How to keep up with medical literature

02/21/2013 14:30:21
tags: medicine

Keeping up with the latest medical research is an increasingly challenging enterprise. For a variety of reasons, the sheer volume of new articles being published every day is growing exponentially. Much of it is, quite frankly, crap — either of use to only a very small group of people, or a rephrasing of ideas that have already been published.

As a medical student, then a resident, and now an academic faculty member, I have tried different strategies over the last 15 years. Most of them didn’t work very well, but fortunately the internet era has brought some new tools that make this task easier.

Define Your Interests

First, you have to decide what areas of interest you want to keep up with. It’s impossible to read everything, so you have to pick and choose. You have two basic strategies: deep or wide.

The deep strategy is to define a very narrow subject matter, and read everything that comes out related to that subject. For example, you might be interested in the diagnosis of systolic heart failure by physical exam. This is a very focused topic — one could reasonably expect to be able to read every journal article published related to this.

The wide strategy is to pick a very general topic and read representative articles as they arrive. For example, as a hospitalist I try to stay on top of things published that related to hospital medicine. I can’t read everything about hospital medicine and internal medicine, but I can try to catch the highlights.

The Deep Strategy

The best way that I have found to keep up with a specific topic is to create a saved search that is run periodically to identify newly published articles that match your criteria. I use the saved search feature at PubMed to receive weekly updates on topics of interest. Once you configure a search query, you don’t have to think about it again. If a new article is published meeting your criteria, you get notified. If there’s nothing new, no email. I have several very specific queries saved that only trigger a couple of times a year, but they notify me about articles I would be very unlikely to find any other way.

An Example

JAMA has a series called The Rational Clinical Examination. It’s a series of articles over the last 15 years that focuses on diagnostic evaluation during the history and physical. This series was one of the first ways by which I was exposed to the idea of an evidence-based physical examination. Each article examines the accuracy and utility of physical examination maneuvers, questions to be asked during the history, or useful laboratory tests when trying to make a specific diagnosis.

I use this as an example for several reasons:

  • It’s a somewhat broad topic, but with a narrow set of matching articles — this means it will require some thought to create an accurate search.
  • The link above points to an actual definitive list of articles in this series. This can help us assess the accuracy of our query.

So, in order to match these articles, we scan the list of title and notice a few things:

  • Every article is published in the journal JAMA, of course. This can narrow our search quite easily.
  • Many (but not all) of the titles include the phrase “this patient”. Others use “this woman” or “this adult”.
  • All articles were published after 1997.

So let’s start building a search for PubMed, refining it as we go. (I assume some knowledge about how to use PubMed, how to use Boolean searches, etc.)

"JAMA"[Journal]

This gets us over 60,000 hits. Way too many.

"JAMA"[Journal] AND ("1997/01/01"[PDAT] : "3000/12/31"[PDAT])

Now we’re down to 19,000. Still too many. (BTW — I figured out the date limits syntax by experimenting with the filters and seeing what showed up in the “Search details” box on the right. This technique is your friend.)

"JAMA"[Journal] AND ("1997/01/01"[PDAT] : "3000/12/31"[PDAT]) AND "patient"[Title]

Now we’re down to 1102 (BTW - a perfect search should result in 64 matches at the time I am writing this.) Still too many, but we’re getting there.

"JAMA"[Journal] AND ("1997/01/01"[PDAT] : "3000/12/31"[PDAT]) AND "patient"[Title] AND (systematic[sb] OR Review[ptyp])

Since these articles are usually either review articles are systematic reviews, this should narrow the focus. This shows 84 matches. Scanning the first page of results shows that most of them are matches, but some are not.

"JAMA"[Journal] AND ("1997/01/01"[PDAT] : "3000/12/31"[PDAT]) AND ("patient"[Title] OR "patient's"[Title] OR "adult"[Title] OR "woman"[Title] OR "man"[Title] OR "I"[Title]) AND (systematic[sb] OR Review[ptyp])

We were missing articles that didn’t have “patient” in the title. This grabs a few more. Now we’re at 114 matches. A few more false positives, but fewer misses. And still a reasonable number of hits for 15 years of articles.

"JAMA"[Journal] AND ("1997/01/01"[PDAT] : "3000/12/31"[PDAT]) AND ("patient"[Title] OR "patient's"[Title] OR "adult"[Title] OR "woman"[Title] OR "man"[Title] OR "I"[Title]) AND (systematic[sb] OR Review[ptyp] OR "Physical Examination"[MeSH Terms])

In order to catch a few missed articles that weren’t classified as review articles, we give the option to find those using “Physical Examination” as a MeSH term.

When I compared the results of this search to the definitive list of articles, I found that it missed around 5 out of the 64. I didn’t see anything obvious in the titles of those articles that was useful when creating a search. We’re still catching over 90% of the articles we wanted, in a search that has averaged just over 10 articles per year for the last 15 years. Hardly an overwhelming pool to sort through.

By using the “Save search” button on PubMed, you can have their servers run this search for you on a recurring basis and notify you of any new hits. So from now on, you should be alerted to any new articles published in this series in JAMA (unless they begin to change their naming habits for titles….)

I’ve created similar searches for specific topics, general topics, and even certain authors. Every Saturday morning I get 4–6 emails with updates, and can usually delete most of them just by scanning the titles. The articles that look like they are potentially interesting merit a second look by pulling them up on PubMed and scanning the abstract.

The Wide Strategy

As an academic hospitalist, I need to keep up with a wide variety of areas — cardiology, gastroenterology, pulmonology, etc. Clearly I can’t read everything, so I need a way to make sure I see the “highlights.”

What I have found that works best for me is to subscribe to a few services that filter new articles and rate them for notification.

Some of the services I have used include:

Most work in similar ways — you specify areas of interest, and set your thresholds for notification. Some services are free, and some charge a fee. Some offer free notification, but you pay if you want access to their summaries of the articles. I personally find that scanning the abstract/article is probably as helpful as reading their review. I need tools to help me find articles, not to read them for me.

By using several of these services and tuning the settings, I tend to get 5–10 emails per week with a list of potentially interesting articles. There are some downsides:

  • I tend to see a lot of duplicates. That’s ok, just means I have to skip some articles because I’ve already been alerted
  • Not all of the articles are interesting to me. Again, that’s ok. The delete key is free to use.
  • It’s a lot of email. The key here is to skim it when you get it. If there isn’t something that really grabs your attention delete it. Don’t save it to come back to later, unless there is an article that you really want to read. Sometimes just reading the headline is enough to plant the seed. If you need the article later, you can come back for it.

Summary

This approach has helped me to do a better job of staying afloat in the deluge of articles that are published every week. By being ruthless about hitting delete when necessary, and skimming headlines to at least know what’s new, I’ve been able to do a fairly good job of staying up on the latest news generally, and being alerted to new articles in my own areas of interest.

Similar Pages

Comments

I use this too, except with Pubmed’s RSS instead of email. I was surprised to see they offered it but it really works quite well to have an RSS feed for each search you want to keep up with.

Leave a comment