheadings instead of specifying part of them.
  • use MeSH term "OR"ed with its equivalent textword.
  • use truncation when searching textwords.
  • omit one "AND"ed concept (the least important one)
  • increase the number of years searched.
  • search another possible CD-ROM database.
  • Narrowing the search when the number of retrieved records is very large:
  • use "AND" to combine with another concept.
  • use proximity operators (where applicable) instead of Boolean operators.
  • omit textword synonyms.
  • omit broader MeSH terms.
  • omit truncation.
  • search for specific subheadings with each MeSH term.
  • limit years, languages, reviews, etc...

In addition to subject (MeSH) and textword searching, other access points are available for searching MEDLINE CD-ROM, such as author, journal name, publication year, etc... Searching through these is usually straight forward, and the results may be combined with any subject search using Boolean operators.

Three facts are to be mentioned here, to clear common misconceptions that usually arise while searching MEDLINE. The first fact is that there is a time lag between the date an article is published in a journal and the date it actually appears in MEDLINE. This time lag varies and may reach one year for some journals. The second one is that some journals are selectively indexed, and not all articles published in these journals appear in MEDLINE. The third fact is that a user should be aware that MEDLINE does not cover all the medical literature in the world [4]; maybe it covers a large portion of the most important ones, but sure does not cover all. As a remedy, a user is advised to use other databases like EMBASE or BIOSIS to cover a wider collection of the world's medical literature.

Efficient MEDLINE searching is of vital importance to improve patient-care [5]. The efficiency of such a search depends on many factors such as the expertise of the searcher, his ability to translate any topic into a strategy that best serves his needs, and his knowledge about MeSH indexing [1]. In most cases, once these basic points are secured, a searcher will search MEDLINE more efficiently, specially if a combination of both MeSH terms and textwords is utilized [2].

About the authors:

I. Aida A. FARHA: Reference Librarian, Saab Medical Library, American University of Beirut (SML-AUB). Back

II. Hilda T. Nassar: Medical Librarian, SML-AUB. Back

Correspondence to : Aida Farha, Saab Medical Library.
American University of Beirut. P.O.Box 11-0236/36. Beirut-Lebanon.


  1. Craig TB. MEDLINE [letter]. J Fam Pract Aug 1991; 33(2):128.
  2. Farbey R. Searching the literature. Be creative with Medline [letter]. BMJ Jul 1993; 307(6895):66.
  3. Feinglos SJ. MEDLINE : a basic guide to searching. Chicago, Ill: Medical Library Association, 1985:26.
  4. Fink M. Using MEDLINE to solve clinical problems [letter]. JAMA Nov 1993; 270(17):2053-4.
  5. Lindberg DA et al. Use of MEDLINE by physicians for clinical problem solving. JAMA Jun 1993; 269(24):3124-9.
  6. National Library of Medicine. Medical Subject Headings. Bethesda, MD: National Library of Medicine, 1997: 897.

This article was published in: J Med Liban 46(1):48-51, 1998.