Stuff that occurs to me

All of my 'how to' posts are tagged here. The most popular posts are about blocking and private accounts on Twitter, also the science communication jobs list. None of the science or medical information I might post to this blog should be taken as medical advice (I'm not medically trained).

Think of this blog as a sort of nursery for my half-baked ideas hence 'stuff that occurs to me'.

Contact: @JoBrodie Email: jo DOT brodie AT gmail DOT com

Science in London: The 2018/19 scientific society talks in London blog post

Showing posts with label PubMed. Show all posts
Showing posts with label PubMed. Show all posts

Saturday, 2 January 2016

PubMed is not for beginners... well, OK, maybe. Making sense of medical research.

Recently I've seen this image (from BasicBiologyBabe on Facebook) crop up a fair bit on 'skeptic Twitter' (of which I'm a member). We are the people who argue with those who claim that [X] can diagnose or cure [Y] despite no good evidence that it can. We also report misleading advertising to Advertising Standards Authority or Trading Standards etc and sometimes get claims removed and dodgy websites closed. As skeptic blogger Guy Chapman once put it we're at the intersection between science advocacy and consumer protection.

While I do try to be polite to people whose opinion differs from mine I find it pretty difficult not to be terse with people who are putting others' health at risk with their bad advice!


Transcript: "PUBMED IS NOT FOR BEGINNERS. Browsing abstracts on PubMed does not take the place of a university education in a scientific field. Experience and knowledge of context are essential to comprehension as well as reading the WHOLE GODDAMNED PAPER. Stop being dicks. - Scientists Everywhere [fb.com/BasicBiologyBabe]"
Clearly there's a big element of truth to this 'PubMed is not for beginners' idea,  but I'm not sure that this sort of advice will convince the people who might need to be convinced.

People who are promoting dangerous or ridiculous treatments or diagnostic tests (think antivaxxers, homeopathy shills, live blood analysis, tuning fork therapy etc) occasionally cherry-pick a PubMed abstract and promote the idea that it supports their favoured treatment or test. I've collected a few examples in this Storify, some are pretty funny but it's sad to think people are being so badly misled. They seem a lost cause to be honest ;)

But what about non-specialist people who have a health condition, or care for someone who does, who want to find out more? 

I'm hoping to go to an event on 'Communicating risk in health information' (London, Tue 26 Jan) which is for the subgroup of science communicator types who communicate health information. This is partly about helping non-specialist people to make sense of statistical information by helping those communicating it to express it more clearly.

A straightforward translation of 'science' to 'plain English' for an abstract (even for a whole article) will never be sufficient because an abstract or article cannot be read in isolation. Readers need to consider how it fits in with what's already known on a topic and also to know what conclusions can reasonably be drawn from a particular type of study design (how many people were involved, were they randomised etc) and just knowing lots about 'biology', for example, isn't enough. In fact it might give people a false impression of knowing / understanding.

A 2011/12 project, 'Patients Participate!', looked at ways in which patients could be more involved with health information providers (for example, medical research charities) and found that patients and the public wanted the following, which are all very reasonable -
  • To find out about the latest medical research and how it relates to their health
  • Accessible, plain English, searchable summaries of research articles that clearly communicate the implications and limitations of the research findings
  • Information they can trust (credible, reliable and unbiased)
  • To engage with researchers and research funders
  • To be involved
  • To learn more about research because theyare interested and they want to be better informed about the latest developments
  • To help researchers be better communicators
It's hard enough to provide a dedicated abstract translation service, let alone one that takes all this into account and I don't think we've comprehensively solved that problem sustainably, though different health charities do this in their own ways.

Here's a comment I left under the picture on Facebook in response to people asking 'well OK, but where can we find this info then?' -
"Where possible I'd suggest, for medical articles, asking reputable health charities for their take on a study. Many of them have dedicated science communicators (of varying job titles) whose job it is to (a) explain the content of new studies AND (b) put it in context with other information known about the topic, as well as highlighting what conclusions can reasonably drawn from a particular study design etc.

There are certainly many efforts made to 'translate' scientific abstracts into plain English but by itself this might not be very useful, as there is usually a requirement for some background information needed, in order to know how much weight to give to some new finding."
 To which I'd add this patient-facing resource (there are other suggestions in this PDF (also linked above))

NHS Choices' Behind the Headlines - stories covered in the press might be spot on but often the end result can be that the study is a bit overhyped. The text of the newspaper article can be very good and clear, but let down by a dramatic headline. While we might all hope that everyone ignores the headline and reads the entire article I think we'd agree that's unlikely. Behind the Headlines considers the context of the story and explains the science for a non-specialist audience. It's fab.

Tuesday, 19 February 2013

How to grab the search results URL from NCBI PubMed

For those familiar with PubMed
It's the 'See more...' link below the rectangular box on the right, once you've done a search - click that and then select the 'URL' button that appears below the box on the new page.

New readers start here
When you run a very basic search on PubMed (eg literally just typing aspartame insulin) the URL of the page changes from

http://www.ncbi.nlm.nih.gov/pubmed to 
http://www.ncbi.nlm.nih.gov/pubmed/?term=aspartame+insulin

If you then add a filter (look on the left hand side for the 'species' filter) so that you're only looking at results of studies done in humans then the URL returns to

http://www.ncbi.nlm.nih.gov/pubmed

which is a bit disheartening if you want to email it to someone and say 'hey look what I found'.

There is a way round that but I think that, in most cases (within reason), the URL of the page should reflect what the page is doing (except where you've had to log in of course). Mind you if I'd been in charge of this from the start we'd have never had frames.

First catch your URL 
Have a look at the right hand side of the page (you might have to scroll down a bit) for the 'Search details' box where you can see how PubMed has converted your typed in search words into a properly built search. Below the box (on the right) is a 'See more...' link - click it.

You're now on the 'Search details' page, there's another box and below it are two buttons one of which says 'URL' - click that and you'll be taken to a page of search results. At the top of the page, where the web address / URL is (ie the address bar) is a plain text version of the URL which has spaces in it and doesn't look that URL-y. However once you click on it and copy it, and paste it anywhere else it should automatically fill in the spaces and even if it doesn't, when pasted into a browser it will work as a web address and take you to a copy of those search results.

http://www.ncbi.nlm.nih.gov/pubmed?term=%28%28%22aspartame%22[MeSH%20Terms]%20OR%20%22aspartame%22[All%20Fields]%29%20AND%20%28%22insulin%22[MeSH%20Terms]%20OR%20%22insulin%22[All%20Fields]%29%29%20AND%20%22humans%22[MeSH%20Terms]&cmd=DetailsSearch




Sunday, 30 December 2012

NCBI PubMed and share buttons - easy way to share abstracts?

Update 15 July 2014
This has (sort of) been implemented. You can now share directly from NBCI PubMed to Google+, Facebook and Twitter with one click (if you're also logged in to the services you want to share to) although there's no information given about how many times a particular abstract's been shared. Useful nonetheless and seems to be quite popular.



------- Original post below -------- 

The other day, while reading the answers to a question on Quora: "What is the funniest research paper you have ever read?" I checked on PubMed to see what the URL was for my all-time favourite "Development of a technique for the in vivo assessment of flatulence in dogs" so that I could include it in an answer myself.

While on PubMed I wondered if folk at NCBI knew that I've shared that paper a few times, on Twitter and my blog, in a collection of abstracts or titles that make me giggle, (and now on Quora). I also wondered if the authors knew that their paper had been shared in this way... and that made me wonder about share buttons on PubMed, so that everyone knows how often an abstract's been shared.
Just for fun really as it's not a particularly stable metric to start weighing one abstract against another (see 'More on share buttons and what they can't do' below). Authors whose names are also the names of rude words will be highly 'cited' as I'm sure will any paper making reference to dogs farting, as my favourite one does. I've also shared abstracts that talk about rats flying in space and kittens being startled by a sort of jack-in-the-box device. So the strangeness or amusingness of the paper is definitely a factor.

After a brief exchange on Twitter with Graham Steel (@McDawg) in response to my tweet above he's investigating how useful it might be, with chums at NCBI, and asking other people what they think - there's a poll which he's shared with people via Twitter and email.

I can't really think of any cons other than jealousy if your paper gets shared more than my paper, so to speak, although there may be the risk that people misuse the numbers to imply that their paper's been shared and used more than the 'data' can really say - I think the benefit is mostly for sharers by making it easy to find the right bloomin' link.

Correct URL: http://www.ncbi.nlm.nih.gov/pubmed/11453473
Incorrect URL: http://www.ncbi.nlm.nih.gov/pubmed?term=assessment%20of%20flatulence%20dogs

More on share buttons and what they can't do
Article authors will also have copies of their abstracts or even whole papers on their own university site or an academic listing so people might share that link rather than the PubMed one. Also PubMed deals only in biomedical stuff.

But I thought it might be fun for anyone to see that something's been shared. I've noticed that the Advertising Standards Authority has very recently begun doing exactly this so that any adjudication shared from the share buttons on the page will show up. It doesn't show up though if you just copy and paste the URL (so if you want to share a page 'covertly' then that's the way to do it).

I provide a similarly pointless-but-kinda-fun metric for my pals on the CHI+MED project. Because all of our published papers (really pre-prints) are available on our document management system (Knowledge Tree) and it lets you know how often a paper's been downloaded I can provide this info for any paper.

It's fraught with irrelevancy and wrong-ness though because (a) papers that are published early are around longer and have had more opportunity to be downloaded (b) later-published papers are more likely to be promoted on Twitter or elsewhere as we started the Twitter feed a while after the project began. Exactly as above, authors have copies of papers on their own academic pages and they're also findable in university repositories and some journals make copies available too.

It also tells you nothing about the citations, or even if people read them... and if they did read them if they thought they were useful!