Hints on using PubMed and MedLine

Helen Stanbro posted this set of hints on searching MedLine to a cancer newsgroup in early 1998 and was kind enough to allow me to add it to CancerGuide. Helen is one of the compilers of the MedLine database so these are tips from an expert’s expert. Don’t expect to get all of this the first time you read it – but play around with MedLine then come back and read it again and then play some more. Your skills will be the better for it – even if you already use MedLine.

If you want references to articles in the scientific/medical literature, the MEDLINE database which has over 9 million articles indexed by subject, is a great place to look and it is now available free of charge at PubMed. It can be tricky to find what you want among all those references, though, especially if you just textword search. Sometimes, for a disease with a rare word in the name like “leiomyosarcoma” you can get away with it, but for things like breast cancer, you have to think of every possible synonym (breast tumor, breast neoplasm, breast neoplasia, mammary cancer, mammary tumor, mammary carcinoma, etc., etc.,) or you miss things. The articles are indexed according to a huge controlled vocabulary called “MeSH” (“MEdical Subject Headings”) with almost 20,000 headings paired with about 70 possible subheadings. The drug and chemical names also are indexed to a controlled Substance Name list with about 200,000 names, so that all the trade names, chemical names, common names, etc. of each drug are linked together (otherwise you’d have to know them all and textword search all of them for each drug/chemical). To take advantage of the indexing, you must select “MeSH Terms” or “Substance Name” from the pull-down menus over the first box on the screen.

I have been an indexer for this database for 18 years, and I have probably processed 120,000 articles, so I am pretty used to the MeSH system. If you are having trouble with PubMed (and I myself find it much more cumbersome than the old ElHill system that everyone had to pay for), here are a few suggestions:

  1. Every cancer, except the leukemias and lymphomas, is listed both by organ site and the histological type, eg. brain neoplasm + glioblastoma multiforme, or lung neoplasm + carcinoma, squamous cell. The word “neoplasm” is almost always used, so don’t bother using “cancer” or even “tumor.” The common English name for the organ is also almost always used, so it’s “breast neoplasms” and “stomach neoplasms”, rather than “mammary neoplasms” or “gastric neoplasms.” If you don’t know the histological type, you can still find it by the organ site neoplasm term. Even though we always index to the most specific term possible, the system at the moment automatically includes all the things included under it in the MeSH hierarchy, so if you’re looking for cancer of the jejunum, you can search “jejunal neoplasms” but if you’re not sure where in the gut it is, you can search “intestinal neoplasms” or even “gastrointestinal neoplasms” and you’ll get all the jejunal ones as well as a lot of other stuff. If you want to be sure of getting everything even at the risk of getting useless postings you can skip, go to a very general term like “digestive system neoplasms” (which picks up stomach, gut, pancreas, liver, etc.). If you want to narrow the search down, try a more specific term if you can think of one.
  2. If you get some articles you like, the system has a nice feature whereby you can ask it for “related articles” and it will find more like the one you selected. If their idea of related articles misses the point you were interested in (which it often does!), and you want to see what MeSH Terms it was indexed under so you can use some of them in your next attempt, choose the “Medline” form of data display, rather than “Citation.” The system will print out all the MeSH Terms assigned to that article, so you will get some ideas for future searches.
  3. If there are still too many articles, you can try to use subheadings to narrow things down. To do this, you stay on the MeSH Terms option and find a MeSH Term that works, eg. “breast neoplasms.” Then you add the subheading of interest with a slash– eg. breast neoplasms/surgery or breast neoplasms/drug therapy. There are a lot of possible subheadings, but the most useful ones for diseases are probably: /therapy, /diagnosis, /surgery, /radiotherapy, /drug therapy, /diet therapy, /etiology, /mortality, /prevention & control, /psychology, /genetics, /pathology, /immunology, /metabolism, and /physiopathology. There is also a subheading “/secondary” for metastatic tumors which goes on the organ TO WHICH the cancer has spread and on the histological type, so for example a prostatic cancer with bone mets would go under “prostatic neoplasms/pathology AND bone neoplasms/secondary AND adenocarcinoma/secondary” or some specific bone like “spinal neoplasms/secondary.” Don’t bother looking under “metastasis” which although it is a MeSH Term, is seldom used since we prefer the specific site with /secondary. There is a term “lymphatic metastasis” which is often used, though, since we don’t have an organ neoplasm heading for lymph nodes.
  4. For chemotherapy protocols, we always use the term “antineoplastic agents, combined” plus all the drugs in the protocol with /administration & dosage as the subheading. We also have many protocols as special terms in the Substance Name File even though they are not really substances per se. Play around with it and see if you find your specific protocol. Try adding the word “protocol” or “regimen” to the name, if you don’t find the acronym by itself, eg. “VAC protocol” –we try to avoid 3-letter acronyms just because they often get used for different meanings in different disciplines, so we usually add another word to indicate if it’s a drug protocol, a gene name, a protein, or whatever.
  5. Any kind of combined treatment will have the term “multimodal treatment” added, so you don’t have to add together a lot of subheadings or agonize over whether it is “adjuvant chemotherapy” or just regular chemotherapy plus surgery, or whatever.
  6. There are lots and lots of useful terms that are not diseases, organs, or drugs like:
    treatment outcome, alternative medicine, palliative treatment, salvage therapy, remission induction, clinical trial, case report, randomized controlled trial, double blind method, patient acceptance, physician-patient relations, drug interactions, graft rejection, graft survival, immunotherapy, survival rate, survival analysis, disease free survival, follow-up studies, cohort studies, case-control studies, tumor markers, and many others I can’t think of right now. Guess, or check the Medline format for some articles to get ideas.
  7. Just about any kind of drug, chemical, protein, enzyme, antigen, Chinese herbal medicine, plant extract, or substance has its own entry in the Substance Name File. They may be listed in a slightly different way, eg. hyphenated or not hyphenated, from your first try, so don’t give up. I couldn’t find St. John’s wort the other day, though I knew it was in there, so I tried the botanical name “Hypericum” and had no luck, then I tried “hypericum extract” and hit pay dirt. I can’t remember what the problem with “St. John’s wort” was–I think they skipped the apostrophe, or something similar–but if you don’t hit it exactly right the first time, keep trying. Especially on transliterations of oriental herbal names, you may have to experiment! Useful subheadings for substances are: /therapeutic use, /adverse effects, /administration & dosage, /pharmacokinetics, /pharmacodynamics (which is the one we use for its effect on anything, eg. morphine’s effect on the immune system might be under “morphine/pharmacodynamics AND immunity, cellular/drug effects”), /chemistry, /metabolism, /genetics, /blood, /urine, and /immunology. All the terms in the Substance Name List are mapped to some MeSH Term (occasionally a plural form or a group term), so if you don’t want to have to search under your disease in the MeSH Terms box and add the drug in the Substance Name option on the Add Terms box, try just using MeSH Terms at first. Thousands of major drugs have their own MeSH Term–the Substance Name List is for things that appear only rarely or even only once in the literature. New drugs usually stay on the Substance Name list until they go through enough clinical trials to make it out into the world, and then they graduate to being a full MeSH Term. Often they just have a drug company number and maybe a long IUPAC chemical name while in the Substance Name file, then they get a trade name and perhaps make it into the MeSH Terms. So unless you know you are looking for SCH 32335 (I made it up, but this is typical for a Schering experimental drug name), you can probably guess something close in MeSH.
  8. Every drug is supposed to be indexed under a chemical individual or group name plus under a term for its pharmaceutical action, so all the anticancer drugs will also be under “antineoplastic agents” (or a specific, like “antineoplastic agents, hormonal” or “antibiotics, antineoplastic”–all of which will be picked up too if you use the general term “antineoplastic agents” as a MeSH Term). If you can’t find the drug you want under guesses at its chemical name, you may pick it up searching on its action. Some of the other useful action headings are: immunomodulators, biological response modifiers, anticarcinogens, carcinogens, mutagens, radiation sensitizing agents, and cancer vaccines.
  9. To link or exclude things, use AND or NOT, respectively, between the MeSH Terms. To further limit your search, you can ask for only specific languages, eg. “AND english” or “NOT chinese”. Don’t automatically exclude foreign articles unless you really need to cut down, as many of them have useful English abstracts. All the titles are translated, so you can at least get an idea of what the article is about.
  10. If you can’t find the full-text articles near you, request them from the National Library of Medicine through your local public library.

I hope this short course in MEDLINE searching is helpful. If you get desperate, you can e-mail me for specific suggestions, but obviously if I get a zillion requests I won’t be able to answer them all (I have to be out there indexing the articles for you to find!). Happy hunting!–Helen Stanbro

This CancerGuide Page By Helen Stanbro. © Helen Stanbro
Last Updated: Feb 24, 1998