Tous les COURS - French version: Info Santé Fiable - Recherche - Denis - Corpet
Everyone wants to know. Everyone would like to know "what is good", "what to eat", to go well, to be fit and handsome, to avoid diseases, and if an illness strikes, to heal better. How can one find the information?
The reflex is to search on the internet, or talk to your doctor. But... your doctor is not an expert on all questions. Internet has surely answers, but are they reliable? Sadly, online pages often show advertisements (to sell a product), or the opinion of an incomptent guy, or a "Guru"'s website. And there are so many pages, which one shall I choose?
My method is to trust only scientifically validated information, published in "good" journals, then indexed by the "U.S. National Library of Medicine," and posted on the website PubMed: It is free, fast and up to date.
But 3 problems still remain: They are 20 millions papers, the writing is "scientific", and most papers only diplay an abstract.
You will thus have to work and spend time: to search, to choose, to read, to understand: it's a true job. I would suggest you "A Method" ...
Here is the six-steps method
- 1- Key-words: Take time to write down on a paper the English key-words describing what you're looking for.
- 2- PubMed: search PubMed database, by testing several key-words combinations (*1)
- 3- News: Do NOT jump on "the" most recent publication (on top of PubMed search): there is no reason why "the last paper" would be better than the previous hundred! (see Cartoon on left)
- 4- Reviews: Start by looking at review papers (click "Review" on the left of PubMed page). Meta-analyses and "big journals" (e.g., NEJM, Lancet, Nature, Science, JNCI, JAMA) are usually more trustworthy.
- 5- Read it all: Select 3 review papers on their title and abstract, written by different authors (because each author has his foibles and prejudices). And reat the entire full paper(s). Abstract can mislead the reader with a conclusion that is not supported by the full article. You will thus need "full text papers": On PubMed page, some articles are labelled "free-full-text" (See link below "Review" on the left. Also written under each "free article"). What can you do if the paper you dream to read is not a "Free article" in PubMed? Also look for it in Google Scholar which indexes many Open Archives and Repositories (*2).
- 6- Finally: When this had been done, it is wise to write a simple summary for yourself. Then only you can focus on some recent articles. Why read them? Because some recent findings might be important, that has not been yet included into a synthetic review * (*3). Also because journalists (and therefore "people") only see what is new, so you must be aware of the news, but not unduly influenced (see cartoon above). For the same reason, once you've established your knowledge base, thanks to PubMed, you may watch some lay websites, to get an idea of lay people beliefs.
- It's long!: Yes, it takes time. I need between 3 hours and 2 days for a first round on a given question using this method. But this is the only serious possibility to find proven knowledge.
Any other approach leads to submit your judgment to someone else: This "other" could be "serious"... or not. It might be fine if you want to have a vague idea about a secondary issue. But it is neither honest nor intelligent if it is a vital point for yourself, or if you need to communicate widely on this topic (e.g., on TV, or by writing an article or a book).
Notes(*1) PubMed Search: It is efficient to "cross" key-words of different "nature" or from different "areas", such as a drug and a disease (cancer x aspirin), or a food and a body part (broccoli x colon), or three words from three areas (otherwise you'd get too many references). It is mandatory to test the spelling (broccoli or broccoli or broccoli, what yields most results?). Do try similar words or synonyms, but in separate searches (broccoli colon, intestine broccoli and cruciferous colon) When you write two words in PubMed search window, PubMed searches for articles containing both words (as if there was "AND"). You can also write a developed search using OR, AND, and parentheses.
(*2) Google Scholar: For instance have a look at the 18 proposals Google Scholar gives for an article that is not "free article" in PubMed : How good are rodent models of carcinogenesis in predicting efficacy in humans?. to arrive on this page, copy paper's title from PubMed into Google Scholar: first ref is the good one. Click the link "the 18 versions" below. At least four out fo 18 are full versions of the article (pdf/Free, pdf/ResearchGate, doc/OpenArchive HAL, and even more if you have access to ScienceDirect or else). Click link [PDF] on right, and get the paper. You may also search for the author's E-mail on the internet, and write him/her to ask the article be sent to you: it often works, since many scientists are kind people!
(*3) Latest paper? Conner Middelmann-Whitney told me "One reason I like to use recent articles as a starting point is that their reference section opens up further avenues to explore, including older papers. When I consult reference sections in older papers citing even older papers I think I might be missing something" Her website: Nutrelan
Conner M.W. also wrote me:
Something many lay readers find challenging is the statistical presentation of data (multivariate odds ratios, risk ratios, hazard ratios, confidence intervals, etc...). This is probably one reason why journalists may get it wrong or sensationalize stories: they misinterpret the data. Mme Conner Middelmann-Whitney - Her website: Nutrelan
=> Yes, CMW is right: I should write something on this point, but (1) several good statistical and epidemiological lessons could be found elsewhere, and (2) it ids impossible to teach this briefly in one page.
=> You may fins useful my simple (too simple) Cours d'épidémiologie nutritionnelle (Master M1 level, en français & in English)
3. Another thing that lay readers of science publications can find difficult to understand are the different types of scientific investigation such as anecdotal reports, case-control studies, case reports, case series, cohort studies, controlled trials, cross-over trials, meta-analyses, non-randomized studies, observational studies, prospective studies, randomized control trials, retrospective studies, reviews, systematic reviews, etc. A glossary of these, and possibly a ranking of their reliability, might be useful. CMW - Nutrelan
=> You may want to watch a good funny web presentation of those different kinds of studies on the AICR website: AICR: Studying Cancer
=> You also may have a look at my Lesson Epidemiological Methods (Master M1 level, French and English)
- A short BBC article "Misleading media reports about cancer" by Richard Evans, the WCRF's head of communications. His conclusion: The solution is simple. Scientists need to get better at explaining their research simply and journalists need to get better at including information about the limitations of the studies, even if it makes their articles less interesting. Conner's conclusion on this article: "Among others, it emphasizes the risks of attaching too much importance to isolated studies, and that these must always be seen in the context of the existing knowledge base!"