Transitioning from ELHILL to PubMed
Transitioning from ELHILL to PubMed
[Editor's Note added 1998/03/26: see comment in the NLM Technical Bulletin. 1998 Jan-Feb;(300):e1]
A previous article in the Technical Bulletin (July-August 1997 issue) highlighted the basics of PubMed searching. A listing of some of the PubMed search rules and syntax appears on page 9 in this article. This page may be photocopied or removed to keep for ready reference. PubMed's search rules can be easily used to construct expert search strategies similar to those that users are accustomed to creating using the ELHILL command language. PubMed's search rules are easy to master. Two of the biggest syntax hurdles will be to break the habit of using parentheses rather than square brackets to qualify search terms, e.g., [pt] not (pt), and to remember that Boolean connectors must be entered in uppercase, e.g., AND, OR, and NOT.
PubMed offers certain advantages such as the ability to search against all years of MEDLINE and PREMEDLINE in one fell swoop. And, "Overflow" errors that arise on ELHILL from conditions such as terms that generate a high number of postings or date ranging don't routinely happen on PubMed.
To illustrate how easy it is to convert ELHILL commands into PubMed commands, this article will transform a Technical Bulletin Gold Standard search strategy into a PubMed search strategy. The search "The client is a layperson who wants information on alternative treatments for breast cancer" was taken from the January-February 1997 issue. Figure 1 is the ELHILL search strategy originally formulated for CANCERLIT which explains search statements 9-11.
The PubMed Boolean Search page was used to recreate the ELHILL search strategy, in Figure 1, into a PubMed search. The Boolean Search page is best suited for constructing complex (or lengthy) search strategies--the query box is larger and scrollable which allows you to view your search strategy in full. To get there, click on Advanced Search from PubMed's Homepage and then click on the Boolean Search link.
SS 1: exp *breast neoplasms SS 2: exp *alternative medicine SS 3: 1 and 2 SS 4: *holistic health or *wit a#d humor or exp *psychotherapy SS 5: 1 and 4 SS 6: exp *breast neoplasms/dh SS 7: 3 or 5 or 6 SS 8: 7 contains eng (la) and 7 contains human (mh) SS 9: 8 and 96 (yr) or 8 and 95 (yr) or 8 and 94 (yr) SS 10: 8 and 93 (yr) or 8 and 92 (yr) or 8 and 91 (yr) SS 11: 9 or 10
Before you continue reading, at this point if you are still not familiar with PubMed, it is recommended that you review the listing of PubMed's search rules and syntax on Page 9. Otherwise, the translation detailed below from ELHILL language to PubMed may not be as clear.
The query box on the Boolean Search page must contain the entire search strategy as a single search statement. The first step in converting the ELHILL search strategy into a single search statement for PubMed was to create the three search statements that would be OR'd together (i.e., 3 or 5 or 6). Each of those search statements converted into a PubMed search statement follows:
SS 3: breast neoplasms [majr] AND alternative medicine [majr]
SS 5: holistic health [majr] OR wit and humor [majr] OR psychotherapy [majr] AND breast neoplasms [majr]
SS 6: breast neoplasms/diet therapy [majr]
Note: PubMed automatically explodes MeSH terms, whereas on ELHILL you must specify whether or not a term is to be exploded (e.g., breast neoplasms [majr] is equivalent to exp *breast neoplasms). The ability to turn off an automatic explode will be available soon. So, keep in mind that the same syntax is used for terms that are explodable as well as for terms that do not have any indentions. For example, the MeSH terms, "Holistic Health" and "Wit and Humor" (neither have indentions) are entered as holistic health [majr] and wit and humor [majr] and the explodable MeSH term "Psychotherapy" is entered as psychotherapy [majr].
Boolean search statements are processed left to right. To change the order in which PubMed processes a search statement, enclose the individual concept in parentheses. The terms inside a set of parentheses are processed as a unit and then incorporated into the overall strategy. Below are two examples of how the above PubMed search statements can be OR'd together with or without parentheses:
PubMed can handle large explosions for over 9 million citations in a blink of an eye. While you should apply logic in formulating a search, all those lessons about segmenting your terms into separate search statements for computer and cost-efficiency are just not vital to the new PubMed technology. That said, however, PubMed's future plans include a feature where you will be able to view a picture of the strategy--similar to Internet Grateful Med's "Details of Search" button. Here we hope to provide clarifying messages such as terms that have no postings. In addition, the plans also call for this feature to include an edit function so that a search strategy can be easily revised and resubmitted; and intermediate postings shown.
The remaining portion of the search strategy is to apply the search limiters for language (i.e., english), study group (i.e., human), and a date range (i.e., articles published during 1991-1996). Whereas, on ELHILL to apply the desired limits in CANCERLIT took an additional four steps (i.e., ELHILL search statements 8-11), this can be easily translated on PubMed into the syntax below and then added (ANDed) to the search strategy.
english [la] AND human [mh] AND 1991:1996 [dp]
Figure 2 contains the ELHILL Gold Standard search converted into a single PubMed search statement.
Figure 2 - ELHILL search converted to PubMed
alternative medicine [majr] OR holistic health [majr] OR wit and humor [majr] OR psychotherapy [majr] AND breast neoplasms [majr] OR breast neoplasms/diet therapy [majr] AND english [la] AND human [mh] AND 1991:1996 [dp]
To ensure that the ELHILL search strategy was correctly converted to PubMed, the search strategies were run in their respective systems to compare search results. The ELHILL search was run in MEDLINE, MED93, and MED90. In both ELHILL and PubMed, the results were the same -- 37 was the total postings.
If you have any questions or comments about the PubMed search strategy, please contact PubMed customer support by clicking on the "Help Desk" link.
While constructing a complex (or lengthy) search strategy for PubMed, it is helpful to use your word processing software as a scratch pad. Use the word processing software to enter your search terms, the correct syntax (e.g., search field tags enclosed in square brackets, Boolean connectors in all uppercase), and spell check. It helps to use extra spaces or blank lines to separate your concepts for ease of proofreading and checking your logic. Once you are satisfied with the search strategy, "copy and paste" it into the PubMed query box. Be sure to remove all extra spaces, hard carriage returns or other special word processing codes. You can then go back to the search strategy in your word processing software to edit or reformulate the search strategy--quickly and easily. Repeating the "copy and paste" step into the PubMed query box will then let you execute the revised strategy with just a few clicks.
Test Your Skills
Now that you have stepped through the process of translating an ELHILL search strategy into PubMed, why not test your skills? Figure 3 is a rather complex and lengthy ELHILL search strategy for you to translate into PubMed. This example is based on a portion of an ELHILL search used to create the MEDLINE subfile of SPACELINE.
Before you get started, please refer to the Helpful Hint box on page 6 that discusses using your word processing software in conjunction with PubMed for complex and lengthy search strategies.
When the ELHILL search was run against MEDLINE back through MED66 there were 20,997 postings. PubMed yielded 21,107 postings--a difference of 110 postings on the test day. The tested ELHILL search differed from the strategy in Figure 3 in only one way--the first two MeSH terms were exploded in order to compare the total postings to PubMed because the automatic explosion of MeSH in PubMed cannot be turned off yet. Even with that adjustment PubMed had higher postings, though. Why the difference you ask? Remember, PubMed is a single database that not only includes all of MEDLINE (1966 to present) but PREMEDLINE as well. In addition, there are citations, labeled as [Record as supplied by publisher], which are electronically-supplied to NLM by the publisher. These citations are first received by PubMed before being processed into PREMEDLINE and eventually MEDLINE. In addition, for those electronically supplied citations, from journals that are selectively indexed (e.g. Science or Nature), the non-indexed citations remain in PubMed even though they never become MEDLINE citations. The additional 110 citations found by PubMed for this search were confirmed to be either PREMEDLINE or publisher supplied citations.
In both the Gold Standard search and the Test Your Skills search you can see how a complex and lengthy ELHILL strategy can be transitioned to PubMed with equivalent results. Learning any new system is a challenge. As we work together with your direct feedback on your needs as information providers, PubMed will keep evolving to make it easier and clearer as evidenced by the future plans for the "Details of Search button."
Figure 3 -Test Your Skills. Convert this ELHILL Search Strategy into PubMed
SS 1: space flight or extraterrestrial environment or aerospace medicine or submarine medicine or cosmic radiation SS 2: (ad) ames and research and center SS 3: 2 and not iowa (ad) and not ia (ad) and not all agricult: (ad) SS 4: (ad) johnson and space and center or kennedy and space and center or goddard and space and center or marshall and space and center or stennis and space and center SS 5: (ad) national and aeronautics and space or nasa SS 6: 5 and not spain (ad) and not france (ad) SS 7: (tw) space and agency SS 8: (tw) canadian or european or all japan: or french or german SS 9: 7 and 8 SS 10: (tw) centre and national and etudes and spatiales SS 11: (tw) deutsche and agentur and raumfahrtangelegenheiten SS 12: blomqvist cg or booth f or booth fw or buckey jc or cintron nm or convertino v or convertino va or czeisler ca or daunton ng or dillaman rm or dudley ga or fitts rh or fortney s or fortney sm or fox ge or ganong wf or genant hk or hargens ar or hockstein li or horrigan dj jr or ingber de or jee ws or jukes th SS 13: kanavarioti a or keil lc or koch kl or kretsinger rh or lacey jc jr or lackner rd or lambertsen cj or lanyi jk or leach cs or loeppky ja or malacinski gm or margulis l or mcfeters ga or meehan rt or moore-ede mc or morey er or morey-holton e or morey-holton er or musacchia xj SS 14: orgel le or pak cy or partridge nc or perachio aa or pierson dl or ponnamperuma c or reschke mf or ross md or rumbaugh dm or schneider vs or schor rh or tipton cm or tischler me or vandenburgh hh or vailas ac or waligora jm or west jb or whalen rt or wilt f or wilt fh or woese cr or wolgemuth dj SS 15: (jc) kxc or ppj or bp5 or 2rq or 9ja or bx7 or ok7 or ok8 or ux6 SS 16: 1 or 3 or 4 or 6 or 9 or 10 or 11 or 12 or 13 or 14 or 15
Figure 4 - Test Your Skills ELHILL Search Converted to PubMed as it Might Appear in a Word Processing Document
space flight [mh] OR extraterrestrial environment [mh] OR aerospace medicine [mh] OR submarine medicine [mh] OR cosmic radiation [mh] OR
(ames [ad] AND research [ad] AND center [ad] NOT (iowa [ad] OR ia [ad] OR agricult* [ad])) OR
((johnson [ad] OR kennedy [ad] OR goddard [ad] OR marshall [ad] OR stennis [ad]) AND space [ad] AND center [ad]) OR
((national [ad] AND aeronautics [ad] AND space [ad] OR nasa [ad]) NOT (spain [ad] OR france [ad])) OR
(space [tw] AND agency [tw] AND (canadian [tw] OR european [tw] OR japan* [tw] OR french [tw] OR german [tw])) OR
(centre [tw] AND national [tw] AND etudes [tw] AND spatiales [tw]) OR
(deutsche [tw] AND agentur [tw] AND raumfahrtangelegenheiten [tw]) OR
blomqvist cg OR booth f OR booth fw OR buckey jc OR cintron nm OR convertino v OR convertino va OR czeisler ca OR daunton ng OR dillaman rm OR dudley ga OR fitts rh OR fortney s OR fortney sm OR fox ge OR ganong wf OR genant hk OR hargens ar OR hockstein li OR horrigan dj jr OR ingber de OR jee ws OR jukes th OR
kanavarioti a OR keil lc OR koch kl OR kretsinger rh OR lacey jc jr OR lackner rd OR lambertsen cj OR lanyi jk OR leach cs OR loeppky ja OR malacinski gm OR margulis l OR mcfeters ga OR meehan rt OR moore-ede mc OR morey er OR morey-holton e OR morey-holton er OR musacchia xj OR
orgel le OR pak cy OR partridge nc OR perachio aa OR pierson dl OR ponnamperuma c OR reschke mf OR ross md OR rumbaugh dm OR schneider vs OR schor rh OR tipton cm OR tischler me OR vandenburgh hh OR vailas ac OR waligora jm OR west jb OR whalen rt OR wilt f OR wilt fh OR woese cr OR wolgemuth dj OR
kosm biol aviakosm med [ta] OR probl kosm biol [ta] OR adv space biol med [ta] OR aviakosm ekolog med [ta] OR aerosp med [ta] OR aviat space environ med [ta] OR orig life [ta] OR orig life evol biosph [ta] OR space life sci [ta]
Figure 5 - Test Your Skills ELHILL Search Converted to Pubmed as it Should Appear in the PubMed Query Box
space flight [mh] OR extraterrestrial environment [mh] OR aerospace medicine [mh] OR submarine medicine [mh] OR cosmic radiation [mh] OR (ames [ad] AND research [ad] AND center [ad] NOT (iowa [ad] OR ia [ad] OR agricult* [ad])) OR ((johnson [ad] OR kennedy [ad] OR goddard [ad] OR marshall [ad] OR stennis [ad]) AND space [ad] AND center [ad]) OR ((national [ad] AND aeronautics [ad] AND space [ad] OR nasa [ad]) NOT (spain [ad] OR france [ad])) OR (space [tw] AND agency [tw] AND (canadian [tw] OR european [tw] OR japan* [tw] OR french [tw] OR german [tw])) OR (centre [tw] AND national [tw] AND etudes [tw] AND spatiales [tw]) OR (deutsche [tw] AND agentur [tw] AND raumfahrtangelegenheiten [tw]) OR blomqvist cg OR booth f OR booth fw OR buckey jc OR cintron nm OR convertino v OR convertino va OR czeisler ca OR daunton ng OR dillaman rm OR dudley ga OR fitts rh OR fortney s OR fortney sm OR fox ge OR ganong wf OR genant hk OR hargens ar OR hockstein li OR horrigan dj jr OR ingber de OR jee ws OR jukes th OR kanavarioti a OR keil lc OR koch kl OR kretsinger rh OR lacey jc jr OR lackner rd OR lambertsen cj OR lanyi jk OR leach cs OR loeppky ja OR malacinski gm OR margulis l OR mcfeters ga OR meehan rt OR moore-ede mc OR morey er OR morey-holton e OR morey-holton er OR musacchia xj OR orgel le OR pak cy OR partridge nc OR perachio aa OR pierson dl OR ponnamperuma c OR reschke mf OR ross md OR rumbaugh dm OR schneider vs OR schor rh OR tipton cm OR tischler me OR vandenburgh hh OR vailas ac OR waligora jm OR west jb OR whalen rt OR wilt f OR wilt fh OR woese cr OR wolgemuth dj OR kosm biol aviakosm med [ta] OR probl kosm biol [ta] OR adv space biol med [ta] OR aviakosm ekolog med [ta] OR aerosp med [ta] OR aviat space environ med [ta] OR orig life [ta] OR orig life evol biosph [ta] OR space life sci [ta]
Author Searching: Currently, PubMed searches an author name having an initial or initials the same way ELHILL does. With the next system update, the search rules for author searching will change to incorporate automatic truncation that can be turned off. This will affect how this search would be translated to PubMed in the future.
Be advised that unqualified author names can retrieve from other fields such as the abstract where references may be cited.
Journal Titles: In Figure 3, the 3-character journal code (e.g., 2rq (jc)) was used to retrieve citations to specific journals on ELHILL. In PubMed, the 3-character journal code is not searchable. Instead, you can use the MEDLINE title abbreviation (e.g., aerosp med [ta]) or the full title (e.g., aerospace medicine [ta]) to retrieve citations from specific journals.
Date and Date Range Format:
Search Field Qualification:
Terms can be qualified using PubMed's search field tags. Note: A list of search field tags is available in PubMed's online help under Search Fields.
MeSH Terms, Subheadings, and Language