Transcript for Search Clinic: Chemicals & Drugs in PubMed

Event ID: 760373
Event Started: 8/23/2007 1:56:23 PM ET
Please stand by for real realtime captioning.

Welcome. Welcome to the search clinic on chemicals and drugs in PubMed. I am Katherine Majewski, and I am from the U.S. National Library of Medicine. With me is Barbara Nicholson from the Network Office here at NLM, and in New York is Sharon Brown from the National Training Center and Clearinghouse and the New York Academy of Medicine.

The goal of this clinic is for participants to understand how the NLM MeSH vocabulary is used to describe chemicals, drugs and other substances and their actions and therefore how to search PubMed for relevant articles. Note that the trainers are not chemistry experts and will not be prepared to answer in-depth questions about chemistry.

The intended audience for this clinic is librarians with little to moderate experience in searching for articles about chemicals or drugs in PubMed. There is one prerequisite, and that is a basic understanding how to search PubMed. Before we begin, please maximize the meeting window if possible use your F11 key. Note the attendee list on the left-hand side of your screen. It should look like this. Please click on the gray my status bar and I have us a thumbs up if you can hear me well and you can see a dark bluish gray PowerPoint slide on the screen. I see a lot of thumbs up. That makes me very happy.

Note that you will occasionally see a blue and white cross-hatched screen as I move between windows. If at any point you experience technical difficulties, please change your status, or if you can let us know what is wrong using the chat pod on the lower left of your screen. If you lose your connection for more than 20 seconds or so, please exit the meeting room and reenter. Usually that resolves the problem. Note that this session will be archived at the URL provided in your e-mail if you wish to review the class at a later date.

Please use the chat pod to ask us questions as they come to you. Sharon will be monitoring the chat and will try to answer your questions as best we can at the end of the clinic. Note that the scope of this clinic is PubMed searching. If you have detailed questions about NLM indexing or MeSH, we are happy to take them, but we may need to defer to experts in those areas and get back to you after the clinic.

First, we have a question for you: If someone asked you to find the latest research on 1,3-bis(2-chloroethyl)-1-nitrosourea and glioblastomas, how confident would you be about your approach? Go ahead and use your mouse and select the confidence level you feel. [Sorry. I did not open the poll. Guess I need to do that.] Still getting a few results in. All right. Very good.

I guess we will close the poll there and I will show you how people voted. It looks like a little over half of are you somewhat confident. Thank you for your honest self appraisal.

Here are our objectives. By the end of this session you should have a basic understanding of:

We're going to start in PubMed. You should see the PubMed screen coming up. [Seems to be a bit of a delay. I will try to share that again because I am not seeing that. There we are. Sorry about that.] We will start in PubMed and do a search for a substance. We're going to look for acetylcarnitine. Chemicals and drugs are part of the MeSH vocabulary used for indexing articles for MEDLINE citations. Acetylcarnitine is a MeSH heading. If we go to detail, we see an untagged search of PubMed will be translated to acetylcarnitine as a MeSH term or acetylcarnitine as a text word. As an untagged search of any other MeSH heading will be translated.

To find more information about a substance and how to search for it in PubMed, use the MeSH database, a database of the medical subject headings, our controlled vocabulary. It is available from the PubMed side bar under PubMed services. In the MeSH database I will return the same search for acetylcarnitine. Note that I have used my NCBI account to set up highlighting so you will see my search terms highlighted in green. Going to the full record for acetylcarnitine in the MeSH database, and then scrolling down, we see the entry terms. Chemical names have many variations, so the controlled vocabulary can be very helpful, even if you aren't aware of it. Searching using any entry term les automatically retrieve citations indexed with the preferred MeSH heading.

Chemicals and drugs make up an entire category in the MeSH hierarchy. You can explore the category from any MeSH heading using the link from the hierarchy at the bottom of the MeSH database record. I am going to scroll down to the hierarchy for acetylcarnitine where I can then go all the way up to the chemicals and drugs category. Here is just a very beginning of the chemicals and drugs category, and I won't make you nauseous by scrolling through the whole thing, but note that chemical and drug classes exist in plural form in MeSH and are also searchable. As you know a search of any broader term will automatically include all narrower terms indented under it. This is called the explosion feature.

In general chemicals in MeSH are entered with their common name rather than their systematic chemical same or structural formula. For example, I will click on amino acids. If I am looking at amino acids, I will find MIMOSINE, not 3-hydroxy-4-oxo-1(4H) pyridinealanine also MIMOSINE or C8H10N204.

Let's look at another substance concept, actually we'll do a search, another search in PubMed. For the substance melamine. It was in the news a lot this spring. If I look at the details here I see something a a little different from my previous search. You may notice some chemicals are not included as MeSH headings but searches of them map to something called a substance or substance names. Going back to my search results, and looking at one of the citations that were indexed, note that melamine occurs only under substances in indexed records. Not all chemicals in the MeSH vocabulary are MeSH headings. MeSH headings are updated annually, but new substance concepts are identified by NLM staff regularly while they are reviewing articles for indexing. Substances not found in MeSH are forwarded to a chemist for review. If a substance is deemed a new concept, it is added to the MeSH database as a substance name.

I am going to now link over to the MeSH database from melamine to take a look at the entry in the MeSH database, and we see that it is labeled with substance name. You may know this as a supplementary concept. We are going to use the term substance name for the purpose of this clinic as that as the label in the MeSH database and PubMed. Currently there are over 170,000 substance names used in MeSH. The substance record always include a map to go a MeSH heading. New citations indexed with the substance are automatically indexed with this MeSH heading. For the example melamine, whenever an indexer adds the substance melamine to a citation, triazines is added automatically. The idea is that this is more general MeSH head to go this substance.

Going back to the indexed record, we see triazine as a MeSH term and melamine as a substance. What this means for searching is searches of the usually broader mapped headings will retrieve citations indexed with the associated substance, and you can search specific substances in PubMed almost immediately after the substance is added to the database as the substances are added to the index weekly.

At some point a substance name may be promoted to a MeSH heading during the annual MeSH update. When this happens, records with the substance are given the new MeSH heading for that substance. The matched MeSH heading like triazine for melamine are also left on the record.

I am going to take a look at another example now in the MeSH database, [and before I do that, I am actually going to -- I neglected to mute the speaker phone I am on and I think it is causing echo for some people. When I did this before it muted me as well, so I am hesitant to do this. I will try it, and I am going to rely on you folks out there to use the chat pod and let me know if I muted myself as well. That would be a bad thing. I am hoping you can still here me.]

I am going to go to the MeSH database. Using the side bar link. And I will do a search for oseltamivir also known as Tamiflu and we're looking at the full record for this substance. Here is a term recently added as a preferred MeSH heading. When you see two dates in MeSH for a substance heading, it means that the term was added as a preferred MeSH heading on the more recent date but the concept is searchable back to the date in parantheses.

To search for substances in PubMed, if you're not tagging a search for a specific substance will find either MeSH terms or substances or instances instances of terms entitled abstract or some other field as we've just seen. A search using [nm], or spelled out as substance name, will retrieve the term as either a substance or a MeSH term. Here I can search for melamine or oseltamivir with the NM search tag and both will work. Conversely, if you search by MeSH heading or with the MH search tag you will restrict your search to preferred MeSH headings and entry terms and will exclude this very large set of substance names. In other words, if you want to tag your search, using NM is safest as that will find a substance whether it is a MeSH heading or a substance name.

Let's take a moment and note the search tags we've seen so far. You can see these in the note pod of this meeting. Actually, it is a pod on the lower right labeled search tags. Note particularly MeSH terms which you can tag as MH, and substance names which you can tag as NM. Remember that PubMed finds these automatically if you enter a MeSH heading, an entry term or a substance name when you use no tags.

Let's look at another sample of a substance in the MeSH database. I am going to do a search for prostaglandins G. If you're looking for work on a substance prior to when it is added as either a stants name or MeSH heading, use an untagged search to find your term entitled ab abstracts or other fields and look up previous indexing for years prior to the addition of your term as a substance name. For this example of prostaglandins G, two broader terms, peroxides and prostaglandins were used for this concept between 1966 and 1977. To search more comprehensively, search the current MeSH heading back to 1978 and one or both of the previous headings for these earlier years. For this particular example it I am looking for literature of prostaglandins G prior to 1978 my PubMed search might look something like this: 1966:1977[mhda] prostaglandins peroxides. I will give you a moment to absorb this. Note that I am using MHDA or MeSH date as I am interested in when the records were indexed, not when the literature was published.

Let's take a look at another example of a substance going back to the MeSH database, and I am doing a search for aspirin. We'll take a look at the full record in the MeSH database for aspirin. As you know, you can use MeSH subheadings to search for the aspect of the topic you're interested in. When using MeSH to search PubMed for chemicals or drugs, you may find some subheadings particularly useful. I would like to briefly point out the top three substance related subheadings that are commonly confused: toxicity, poisoning, and adverse effects. Firstly, toxicity. This subheading is reserved for studies designed to determine the ill effects of a substance. These are usually animal studies. This would include something like the study of the effect teratogenic effect of aspirin on rat fetuses. Poisoning. Poisoning is used for a life-threatening intoxication or overdose which can be accidental, occupational, suicidal, by medication error or by environmental exposure. An example might be a case study of an aspirin overdose. Adverse effects is used for side effects or complications of prescribed or over-the-counter pharmaceuticals used in accepted doseage when intended for diagnostic, diagnostic, therapeutic or prophylactic air an thet I can purposes. An example might be a study of whether aspirin use increases bleeding in risk in surgery. Note that in PubMed a search for adverse effects automatically includes toxicity and poisoning, so remember to use the most specific subheading that describes the concept that you want to search.

I am still on the aspirin record in the MeSH database, and I am scrolling down to mention this. Most substances in MeSH are assigned one or more pharmacologic action terms. These are terms used to describe the action of the substance. Note the four pharmacological actions for aspirin. Since 1996 and when appropriate, the indexer will include in the citation the specific substance term as well as one or more pharmacologic action or actions of the substance discussed in the article. You can search the pharmacological action terms. For example, to find articles about aspirin as a pharmacologic agent, what you can do in PubMed is search aspirin and fibrinolytic agent as a MeSH heading. Here I have my search and we'll take a look at the first here, and you can see your substance term, aspirin, as well the pharmacological action, fibrinolytic.

If I go back to the aspirin record in the MeSH database, I can show you from the MeSH hierarchy I can go to all MeSH categories and show you the pharmacological actions category. You can see a list of all of pharmacologic actions here. These records list the substances identified as having that action, so if I go to air pollutants, I will see the substances assigned to this action.

A search of a pharmacologic action term either untagged or with a PA tag will automatically retrieve citations indexed with any of the substances listed in that term. In this example a search of PubMed citations indexed with any of these substances. Note that substances are often a mix of MeSH headings and substance names.

You can combine a search of, for example, a disease topic and a pharmacological action of interest to find relevant articles. Again, in PubMed, I can combine a search of asthma with a search of air pollutants as a pharmacological action using this PA tag. This search will retrieve citations indexed with asthma or a more specific term and any one of the substances listed as air pollutants. Again, I will take a look at the first index record here. We have asthma. We have RADON listed as an air pollutant.

Again, if you want to search for literature about a specific action of a particular substance, let's say you're interested in literature on bisphenol A. If you want to search for the pharmacologic action only as a MeSH heading,, if I am interested in bisphenol A as an air pollutant says, I want to tag air pollutants with MH or MeSH heading. You are not including all substances with that action. You're interested in article that is identify bisphenol A as an air pollutant. Therefore you want to retrieve citation includes both the MeSH heading air pollutants and the term bisphenol A.

Note that if not tagged, (I will do a search of air pollutants not tagged,) PubMed will search a pharmacological action term as a PA or as a MeSH term or as a text word. The MH or the MeSH term search tag instructs PubMed to search for the action as a subject, the PA or pharmacological search tag instructs PubMed to include all substances with that action.

I am going to switch gears a moment and go back to my PowerPoint. To talk about steroisomerism. Steroisomerism or the spatial arrangement of atoms is generally ignored in indexing with some exceptions. Therefore the letters D, L, DL, R, or S, or the symbols plus or minus before a substance name may be ignored. These refer to stereoisomers. For example, D L tryptophan is a specific steroisomerism of tryptophan. They have the same molecular structure, but the DL refers to a specific spatial arrangement different from L tryptophan. In the MeSH vocabulary either would be indexed as tryptophan. In searching you may want to first search the specific stereoisomers, but strip the D, L, R, S or plus or minus signs in a second search to get more comprehensive results. If a specific stereoisomers is of interest, the indexer will add the term steroisomerism to the record. For example, recently will has been many interest in a stereoisomers of thalidomide. Let me go back to PubMed here. This is an anti-angiogenesis agent for use against cancer. For this you might search thalidomide steroisomerism or actually what I would do is be even more specific: take advantage of the pharmacological action term available, angiogenesis inhibitors.

Another thing of note about searching for substances, organic salts, which, for example, could be substance terms starting with sodium or potassium or ending in hydrochloride or sulfate and often use the term salt, are usually indexed under the general compound. So, for example, if you search for sodium iodoacetate, excuse me, sodium iodoacetate, you'll get results, and if you take a look at the details, you'll see that it maps to iodoacetates.

Also, an ATE ending such such as nitrates is used interchangeably with the parent acid term like nitric acid. You may be searching for the bile salt tauroursodeoxycholate. I am getting results, but if I check my details, I see that it didn't map to anything. If PubMed doesn't map to a substance name when you're looking for a substance, you may be missing relevant results. I have a little trick for you. If you go to Preview/Index, scroll down to where you can view the index and if you select the substance name index you can search for everything in substance name. So if I try to find this term, click on index, it is going to be an alphabetical list in this index, and instead of using ATE, we actually used the acid for this concept, so we're searching for tauroursodeoxycholate acid, so this is the term I want to search. I can do this from here and will clear my search box and run this search so that's a little trick you can use to find your terms in PubMed.

Note we do have a large number of specific salts in MeSH, so follow the general rule of searching your specific terms in MeSH first. If the specific term is not included, try the closest more general concept.

Some of you may have searched PubMed with registry numbers. Registry numbers are assigned by the Chemical Abstract Service to specific drugs or chemicals. I am jumping to the MeSH database record for thalidomide. And we're going to the full MeSH database record for thalidomide so I can show you this. In the past registry numbers and enzyme commission numbers were added to the MeSH database for chemicals. When they are available, you can search by them in PubMed. The RN search tag is optional. However, not all substances have registry numbers or enzyme commission numbers. NLM still adds enzyme commission numbers but ceased adding registry numbers almost ten years ago. Check the MeSH database to see if a registry number for enzyme commission number is assigned.

Now I would like to talk about syntax issues in searching for substances in PubMed. We'll go back to PubMed again. Let's say I am looking for literature about compound 14. I run my search in pubmed, I get lots of results, but if I click on details, it didn't do what I wanted it to. It is tempting to use quotes for phrase searching in PubMed, but use them sparingly as this will turn off the automatic term mapping process and bypass a search of MeSH, so try your search without quotes first like I did in this example. If your term is broken up because PubMed doesn't find it as a phrase, only then use quotes to force a phrase search. If I look at details I will see it forced the phrase search in this case.

Also, there are times when all you have is a systematic chemical name, something like this nasty thing right here: 1-[(3,5-dichloro)-2,6-dihydroxy-4-methoxyphenyl]-1-hexanone. If you want to search with a systematic name, you need to follow some rules. If I go to details, you will see that it did some very strange things with my search. Here are the rules:

First of all, remove square brackets and open and closed parantheses, punctuation in chemical names, so I need to get rid of the brackets and parantheses. Square brackets are used to indicate a search field tag in PubMed as you know. Substitute a space unless there is already a hyphen. Parantheses are interpreted as a nesting operation in PubMed. For example, to nest or things. Therefore you should not include them in your chemical names. Again, substitute a space unless there is already a hyphen, in which case you can eliminate. Hyphens are interpreted as spaces. You can include them or not. I will leave the hyphens in this example: 1-3,5-dichloro-2,6-dihydroxy-4-methoxyphenyl-1-hexanone. That worked much better and actually mapped to a substance name.

Let's look at another example: 2,2-(2-chlorophenyl-4'-chlorophenyl)-1,1-dichloroethene. I am just using this details box so you can see more of the search. Here is another example. It does have parantheses so I know I need to get rid of those: 2,2-2-chlorophenyl-4'-chlorophenyl-1,1-dichloroethene. Always include commas and prime symbols which included as part of a substance name. These are indexed by PubMed. I will check my details. It mapped properly.

Another useful search technique is try searching the database using fragments of the chemical compound you're searching. Using this same example, if I go to the MeSH database, I can take this compound, and I have chlorophenyl and dichloroethene. I can remove the numbers and just include these fragments. This does a nice search of the MeSH database for compounds that include these fragments, and here is what I was searching for. Now that I identified my compound I can use the links menu on the right to run my search in PubMed. It avoids a number of syntax problems and allows you to identify the compound before searching PubMed.

Here is another example: 3-acetyloxy-4-trimethylazaniumylbutanoate. I will go to Details so you can see a little bit more of what I am putting in here. Here is another example of a substance. I don't have any syntax problems with this particular substance. There are no parantheses or brackets. I will try my search in PubMed, but no dice. I don't have anything that matches this. Then following my own advice, I will try the MeSH database. Trying just the fragments: acetyloxy trimethylazaniumylbutanoate. No luck.

Sometimes you will simply not have luck finding your chemical name in PubMed. I am sure we've all had this experience. Another option which you may find very useful is to use the NCBI PubChem Substance database which you can find from the database selection box at the top of your screen or on the NCBI All Databases menu.

There are three PubChem Databases, but we'll use PubChem Substance, as it includes from a number of sources and allows to you search by common name, systematic name, molecular weight and more. I am going to put in my original compound here (3-acetyloxy-4-trimethylazaniumylbutanoate), this is a large database of many chemicals. I want to be as specific as possible. I am running a search for this compound I could not find in the PubMed database, and I am running it in PubChem and I have matches. There are duplicates. It can be confusing at first. Look for a record with some detail and especially those with the literature link on the right-hand side as this means there are PubMed citations link to do this substance. Identify the common name and in this case we found our friend acetylcarnitine, and you can search for it in PubMed using the links menu off of literature.

I am going to go back to my PowerPoint presentation. And to summarize:

It is time for me to take questions. I know you've been entering questions all along. I am going to take a look now to see what questions you have. Okay.

I do have a question here which is a very good question. I don't know if I can personally address The question is: When or why does a substance name change to a MeSH term?

That is a difficult question for me as I am not in the MeSH department. I think what I am going to do is going to ask the MeSH section here at NLM if they can give us an answer to that.

Do all substance records include a registry number?

No. In fact, as of about ten years ago we stopped adding registry numbers to the MeSH records, so in fact a lot of these substances no longer contain a registry number, so you do need to check the MeSH data database to see whether a registry number is included for a specific substance.

Okay. There is a question about using hyphens.

A hypen is treated as a space in PubMed so you can include a hypen if you would like. Yes, you can use a space instead.

Okay. Here is a question, a very good question.

How do you search for melamine poisoning?

[Let me go back to PubMed and bring up PubMed on your screen and address this question. Here we go. Okay. A point which I may have too quickly gone by when I was talking about substance names. Okay. I am going to make sure that you can see this screen before I go on. I may need to reshare this pod, so let me try this again, pardon me. Okay. All right.]

Here we are in the MeSH database. I am going to bring up that melamine record again. Melamine is a substance name. It is not a preferred MeSH heading, so how would you apply what is a subheading concept to a substance name? Well, what happens when articles comes into an indexer and they want to describe the article using a substance, they will add the substance, and they will actually behind the scenes apply the subheading to the substance name, but what that does in practice is apply the subheading to this mapped heading, to triazine in this example. If you wanted to search by poisoning by melamine, what you would do is search by triazine poisoning and then add melamine to your search, so in PubMed I would do a search like this. What did I do? Hang on. Let me go from the MeSH database. I am working too -- okay. I am going to select the subheading on the mapped MeSH heading, and I am sure you're all laughing at me because I made some stupid error before, but that's all right. Then I will add melamine, okay, maybe there are simply no results. Oh, see what happens when you're running one of these clinics and you're trying to answer a question but you don't stop to think for a moment. All right. What's happening here is that we simply have chosen the wrong subheading because we need to do broaden the search a little bit to toxicity as well as poison to go include what we have for results, it is not that what I told you was incorrect, the concept is correct, that you would include the subheading on the mapped MeSH heading, but in this particular example we're not going to get results, we have some with toxicity, and there are very few results overall, so for that example we simply don't have a lot of results.

So let me spend some time thinking about that and maybe I can think of a better way to search for that idea, but technically that's your answer is you would apply the subheading to the mapped MeSH heading. Let's see.

Do we have other questions? Okay.

Here is a question in the aspirin example you used pharmacological action with an MH page but there are PA tags. How do you know which to use?

They do very different things. Let's see. Let's take a look at we're going to go to the MeSH database, and let's see, and we'll take a look at air pollutants as a pharmacological action. When you're searching with a PA tag or pharmacological action, what you're searching is this list of substance names and MeSH terms, this list of substances that have been assigned this pharmacological action so that's what the PA search tag does. When you search for a pharmacological action as a MeSH heading, then you're searching for the action as a subject, so if I go back to my results, then I am searching for records where the indexer believed that whatever substances were involved, an air pollutant was involved, so if I search -- what was my example? Let's see. Air pollutants as a MeSH heading, I would use this to combine with a specific substance term, for example, to find the literature where a specific substance was discussed as an air pollutant, so here is an example, example, RADON as an air pollutant. Excuse me, I am getting flustered here. I am to search PubMed. Here I am searching for RADON as an air pollutant. I don't want the whole list of air pollutants because I already know my substance.

Taking a -- we're almost out of time. I will see if I can find a real quick question. A quick question. Okay.

The MHDA tag.

I want to address that. This is when I searched for a date, and this is actually used when a record has been given MeSH terms. This is the date the mesh terms were applied. That's what that date is. In essence I am looking for the date that a record was indexed.

Okay. I am completely out of time.

For those of you still with us, take a moment and let me know how you feel now. If someone asks to you find the latest research on 1,3-bis(2-chloroethyl)-1-nitrosourea and glioblastomas, how confident would you feel about your approach? Have we improved at all?

While people are completing that poll and before I lose any more participants, we would greatly appreciate it if you would complete an evaluation of this clinic and we are especially interested in your ideas for future clinics. I am going to try to give you a link to an evaluation form. Hang on. I will -- so the clinic evaluation link, I am going to open for you in just a moment, and we would really appreciate it if you would complete that evaluation. I am just going to real quick broadcast the results of the poll. Seems like we might not have improved your confidence as much as I would have hoped [Note: about half of respondents answered "moderately confident], but thank you for completing that poll.

Here is a link to the evaluation: If you have additional questions, please don't hesitate to contact NLM's customer service. There is a link on every page, every National Library of Medicine web page has a link to the customer service, and I will look through the script of this clinic, and I know that I did not address everyone's questions, so I am going to try to compile those questions and our answers and make them available to you, and I probably will post them on the clinic's Web site, and along with those you will find a recording of this clinic so that you can review it at a later time.

Thank you very much for participating, and we look forward to seeing you virtually in future clinics.