Occupational and Environmental Health Searching for Information - - PDF document

Searching for Information

Occupational and Environmental Health

Evans Whitaker, MD, MLIS UCSF Library evans.whitaker@ucsf.edu

Disclosures

I have no disclosures to report

Goals

Build a search from a question using new PubMed. Along the way I will introduce some tools librarians use to help with search development and getting the full text of articles

Objectives

By the end of this session you will:

• Be more familiar with the structure of PubMed
• Will see new PubMed in action
• See how to create a thorough search starting with a question
• Hear how to you may generalize the approach used in PubMed to
• ther databases
• Be introduced to some tools which will help you develop your

search, save your work, and find full text of articles

Legacy to New

5

New PubMed

New PubMed appeared about a year ago as PubMed Labs Now linked from a larger banner in “legacy” PubMed It will become default interface in the next few months Legacy PubMed will be “retired” by end of 2020…

Terminology

PubMed MEDLINE Controlled vocabulary Keywords PICO 5 A’s of EBM P = Problem, Patient, Population I = Intervention or Exposure C = Comparison group O = Outcome Ask Acquire Appraise Apply Assess MeSH or Emtree

Controlled vocabulary

Most databases have a dictionary or glossary of preferred terms. If a search uses these words, the database will look for synonyms at the same time it looks for the preferred term: Example from MeSH in PubMed: Occupational Medicine

 Entry Terms (= synonyms)

○

Medicine, Occupational

○

Medicine, Industrial

○

Industrial Medicine

MEDLINE vs. PubMed

PubMed 30.5M MEDLINE 26.5M

Publishers supply information

Searching for information for biomedicine

Searches for information are usually of two types:

 Quick search for a few good things for patient care or

verification

 Thorough search for detailed information for research or

publication Thorough literature reviews, especially those for systematic reviews, search in more than one database. Cochrane handbook states that at a bare minimum PubMed and Embase should be included.

Presentation Title 13

A question to work with…

Courtesy of Dr. Blanc

Is carbon disulfide associated with adverse reproductive outcomes?

• How many concepts?
• What are they?
• How would you use PICO on this?
• What issues do you see with this

search? Fully defining adverse reproductive outcomes will be challenging 2, CS2 and adverse reproductive outcomes P - may not need to define I (OR E) - CS2 C – not defined O – adverse reproductive

• utcomes

Problems with this search…

 The adverse reproductive outcomes will be hard to full characterize  Switch to new PubMed

to explore

So what did we do there?

 We developed a question, this might have come from a patient, from research we want to do, or

the topic for an invited paper…

 We parsed the question into concepts. We might have used PICO to help  We found controlled vocabulary (MeSH), thought of synonyms/ keywords to describe each

concept

 We used Boolean operators (AND, OR, rarely NOT) to assemble the search and added

parentheses and quotation marks as needed.

 We tuned up the search based on results. A good search should find a list of “known” good

articles)

 We saved our work

15

This approach will work for any database!

 Databases each have their own controlled vocabulary (e.g., Embase uses Emtree)  Databases in which you would look for information related to biomedicine have extensive

• verlap. Each database will contribute a few things to the good results.

Tools to help

 Yale MeSH Analyzer can take up to 20 PMIDs and create a table showing MeSH, author keywords,

title and abstract to help find the terminology in common between articles you like

 Unpaywall, Kopernio and OA Button help find full text of articles for you  MyNCBI saves your searches, keeps you updated, can send tables of contents of selected journals

to you, and can save collections of articles

 Embase and other databases have functionality similar to MyNCBI

Tools to help

 That was a whirlwind tour  If you have questions about the content or need clarification feel free to email me at

evans.whitaker@ucsf.edu

Thanks for your attention!

• Welcome to your template!
• To explore slide and style options, head to the ‘New Slide’ button on your toolbar and select the arrow next to

it.

• You will find two design themes: classic & contemporary.

Questions for Turning Point

MeSH is ______ (pick all correct answers)

1.

A controlled vocabulary of preferred subject terms used in PubMed.

2.

A means to remove leaves from brewed tea.

3.

Stands for Medical Subject Headings.

4.

Used by all databases.

5.

Includes a list of “Entry terms” for each MeSH

6.

Entry terms are synonyms for MeSH

PubMed and MEDLINE are the same thing.

1.

• 1. True

2.

• 2. False

Presentation Title 24

Once you have run a search what tool built into PubMed will help you understand how PubMed interpreted your search?

1.

• 1. Details

2.

• 2. MeSH database

3.

• 3. MyNCBI

4.

• 4. Advanced

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 1 Hi everyone, I am Evans Whitaker from UCSF Library. If you have questions after taking a look at this please contact me at evans.whitaker@ucsf.edu We have 40 minutes to talk about how to search the scientific literature for the information you need to help take care of your patients, teach students, do your research, and write your

• papers. As you know, and as I have found out while preparing this talk, the information for your

field of Occupational and Environmental Health is widely spread throughout the biomedical literature and not conveniently located in a few key journals. I expect there is a great deal of diversity in the group. Some of you have access to multiple databases and extensive full-text journal article access, others will be limited to PubMed and Google Scholar for searching and whatever full text to which you, your practice/medical group,

• r your hospital/medical system purchase. I have tried to make this presentation relevant to this

broader spectrum of information seekers. Today we will focus on:

• 1. New PubMed, we can imagine two kinds of searches…
• a. A search for a few good things, I suspect you all are already good at this…
• b. A thorough search as if for a systematic review or to become well-versed in a

particular topic or as background for conducting your own research.

• c. I will try to emphasize principles you can apply to the search of any database.
• 2. Tools:
• a. Use of MyNCBI and other database-associated personal accounts to save your

work and receive updates

• b. Google Scholar can be used in that same way.
• c. Other useful tools I will mention include Unpaywall, Kopernio, OA Button to help

find legal full-text. We do not recommend or condone the use of SciHub. PICO as a framework for creating questions. Yale MeSH Analyzer to help with finding terms to use in your search.

• d. A reference manager (like EndNote or Zotero) to store, organize, remove

duplicates and add citations and a reference list to document you are writing.

• e. Evernote or OneNote adds a second way to quickly store useful information you

find that reference managers do not do as well with (images or tables from an article, parts of a website, etc. I want you to leave this session today ready to try out the new PubMed and with an approach to the process that carries you from identifying a question to storing and organizing the information you found in a reference manager. Background and introduction which applies to old and new PubMed. PubMed is made of two parts. MEDLINE (26.5 M) is 86% of PubMed and is well organized and easier in which to search.“The part of PubMed not in MEDLINE” (4 M) is where new material is housed until indexed (which can take days to years depending on the journal). See graphic next page to explain.

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 2 MEDLINE articles have MeSH terms (which describe what the article is about) and other added information added. The added information includes species and sex of subjects, language of the article, publication type, subset information, etc.). dPubMed has a very attractive set of “filters” and an even more enticing set of subheadings to

• apply. The only filter you can use without eliminating the newest articles from your results in the

blue crescent of the graphic from your results is date and the various subsets. This two-part structure requires thorough PubMed searches include MeSH terms keywords/synonyms. No other database has this structure This warning does apply to Embase, Web of Science, etc. New PubMed: PubMed is freely available worldwide beginning in 1996. The graphic below shows number of articles added by year to PubMed and key events in the evolution of PubMed. PubMed now holds 30.6 million articles. It is estimated it indexes about ½ of the worldwide “reputable” biomedical journal literature. The ½ it holds is generally high-quality and high-utility. Other databases, like Embase, will add incrementally to what is found through a PubMed search. It is generally considered a bare minimum for a systematic review to search both PubMed and

• Embase. For those of you relying on PubMed alone for clinical information, it will serve you well.

About 25% of PubMed content is freely available to all and that amount is increasing rapidly. See image top of next page, notice that ~7.6 M of ~30.6 M articles are freely available (25%). About 43% of articles added to PubMed in 2019 are freely available.

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 3 Access to the rest of those articles depends on personal or institutional subscriptions or professional association memberships. Purchase of the full text of journal articles is expensive. It cost $25 to $60 dollars per article for direct purchase from a publishers website. Open access journals are on the rise and there is increasing pressure on traditional publishers to control prices to end-users and institutions and increase access to the results of biomedical research. The ongoing UC system boycott of Elsevier articles is one example in which a line was drawn to stop price gouging. The UC system has joined the nations of Germany, Sweden and others. PubMed has had about the same functionality and appearance since arrival in 1996. New PubMed is a major upgrade. About a year ago “new PubMed” appeared as PubMed Labs. Now there is a large banner directing users to new PubMed from what is now being called “legacy PubMed”. Sometime this year new PubMed will become the default PubMed interface and the legacy PubMed will disappear 6 months or so after that change according to the National Library of Medicine. Librarians have mixed feelings about new PubMed as some of the tools we use all the time are not (yet) easily available in new PubMed. For end-users, I believe you will like the design and the function of the new PubMed. The algorithms that run in the background are more robust making the use of PubMed easier. They are still working on integrating some of the functions beloved by librarians and as a result what we look at today will change over time. Tools to help (links to more information can be found at the end of this document)  MyNCBI is your personal space at the National Library of Medicine. From here you can save searches, receive notifications of new articles found by the search your saved. MyNCBI will make seeing what kind of publications your search finds. You can set up MyNCBI to receive selected journal table of contents in email or RSS readers.  UnPayWall, Kopernio, OA Button are plugins that looks for legal copies of full text of articles.  Zotero is a free, open source reference manager that works very well. It is a worthy competitor and easier to use than the industry standard product, EndNote. It allows you to rapidly save results from searches so that you have them available to you when you want them again and it will add citations and references to Word, OpenOffice and Google Docs.  Consider EverNote as a way to say information that does not easily go into Zotero.

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 4  Yale MeSH Analyzer simplifies finding words to use in a search by analyzing title, abstract, MeSH and author keywords of article you have already identified about your topic. Use of PubMed: We will use an example. This section attempts to make your unconscious thought processes overt and is based

• n the 5 steps suggested for Evidence Based Health Care (= EBM and other wordings):

Ask – Create your question Acquire – Search for information Appraise – What is the quality of the information you found Apply – Use it for clinical care, write your review Assess… repeat the preceding steps as needed. We will focus on the first two steps today. Depending on your level of experience and how often you search for journal literature this may all be self-evident Let’s try this with a simple problem: I like to work with a specific example and Dr. Blanc provided this idea. Is carbon disulfide associated with adverse reproductive outcomes? This is a succinct question. It is composed of two concepts:  Carbon disulfide  Adverse reproductive outcomes The first concept is straightforward. The second concept covers a lot of ground and will need to be more fully specified. What that specification looks like will depend on the controlled vocabulary of the database. Taking a quick look in MeSH and Emtree there is no exact match for this wording. Let’s take this step-by-step. A first shot at a search in new PubMed: carbon disulfide AND adverse reproductive outcomes Note: Librarians recommend avoiding quotation marks, asterisks and other punctuation in the first attempt. Wait to see how PubMed translated the search and iteratively adjust from there. The first shot is translated by the new PubMed algorithm as: ((("carbon disulphide"[All Fields] OR "carbon disulfide"[MeSH Terms]) OR ("carbon"[All Fields] AND "disulfide"[All Fields])) OR "carbon disulfide"[All Fields])

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 5 AND ((("adverse"[All Fields] OR "adversely"[All Fields]) OR "adverses"[All Fields]) AND (((((("reproduction"[MeSH Terms] OR "reproduction"[All Fields]) OR "reproductions"[All Fields]) OR "reproductive"[All Fields]) OR "reproductively"[All Fields]) OR "reproductives"[All Fields]) OR "reproductivity"[All Fields]) AND ("outcome"[All Fields] OR "outcomes"[All Fields])) The search finds 4 results. They are not great. We need to work further on this search. As you can see PubMed found the MeSH term for carbon disulfide and reproduction which is

• good. It provided some variations on the wording as well. The CS2 concept looks good. For

adverse reproductive outcomes, PubMed requires each article have all three words (or variants) in the title or abstract. This is why the result number is low. We can do better than this. We can take the clue PubMed gave us, search MeSH for reproduction and variants. Think about what else you would care about. The 4 articles found use the terms infertility, spermatogenesis, and miscarriage in title or abstract. By looking these up in MeSH, you can find good terms and synonyms (called “entry terms”). MeSH is not yet well integrated with new PubMed. We need to get back to the old MeSH interface and transfer what we found there to new PubMed. From new PubMed we can get back to legacy PubMed by clicking the banner at the top of the page. There are two links to the MeSH database here (see image top of next page): Look up reproduction or reproductive. We find: (("Reproductive Health"[Mesh] OR "Reproduction"[Mesh] OR "Reproductive Physiological Phenomena"[Mesh] We might think about things not found yet in MeSH. We mentioned words seen in the original 4 results:

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 6 infertility OR miscarriage OR spermatogenesis We might also think about "Mutagenesis"[Mesh] OR "Teratogenesis"[Mesh] We would add keywords as well to make sure we search in the part of PubMed not in MEDLINE. Our more complete search might then be: ((("carbon disulphide"[All Fields] OR "carbon disulfide"[MeSH Terms]) OR ("carbon"[All Fields] AND "disulfide"[All Fields])) OR "carbon disulfide"[All Fields]) AND ("Reproductive Health"[Mesh] OR "Reproduction"[Mesh] OR "Reproductive Physiological Phenomena"[Mesh] OR reproducti* OR “infertility”[mh] OR infertility OR spermatogenesis OR “spermatogenesis”[mh] OR “abortion, spontaneous”[mh] OR abortion OR “pregnancy loss” OR miscarriage OR "Mutagenesis"[Mesh] OR mutagen* OR "Teratogenesis"[Mesh] OR teratogen*) Note: If not familiar, “ ” around a phrase makes PubMed find just exactly those words in that

• rder. Do not use quotation marks around single words (usually). The asterisk at the end of a

word searches for all word endings for that word. For instance, child* finds child, children, childhood and many others. The use of both quotation marks and the asterisk come with caveats, but that is too deep a level of minutiae for this presentation. This search finds 217 without any limits. A limit you can apply in PubMed that cuts out most animal studies is from Cochrane and goes like this: NOT (animals[mh] NOT humans[mh]. Applying this limit cuts 217 to 139. Now look at the results page, at the top right is the sign in to MyNCBI. Beneath this are options for how the results are sorted and how the page will appear. Default is now relevance ranking (PubMed calls this Best Match) based on the new and improved algorithm, there is also date sorting as well. You can opt to show the Abstract. You can show anywhere from 10-200 results

• n a page.

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 7 Red arrow link back to legacy PubMed Green arrow  sign in to MyNCBI Purple arrow shows display option If you look at a single article, there is navigation on the right column that shows you a variety of useful

• information. See image to right.

Generalizable points from our PubMed example you can use for creating a comprehensive search in any database:

• 1. “Think before you type”, try to come up with

a single sentence that summarizes your question.

• 2. Identify the concepts (usually the nouns in

the sentence) that make up your search. The PICO framework is a way to specify your question and identify the concepts that make it up. It works very well for intervention and exposure questions. Make your question pretty specific or, given the size of PubMed (>30 M, Embase >38 M, and the reach of Google Scholar (who knows?)) you will be overwhelmed with results.

• 3. Identify controlled vocabulary for the database and use those terms in your search. This

is MeSH for PubMed, Emtree for Embase, other databases have their own. Sadly, there is no one standard vocabulary. The advantage to taking the time for this step is that the database will do a synonym search for the controlled vocabulary terms, increasing the sensitivity of the search.

• 4. Identify all likely synonyms, come as close as you can to listing all the various ways an

author might word a concept. MeSH and Emtree can help with this.

• 5. String the controlled vocabulary and synonyms together with OR (you do not have to

capitalize) to create thorough descriptions of each concept in your question.

• 6. String the concepts together with AND.
• 7. Run the search, and prepare to revise, as no one gets it right the first timeI
• 8. Avoid adding filters by year, language, publication type to start with. There may be very

logical reasons to add limits to your searches. However what I see is that searchers

• ften use this approach to make up for a poorly constructed search that finds too much.

You can add limits later. Additional Notes: I thought this was interesting. Ioannidis JPA. Meta-analyses in environmental and occupational health. Occup Environ

• Med. 2018;75(6):443–445. doi:10.1136/oemed-2016-104128

UCSF Occupational and Environmental Health Conference – Library presentation 5-7 March 2020

Have questions about this handout or the presentation? evans.whitaker@ucsf.edu 8 (Meta-analysis [sb]) OR meta-analysis [ti]) AND (work [ti] OR worker* [ti] OR working [ti] OR occupation* [ti] OR pesticide* [ti] OR dust [ti] OR hydrocarbon [ti] OR chromium [ti] OR cement [ti] OR employment [ti] OR job [ti] OR trichloroethylene [ti] OR solvent* [ti] OR pilots [ti] OR phone [ti] OR phones [ti] OR telephones [ti] OR particulate [ti] OR DDT [ti] OR professional [ti] OR by-product* [ti] OR benzene [ti] OR exposure [ti] OR electric [ti] OR industry [ti] OR voltage [ti] OR pollut* [ti] OR hairdresser* [ti] OR “passive smoking” [ti] OR “environmental tobacco” [ti] OR cadmium [ti] OR arsenic [ti] OR copper [ti] OR lead [ti] OR ‘heavy metal’ [ti] OR dioxin* [ti] OR biphenyl [ti] OR chlorpyrifos [ti] OR toluene [ti] OR herbicide [ti]) Links to related content mentioned above

• 1. Set up MyNCBI, to save searches, receive updates, set filters, and receive electronic

tables of contents of selected journals. Link to handout with instructions: https://ucsf.box.com/s/cpeupidiw7uh70sfrvgeuyidf574l3no

• 2. Set up and use Zotero: link to handout for install and use:

https://ucsf.box.com/s/lb5pb4klhi3mdtpj9zn8sueezotxvcyl

• 3. Kopernio: information and install here: https://kopernio.com/
• 4. Unpaywall information and install here: https://unpaywall.org/
• 5. OA Button information and install here: https://openaccessbutton.org/
• 6. Evernote, there are free and paid versions. Combined with Zotero (or other reference

manager) gives you a complete information storage solution. https://evernote.com/ . OneNote is a viable alternative and is often included in MicrosoftOffice suite. As you have likely seen, ToxNet has “moved”. NCBI has put the content into different

• databases. including PubChem, PubMed, etc. There is now a toxicology subset in PubMed you

can invoke by adding AND tox[sb] to your PubMed search. Thanks!