5968 COLLEGE ST, HALIFAX, NOVA SCOTIA, B3H 3J5 | 494-2378

Searching for Drugs in PubMed: MeSH, Substance Name, Pharmacological Action and Textwords

This tutorial assumes that our readers are familiar with basic techniques of searching PubMed. More information about these, including PubMed tutorials, are linked from the blue sidebar on the PubMed homepage.  Additional tutorials for using PubMed to search drug- and pharmacy-related topics can be found in this series.

Scenario:

A national meeting of dieticians is being held in your city in a few months. You are a pharmacist specializing in mental health issues and drugs used in psychiatry; they have asked you to make a presentation on the relationship between atypical (sometimes referred to as novel) antipsychotic agents and cholesterol and conditions caused by high levels of cholesterol. As they expect some American dieticians to attend the meeting, they want you to include all the atypical antipsychotics marketed in the US as well as in Canada.

Goal:

Your goal is to find recent information – hopefully evidence-based clinical information on the relationship between atypical antipsychotics and cholesterol, and hypercholesterolemia, hyperlipidemia and dyslipidemia. In addition, you want to make sure that your searches include all the atypical antipsychotic agents marketed in both the US and in Canada.

Strategy:

Identifying the Relevant Drugs:

The following table lists the atypical antipsychotics currently marketed in the US and Canada and their status in PubMed – the database you plan to use to find relevant journal articles. The marketed antipsychotics were identified by using of the CPS and Health Canada’s Drug Product Database (DPD) for Canada and Drug Facts and Comparisons for the US. The status in PubMed was found in the MeSH Database

Canada(1,2) - generic name US(3) - generic name Status in PubMed
not marketed aripiprazole Substance Name
clozapine clozapine MeSH Term
olanzapine olanzapine Substance Name
quetiapine quetiapine Substance Name
risperidone risperidone MeSH Term
not marketed ziprasidone Substance Name

It is fortunate that you investigated which atypical antispychotics are marketed in the US as there are two not marketed in Canada. That part of your research was easy – now comes the part where you need to make sure that your search(es) in PubMed are comprehensive and include all six of the drugs.

PubMed/Medline and Drug Names

PubMed is the name for the entire database published by the US National Library of Medicine (NLM). Medline is the largest portion of PubMed containing records for those journal articles that have received indexing with Medical Subject Headings (MeSH). PubMed also contains “in process” records – the latest records entered into the database that have not yet been assigned MeSH. “As supplied by publisher” records also have not received indexing. To capture both these categories of records we need to do textword searching. Here, we refer to textwords as natural language used by authors in the titles/abstracts of journal articles.

NLM is very progressive – there is staff there who constantly are thinking about MeSH – adding new MeSH and deleting obsolete MeSH. While this “thinking” goes on year ‘round – MeSH are only updated annually – in December. There is a great volume of journal articles published on drugs – whether the topic be medicinal chemistry, pharmacology, therapeutics, pharmacoeconomics or pharmacoepidemiology. Drugs are “discovered” constantly and the persons who develop these drugs publish journal articles. It is no doubt hoped by all researchers that the drugs they develop from scratch will be found to have pharmacological action – pass toxicity testing – prove to be safe and effective and eventually be marketed. All this usually takes many years and all the while journal articles are being published. In most cases the drugs never reach the marketplace and interest in them and in publishing about them tapers off. PubMed doesn’t want to get “stuck” assigning all these drugs MeSH; this would make the section of MeSH (D) where drugs and chemicals are listed overwhelmingly large. If a drug stays around and is marketed it probably eventually will be assigned a MeSH. Meanwhile, what to do with all those drug names? Newly published journal articles cannot be ignored just because there is no MeSH related to the drug being discussed.

Substance Name/Supplementary Concept Records/Pharmacological Action

All drugs in PubMed not having MeSH have been assigned a Substance Name and information about these can be found in Supplementary Concept Records located in the MeSH Browser; the MeSH Browser also contains records for all MeSH – giving history of the use of the term, etc. An indexer faced with a journal article, can go to the searchable MeSH Browser to find out how a particular drug is “handled” – if it has a MeSH it is treated one way – if it has no MeSH it is designated as a Substance Name. New records need to be created the first time a drug is identified in the journal literature. Each drug also is assigned a Pharmacological Action so that similar drugs can be grouped together.

In 2003 the drugs assigned a Substance Name were added to the MeSH Database and also grouped – along with those drugs having MeSH - under their Pharmacological Action classification. This is all “deep thinking” for the average searcher – those interested in learning more are referred to the NLM Technical Bulletin. On the homepage of the Bulletin there is a link to an Index. Select the letter “P” on the alphabet bar to find articles on Pharmacological Action. Reading the NLM Technical Bulletin is an excellent way to keep up-to-date with new features and changes to PubMed.

Beginning the Search – Using MeSH

Let’s start our search in the MeSH Database – accessible from the blue sidebar on the PubMed homepage. Type antipsychotics in the browse window; click the Go button. You will see the preferred term – Antipsychotic Agents. The first occurrence is the name for the group of drugs; although not stated, this is the MESH. This is followed by Antipsychotic Agents (Pharmacological Action). (Please see below.)

Follow the link from the latter term and you will see a list of all the antipsychotic agents covered by PubMed – almost seven dozen drugs. The list is comprehensive for all antipsychotics identified by PubMed – not just the atypicals. Here is the first part of the list as of August 17, 2006.

Following each drug name is the designation with either MeSH Term or Substance Name. All six atypical antipsychotics marketed in either Canada or the US are listed; two have MeSH and four have Substance Names. In both cases, the group terms Antipsychotic Agents are “explosions”; if you choose to search the MeSH you will retrieve only those antipsychotics that are MeSH. If you choose Antipsychotic Agents (Pharmacological Action) you will retrieve ALL antipsychotics. This was the retrieval on August 17, 2006 with the number of citations using Pharmacological Action (PA) being more than three times greater than the number when using MeSH:

Finding MeSH for Cholesterol/Hppercholesterolemia/Hyperlipidemia/Dyslipidemia

Next we look in the MeSH database for the MeSH for cholesterol and the three disease terms, finding that dyslipidemias is an explosion that includes both hyperlipidemia and hypercholesterolemia. We therefore decide to use the MeSH dyslipidemias to capture all three conditions. (Please see below.)

We combined our search statement for cholesterol/dylipidemias and our search statement for antipsychotic agents (Pharmacological Action). We chose the set of Pharmacological Action antipsychotics as the retrieval was more than three times greater than the retrieval using MeSH for that group of drugs; therefore the greater possibility of retrieving more citations for our particular topic. We then limited our retrieval to English, Humans and articles published in the last five years. (Please see below.)

Further Refining Our Retrieval by Adding Textwords Identifying Atypical Antipsychotics

We have retrieved articles about all antipsychotics – not just the atypicals. You could make your retrieval smaller by textwording the words new/newer/atypical*/novel/modern and/or you could textword the name of each of the six drugs and combine this search with the search having the 96 citations. You could then OR the two searches with the textword approaches to get rid of any duplicate citations. This reduced our retrieval from 96 citations to 77 citations. (Please see below for how these two options narrowed our retrieval.) Or you could scan the titles/abstracts of the 96 – it wouldn’t take all that long.

Textwording Drug Names vs. Using MeSH or Substance Name

Our search so far has retrieved citations that have received indexing; we also want to find new “in process” citations because ours is a “hot”/”in the news” topic. The following table will show you the difference in numbers of citations retrieved when you search for our six atypical antipsychotics using either MeSH or Substance Name compared with the retrieval using textwords:

Drug MeSH/Substance Name Textword
Aripipraxole 239 421
Clozapine 4804 6732
Olanzapine 2266 3302
Quetiapine 815 1304
Risperidone 2718 3823
Ziprasidone 342 655

Our textword search is formulated with truncation of the disease concepts taking into consideration variation between American and British spelling – e.g. – dyslipidemia vs. dyslipidaemia. When we search our terms as textwords we are not qualifying particular fields to search – e.g. when you perform your search in the MeSH Database you are qualifying that your terms only are searched among the MeSH. In an unqualified search the terms are searched everywhere in the database; if your terms are synonymous with MeSH those citations will also be retrieved. As we already have looked at those citations, we “subtract” (remove) these from our set of textword citations by using the Boolean operator NOT. Please see search statement #36 below.

It is important to remember to not limit a textword search to Species as species is added when indexing is done and MeSH are assigned. If you limit to Humans in a textword search you automatically are getting rid of “in process” (non-indexed) journal articles and defeating the purpose of textword searching to find the latest references.

Clinical Queries – Finding Evidence-Based Journal Articles

We are going to use our same textword search strategy we used above to search Clincal Queries in both the Clinical Study and Systematic Reviews categories. There is a link to the Clinical Queries feature from the blue sidebar on the PubMed homepage.  Instead of retyping the strategy we copy and paste. Following is how PubMed “interpreted” these searches and our retrieval. For a definition of evidence-based medicine please see the entry for that MeSH in the MeSH Database. Searching PubMed’s Clinical Queries feature is covered in detail in another tutorial in this series.

Conclusion

When performing drug-related searches it is important to be aware of the differences that the use of MeSH, Substance Name, Pharmacological Action and textword searching will have in your retrieval of journal articles. Textword searching is always necessary to capture the very latest citations added to PubMed.

Outcomes

You have identified a number of recent journal articles on atypical antipsychotics and cholesterol/dylipidemias by using MeSH/Substance Names as well as by using textwords. In addition, you have identified some recent evidence-based references by using textwords in Clinical Queries in both the Clinical Study and Systematic Review categories.

(1) Compendium of Pharmaceuticals and Specialties (CPS). Ottawa: Canadian Pharmacists Association; 2006. p. P18.
(2) Health Canada. Drugs & Health Products. Drug Product Database (DPD) - http://www.hc-sc.gc.ca/dhp-mps/prodpharma/databasdon/dpd_index_e.html Accessed August 17, 2006.
(3) Drug Facts and Comparisons. St. Louis MOS: Wolters Kluwer Health; March 2005, p. 936.

This tutorial was prepared in August 2006 by Elizabeth Foy with the assistance of Jeff Nicholson and with advice from Robin Featherstone and David Gardner.