by Cindy Schmidt, M.D., M.L.S. and Roxanne Cox, M.L.S.
Updated 8/6/2015
Navigating this tutorial:
Accessing PubMed:
Adjust the size of your screen so that you can see the right hand side of the McGoogan Library homepage.
If you are working on this tutorial from an off-campus location, you may be asked to log in with your UNMC Net ID (e-mail username and password).
Clinical Situation:
You are a student completing a rotation in a clinical laboratory. Your supervisor is considering adding one of the relatively rapid, C. difficile tests to the lab's offerings. She wants a test with a turn-around time that is less than 24 hours. She asks that you determine which of the relatively rapid C. difficile assays has the best balance of sensitivity and specificity when compared to the "gold standard" reference testing method.
Do you know what the reference testing method is for C. difficile? Do you know what types of new, rapid tests are available? If not, you will need to read a bit before beginning to compose your search. Right click on this link and select the "new window" option . A new window containing the CDC's C. difficile FAQs will appear. Scroll down to find the "Which laboratory tests..." section and read the information about the available tests. When you've finished your reading return to this browser window and click on the arrow to go to the next screen.
What is the gold-standard or reference method for C. difficile testing?
Which of the methods below would you include among the more rapid tests that your supervisor might consider?
After reading the brief summary from the CDC, you should have enough information to begin creating your search strategy. The first step in any search is identifying the clinical situation/question. Your supervisor has done this for you. The second step is identifying the important search concepts. Sometimes these concepts are very obvious. At other times the PICO acronym can be helpful. This search benefits from the application of PICO. The "P", "I", "C", and "O" stand for:
P-Patient characteristic/disease
I-Intervention
C-Comparison intervention
O-Outcome of interest
What are the P, I, C, and O concepts in your search?
In case you need a reminder, here's the situation of interest:
Your supervisor is considering adding one of the relatively rapid, C. difficile tests to the lab's offerings. She wants a test with a turn around time that is less than 24 hours. She asks that you determine which of the relatively rapid C. difficile assays has the best balance of sensitivity and specificity when compared to the "gold standard" reference testing method.
Did you come up with the following PICO search concepts?
P -- C. difficile infection
I -- immunologic or genetic/PCR detection of toxin
C -- toxigenic culture
O -- sensitivity and specificity
It's not important that the wording be identical to the above. Also, the I and C concepts could be switched.
In the future, you may work on searches that do not require all aspects of PICO. In some searches there is no essential outcome concept, for example. Never include a concept that is not essential to your search.
You have now
1. Identified an answerable clinical question
2. Identified the important search concepts.
The tutorial will guide you as you go on to complete the remaining steps in the searching process.
3. Identify subject headings or keywords for each concept.
4. Join alternate subject headings/keywords for a single concept with OR
(OR tells the search engine to retrieve items with one or the other of the adjacent terms)
5. Enclose OR'd terms with parentheses
(Parentheses tell the search engine to consider the items retrieved by the enclosed terms as a set.)
6. Join terms for disparate concepts with AND
(AND tells the search engine to only retrieve items that contain both of the adjacent terms or only those that are members of both of the adjacent sets.)
7. Use the strategy that has been created to search PubMed
8. Add filters/limits. Filter/limit concepts include:
Publication characteristics-publication language (e.g. English), article type (e.g. review vs. randomized controlled trial)
Experimental subject characteristics - species (e.g. human vs. animal), gender, age
Do not add limits unless they are central to your goals for the search.
9. Attempt to gain access to full-text articles
Most literature databases can be searched with either keywords or subject headings.
A keyword search is the type of search you do in Google or another internet search engine. You enter words that you think the author might have used to indicate that he/she is talking about the concepts of interest.
If you are trying to do a thorough search with keywords, you have to try to list all the words an author might use to indicate that he/she is writing about a concept. The alternate terms for a single concept are OR'd together.
Why use OR?
Because you don't care which of the alternate terms is present.
"immunologic assay" OR "immunologic assays" OR "immunologic test" OR "immunologic tests" OR immunoassay* OR immunosorbent OR EIA OR EIAs OR ELISA OR ELISAs OR RIA OR RIAs OR radioimmunoassay* OR PCR OR RT-PCR OR RTPCR OR "polymerase chain reaction"
Notice that:
You might know of other types of immunologic or genetic assays or the brand names of specific tests. These other terms/names could be OR'd into the above list.
As you can see, doing a thorough keyword search is not a simple process. Using subject headings can make searching easier.
However, to use subject headings correctly, you need a little background.
Select "MeSH" from the drop-down menu next to the PubMed search box (a in figure below). Type --
ELISA
-- in the search box (b in figure) and hit the "Search" button (c in figure).
Did your search find the entry shown in the screenshot below? This is the single subject heading used for articles focused on the ELISA concept.
What's so great about searching with subject headings?
1. When you search with a broad subject heading like "immunologic techniques," you retrieve all records indexed with that heading and you also retrieve all the articles indexed with narrower headings (like "immunoassay", "Enzyme-linked Immunosorbent Assay", "Enzyme-linked Immunospot assay", etc). This is called heading explosion. It happens by default in PubMed. Click on the "Immunologic Techniques" heading and scroll down to the "Immunologic Techniques trees. Take a look at all the types of assays you can search for with this one heading!
2. If we use the "immunologic techniques"[mesh] to search, the articles we retrieve will focus some attention on this topic. If we search with an "immunologic techniques" keyphrase/keyword, we may retrieve a record that states that "all the newer diagnostic techniques except immunologic techniques will be discussed." The article includes our keyword/keyphrase, but focuses no attention on that concept.
3. If we wish, we can combine our heading with a subheading, like "methods" or "instrumentation" to indicate that only a specific aspect of the topic is of interest to us.
There is one big disadvantage of using MeSH in search. A MeSH search cannot retrieve records that are not indexed yet.
Most of the records in PubMed are part of the fully-indexed MEDLINE record set. These records carry the designation "indexed for MEDLINE." They have MeSH and can be retrieved by a MeSH search.
However, PubMed also contains three other record sets.
1. PREMEDLINE -- PREMEDLINE is composed of records that have been sent to the National Library of Medicine by journal publishers and are still awaiting the attention of an indexer. These records are designated as "as supplied by publisher" or "in process" in PubMed. Records for articles from the less important journals may sit in PREMEDLINE awaiting indexing for over a year.
2. OLD MEDLINE -- OLD MEDLINE records have been created by keying entries from old print indexes into a computer. The old subject heading terminology in these records is different from current-day MeSH.
3. Other PubMed -- U.S. law states that the full-text articles based on research that has been funded by the NIH must be made freely available to the public within a year of publication. The NIH-funded articles are made available through a full-text depository called PubMed Central. All PubMed Central articles have records in PubMed but not all have been published in MEDLINE-indexed journals. So some of these records are never indexed. Other items that will never be indexed include records for the book chapters available through the NCBI bookshelf, and records for Oregon Drug Effectiveness Project reports.
Items from these three, non-MEDLINE PubMed record sets can only be retrieved using keyword searches and not by MeSH searches.
Why use MeSH if some records can only be retrieved by keyword searches? Use of MeSH makes Medical Laboratory Science searches so much easier that it's usually worth starting with a MeSH search.
Now it's time to try to find MeSH terms for each of your search concepts:
1. Clostridium difficile infection
2. immunologic or genetic assays
3. toxigenic culture
4. sensitivity and specificity
Since you're already looking at the "immunologic techniques" section of MeSH, start by using this heading for the "immunologic assays" portion of the immunologic or genetic assays concept.
Click the "Add to PubMed search builder" button on the right-hand side of the page (see arrow in figure below).
PubMed will enclose the MeSH in quotes and add the "MeSH" tag (see blue box in figure below).
Perhaps you're wondering why you should use "immunologic techniques"[mesh]? Why not use "immunoassays"[mesh]?
Look at the MeSH trees at the bottom of the "immunologic techniques" page again. Do you see any important narrower terms under "immunologic techniques" that are not also narrower terms under "immunoassays?"
Did you say that the "immunoprecipitation" heading might be important?
If so, you are right. "Immunoprecipitation" is a narrower heading under "Immunologic Techniques" but (for reasons unknown to this tutorial's author) it is not also in the "Immunoassays" tree.
If you search with the "Immunoassays" heading you will not retrieve any articles indexed with an "immunoprecipitation" heading or a "Radioimmunoprecipitation assay" heading.
If, on the other hand, you search with the "immunologic techniques" heading, you will retrieve articles indexed with:
Of course, if you know that there are no important immunoprecipitation or radioimmunoprecipitation assays for C. Difficile antigens, you could use "Immunoassays" instead.
You could now search for a genetic techniques heading, but since we don't know the exact wording for the heading, it may be easier to find it by looking around in the "investigative Techniques" MeSH tree. Click on "Investigative Techniques"
You should now be looking at the entry for the "investigative techniques" (see figure below).
Scroll down to see the list of narrower headings in the "Investigative Techniques" trees. The first heading you'll see for the "genetic techniques concept is "Genetic testing." Click on this heading.
When you reach the "Genetic Testing" entry, scroll down to see the trees for "Genetic testing." Is there a broader heading that might include important tests that are not included under "Genetic testing?"
Did you say that the "Genetic Techniques" heading might include other important genetic assays? If so, you are correct. Click on the "Genetic Techniques" link (see blue arrow in figure below).
When you reach the entry for "Genetic Techniques," think about how you want to join the "Immunologic Techniques"[MeSH] that is already in your search builder with the new "Genetic Techniques"[MeSH]. Do you want to use AND, OR, or NOT? Chose the correct option from the drop-down menu under the "PubMed Search Builder" (see arrow "a" in the figure below). Then click on "Add to Search Builder" (see arrow "b" in the figure).
Did you choose to join the headings with OR? If so, you are correct. You want articles that either contain information about immunologic techniques or contain information about genetic techniques. It isn't essential that both types of assay be discussed.
If you chose either AND or NOT, highlight the AND or NOT in the search builder and replace your choice with OR. Be sure to leave a space between the OR and each of the adjacent terms. Be sure to type the OR in all caps.
If you haven't seen PCR assays in the "genetic techniques" tree, you may be worried. Type PCR in the MeSH search box and hit the "Search" button.
When the result list appears, click on the "Polymerase Chain Reaction" heading.
When the detailed entry for the "polymerase chain reaction" heading appears, scroll to the bottom to look at the "polymerase chain reaction" trees. Notice that "Genetic Techniques" is a broader heading in a "Polymerase Chain Reaction" tree.
Will our "Genetic Techniques"[Mesh] retrieve articles indexed with "Polymerase chain reaction"[MeSH]?
The use of subheadings makes things a bit more complicated. As long as we search without subheadings, our search will retrieve all articles indexed with the headings even if the indexer used a subheading.
If we use a subheading, our search will retrieve articles indexed with narrower headings, but only if the same subheading (or a narrower subheading) has been used.
If you had more time, you could check on additional assays. For the purposes of this tutorial, we will assume that all assays of interest would be indexed with the two MeSH we've chosen or narrower terms. Remember, you don't need to add any of the narrower terms to your search because PubMed automatically explodes your search terms and searches for all those narrower terms!
We could go on to add terms for other concepts to the same search box, but it's often easier to join concepts using the PubMed search history.
Go ahead and click the "Search Pubmed" button to run the PubMed/MEDLINE search for your "immunologic/genetic assays" concept.
You should now be looking at a list of over 2 million PubMed records. (Remember you've just searched one concept so far). Click on the "Advanced" link under the search box (blue arrow in figure) to see how your search has been stored in the PubMed search history.
The history table should show the number of results (should be over 2 million for this search--your number may be slightly higher) in a column on the right (see blue box in figure below).
You're currently on the PubMed database history/"advanced search" page. Now it's time to return to the MeSH database/browser to find the MeSH for clostridial infections.
Click on the "Resources" option in the blue bar at the top of the page. A drop-down menu will appear. Select "Literature" and then the "MeSH Database" (see figure below).
Before searching for the clostridia difficile infections MeSH, you need to know that 1) organisms and 2) the infections caused by organisms have separate headings in MeSH.
If you were looking for articles about how to treat an infection, the heading used for "clostridial difficile infections" would be the only really important heading.
The clostridial difficile infections" heading is still the most important heading when searching for information about diagnostic lab tests. However, the organism heading (the "clostridia difficile" heading) may also be used in indexing articles about typing or differentiating one organism from another (even when specimens were obtained from patients).
The upshot: You will have to find both the infection heading and the organism heading.
When searching for infection headings, it's often easiest to find the needed heading if you "start broad," i.e. if you start with "clostridial infections" rather than with "clostridia difficile infections"
Enter
clostridial infections
in the MeSH search box and hit the "Search" button.
Look at the "Clostridium infections" tree, is there a narrower heading for the "Clostridium difficile infections" concept?
If you said yes or no, you are partially correct.
There is a heading for pseudomembranous colitis, a condition caused by C. difficile infection.
However, you can't be sure that all the C.difficile infections you might want to diagnose would be causing pseudomembranous colitis. Consequently, you'll have to use the broader "Clostridium infections" heading. Later, when you use the "PubMed advanced search" screen to combine your MeSH searches, you'll add a difficile keyword.
Look at the list of subheadings that are available with "Clostridium Infections." Is one or more of these subheadings of central importance to your search?
If you said that the "diagnosis" subheading was of central importance, you are correct. Click to check the box in front of diagnosis.
Click the "Add to search builder" button.
The formatted heading/subheading combination should now be present in your PubMed search builder (see blue box in figure below).
clostricium difficile
in the MeSH search box and hit the "Search button.
Click on the "Clostridium difficile" heading.
Look at the list of subheadings. Do any appear of central importance?
If you said "no." I'd have to agree. None of these look especially relevant to me. However, the "isolation and purification" subheading is also used to index articles about identification (Some time when you have more time, you can search the MeSH browser for ---isolation and purificiation -- and read the definition/scope note : - ).).
Click to check the box in front of "isolation and purification" (see arrow in figure below).
Make your choice from the drop-down under the "PubMed Search Builder" (see arrow in figure)
Did you select OR? If so you are correct! If not, edit the text in the "PubMed Search Builder."
Hit the "Search PubMed" button
You may want to click the "Advanced" link just to be sure you now have two concepts in your search history. This step isn't important when making normal use of PubMed, but we've found that using PubMed inside a tutorial sometimes causes problems with history storage -- so please check on your history.
Do you see the greater than 2 million results for the assay concept and the greater than 6,000 results for the clostridium concept in your search history? If not, please ask for help. You can call 402-559-6221 and ask for Roxanne or Cindy or e-mail rcox@unmc.edu or cmschmidt@unmc.edu.
If you do see both concepts with the expected numbers of results in your search history, go back to the MeSH database to search for the next concept.
Use the "Resources" drop-down to select "Literature" and then "MeSH database."
It would make sense to search for the toxigenic culture concept next. However, one of the things that the experienced MeSH searcher knows (and that you should learn) is that there are no good headings for bacterial cultures. Articles about bacterial culture are generally indexed with the organism name (or name for the group of organisms of interest) and the "isolation and purification" subheading. However, this same combination is used for most articles about identifying an organism -- regardless of the type of assay used.
Later, when you combine your MeSH searches using the PubMed "Advanced Search" page, you will use a keyword search for the toxigenic culture concept.
This leaves the sensitivity and specificity concept.
Search the MeSH database for
sensitivity
Use the "Add to Search Builder" button to add the formatted heading to your "PubMed Search Builder" and then click the "Search PubMed button.
There are over 400,000 results for this search.
Now click on the "Advanced" link to go to the "Advanced search" page.
It's now time to join the "sensitivity and specificity," "clostridial infections" and the immunologic/genetic assay" concepts.
Will you use AND, OR, or NOT to join these concepts?
You want at least one term for each of these concepts to be present in the each PubMed record you retrieve.
(Some of you may find thinking about the commercial that asks whether you want "Bed AND breakfast" or Bed OR breakfast" -- you may not care whether the bed is a double, king OR single, but you want a bed. You may not care whether the breakfast is ham, toast, croissants, OR eggs, etc. but you want a breakfast.) When you join disparate concepts and want one of each, use AND.
Use the "Builder" on the "Advanced Search" page to join the three MeSH searches. You can either left click on the search numbers in your history table and choose the desired AND, or you can type into the search box.
If you click on the numbers, a lot of extra unneeded parentheses will be added to your search strategy.
If you type the numbers, be sure to put a numeral sign directly in front of each of your search numbers (your numbers may be different than the numbers in the figure below). Also, be sure you leave a space on each side of each AND and be sure that you type the AND's in all caps.
l
You may remember that we still need to represent the toxigenic culture and difficile concepts with keywords.
In the second search box, type
cultur*
Perhaps you're wondering why you should use --
--rather than--
"toxigenic culture"?
It's a good question.
The answer is that there are so many ways an author might express the toxigenic culture concept in a title or abstract , e.g.
that it's easier to just use the term cultur*
-- instead of --
culture
Another good question.
Culture just retrieves records containing the word culture.
Cultur* retrieves records containing the words:
cultures
culturing
cultured
cultural
culturally
Obviously, we don't want articles about cultural or culturally, but these probably will not come up in articles about specificity or sensitivity of clostridium assays. The other terms could all be used in relevant articles.
After you type cultur* in the fourth search box, move on to the third search box.
You want articles that are focused on C. difficile. Most articles (almost all) that are focused on C. difficile identification will have the term difficile in their titles. Use the drop-down next to the third search box to select "Title." Type the term --
difficile
-- in the fifth search box (see red box in figure below).
Hit the "Search" button (see green arrow in figure above).
You are now looking at the results for your search. There should be over 140 results for this search. Don't be discouraged if the first one or two results seems irrelevant. Keep browsing through the results. You'll find that most are highly relevant.
Perhaps you want to start by looking at the highest quality evidence. Well-done systematic reviews or meta-analyses are considered the highest quality evidence. The authors of these articles have tried to find all high quality studies published on a topic, critique the quality of the studies, and then use the results of the best studies to come up with a recommendation or overall finding. You can limit your search to systematic reviews, meta-analyses, consensus statements, and practice guidelines (with a few undesired types of articles occasionally retrieved) by using a "systematic review" "article types" limit.
PubMed's limits are called "Filters" and are shown on the left-hand side of the results page. Because there are so many available filters, not all are shown in the default view.
Click the "More" under "Article Types"
In the menu that appears, select "systematic reviews" and then click "Show". This just makes the limit you want available.
f
To actually use the filter/limit, you have to click on it after it appears in the list on the left-hand side of the page.
When you've successfully chosen a filter, the page will refresh and a check mark will appear in front of the filter.
You may want to read one or more of these.
Start with the article from "Lancet Infect Dis" (#3 on 07/08/2015). Click on the link to the detailed record.
Look at the buttons.
There is a "Free Summary in DARE" button. These DARE buttons are new addition to PubMed. They're slightly misleading. Rather than taking you to a free full text version of the article, they take you to the full text of a CRITIQUE of the article from DARE (Database of Abstracts of Reviews of Effectiveness).
There is a Lancet button on top. This button doesn't say "free" or "open access" If you are on campus and the Library happens to buy access from the publisher, this button may work for you. However, the Library buys full-text from many non-publisher sources.
it's best, when you don't see a "free" or "open access" button, to start with the "GetIt!@UNMC" button. Be Warned, when you click on the "GetIt!@UNMC" button a new window or tab will open. You will need to click back and forth between the new window/tab and the tutorial window/tab.
If you haven't already done so, click on the "GetIT!@UNMC" button.
In the left-hand frame, you should see a page from Clinical Key that contains the abstract of the article of interest. Sometimes, rather than going straight to the abstract for the article of interest, you will end up at the homepage of the full-text source and have to search or browse that website for the article you need.
The citation for the article (article title, authors, journal title, journal issn, volume, issue in parentheses, year, pages) appears toward the top.
Find and click on the .pdf link/icon to see the full-text article. Each journal/publisher has it's own way of displaying these .pdf icons/links. At times, you may have to hunt a bit to find the needed "full-text" or "pdf" icon or link.
You now probably have several new windows or tabs open. Close these to return to your PubMed search results.
Perhaps you want to see more articles in which toxin detection kits are compared to the gold standard assay for C. difficile. You may be able to find additional relevant articles using PubMed's "Related citations" feature.
"Related citations" searches use a combination of MeSH and keywords from a record to find similar articles. Because keywords are included, you can find unindexed (non-MEDLINE) records as well as indexed records using a "Related citations" search.
To do a "related citations" search:
If you had been looking at a list of search results, you would have seen a "Similar Articles" link under each of the search results and could have used that link instead of the "see all" link shown above.
Use the "Sort by..." drop-down above the result list to select "Pub Date"
and then use the "Summary" drop-down menu to select "Abstract".
When this tutorial's author sorted the results of the "related citations" search by "Pub Date," it was possible to see that some of the retrieved results were "in process" or "as supplied by publisher". These are some of the unindexed results that can be retrieved by a "Related citations" search.
The new tab or window that appears will display the message "Sorry no holdings were found for this journal" (see red box in screenshot below). This means that the McGoogan Library does not purchase electronic access to this full-text article.
The Mcgoogan Docs sign-in page will appear in a new tab/window. If you really wanted this article and hadn't previously created a "McGoogan Docs" account, you would have to start that process by hitting the "Create a McGoogan Docs account" link. You don't need to do this now. You can wait to sign up for an account until you actually need an article. When that day comes, just follow the instructions on the "Create a McGoogan Docs account" page, and -- when asked for a billing method-- add the "UNMC free" option to your billing method table.
If this is an article you want, don't despair. The library can almost certainly obtain the article for you through interlibrary loan. As students, there is no charge for requesting articles from interlibrary loan.
To start the article ordering process, close the new tab/window for the catalog search. When you're back at the pop-up produced by the "GetIt@UNMC" button, click on the "McGoogan Docs" link.
The Mcgoogan Docs sign-in page will appear in a new tab/window. If you really wanted this article and didn't have an account yet, you would have to start by hitting the "Create a McGoogan Docs account" link. You don't need to do this now. You can wait to sign up for an account until you actually need an article that the library doesn't license/have. When that day comes, just follow the instructions on the "Create a McGoogan Docs account" page, and -- when asked for a billing method-- add the "UNMC Free" to your billing method table.
If a request form isn't automatically filled in when you finish creating your account,
When you request articles, be sure to watch your e-mail. Download and save the articles to your computer and/or print the article. The library can only post articles for 30 days. After that time you would have to re-request the article.
Over 90% of routine interlibrary loan requests are filled within 2 weekdays.
When looking for full-text always watch for "free" or "open access" buttons. PubMed includes MANY of these.
Sometimes it's useful to look at a PubMed records subject heading. Perhaps you've found one really relevant articles and think you might be able to construct a good search if you knew how that article was indexed.
To look at the subject headings of a record that appears among your search results, you must look at the "Abstract" view. Either, use the "Display Settings" menu to select and "Apply" the "Abstract" view or click on the title of an article of interest.
When the "Abstract" view appears, click on the "Publication Types, MeSH Terms" link to see the MeSH and other terms added by the indexer.
Right click on this link and choose the "new window" option to see the abstract view of a record with its various parts labelled.
This is also the area where the indexer adds information about whether the article is about humans or animals, females or males, children or adults, whether it's a review article or is reporting the results of a clinical trial, etc. The filters for humans, age, gender, and most article types depend on indexing.
Point to remember: You can't retrieve an unindexed record when you apply a "humans" limit.
Of the frequently used filters/limits, only the "systematic review" filter, the publication date filter, and language filter allow retrieval of unindexed records.
Congratulations! You have reached the end of this tutorial.
The next page of this tutorial mentions a quiz. There is no quiz included in this tutorial -- ignore that portion of the following page.
We will pay attention to any comments you would care to make on the tutorial. You can make comments anonymously using the "What did you think of this tutorial?" link on the following page.
If you want to save a copy of the entire text of this tutorial, just click on the print icon at the top of this panel. The contents of the tutorial will appear on one web page for printing, saving, or copying and pasting into a Word document.
Please enter your name and email address to retrieve a copy of your completed quiz.
You can enter multiple email addresses separated by commas. If you are doing this for a class, you may need to enter your instructor's email address also.