ONE WEIRD TRICK THAT OVID WON’T TELL YOU

761bf8f77c17cc26a07f837501f75850913c192227b19aabaec2a3910e5c6f99No, it’s not a food that will give you a slimmer stomach or boost your manly prowess.
I’m talking, of course, about the ability to find the total number of citations in the Medline database. Why on earth would someone want to know how many citations are in a database, you ask?
  • To compare and contrast the size with other databases
  • For FUN, because you’re a nerd like me
  • Um… because?
It’s relatively straightforward to find the total number of citations in PubMed. Their documentation helpfully tells us: “To search for the total number of PubMed citations, enter all [sb] in the search box.”
However, a few days ago I was struck with an awkward problem. I needed to find the total number of citations in Ovid Medline. Why? I had conducted a straightforward scoping search for a researcher and created a basic frequency analysis of the number of citations retrieved in the search per year to show the publishing trends in the topic over time.
blog1
frequency analysis, non-normalised (raw count of citations)
The researcher asked me to normalise the data…. say what?? Do I look like a statistician?
I knew I couldn’t use the numbers from PubMed, because the two have slight differences in content. And I couldn’t translate my strategy into PubMed because it relied heavily on the adjacent operator (which is absent in PubMed).
After some frantic searching, I found out that this was not such a difficult task: all I needed to do was take the number of citations retrieved from the search in a given year, and divide this by the number of total citations published in the database that year. This would even out any potential errors in the chart from anomalies in the database as a whole.
The problem: I could not find an equivalent operation in Ovid Medline to PubMed’s all[sb] command. After combing through Ovid’s documentation, I finally broke down and tweeted them… and received a response within a few hours.
I know everyone’s been waiting with bated breath to find the answer: it’s docz.dz.
What does the .dz field code stand for? No idea. But anyway, it seems to get the trick done, and now I have my nicely normalised graph. In the second image, below, you can see that the downtick in citations for the year 2016 has vanished, because the number of citations retrieved from the search is proportionate to the total citations published this year.

 

blog2
frequency analysis, normalised (results as a percentage of total citations in database)
Happy story! The end.
PS Cheers to Ovid’s social media team! They are totally on the ball.