We’re in a transitional phase here at PubTrawlr as we move to get our 101 Days of Science and monthly newsletters up and running. Part of that will include much more dynamic and interactive formatting.
So, in the meanwhile, I decided to do something cheekily meta this afternoon: a review of journals focused on reviews! For the uninitiated, a systematic review attempts to gather all of the article focuses on a specific research question, and then evaluate the quality of the evidence in its totality. Needless to say, these are great when starting our research on a topic.
For this specific article, though, I pulled all the articles from the last 30ish days from four specific journals:
- The Cochrane Database of Systematic Reviews
- Systematic Reviews
- JBI Database of Systematic Reviews and Implementation Reports
- World Journal of Meta-Analysis
You can click on the title and go straight to the journal’s webpage. There are a lot more that I could conceivably have pulled by specifically searching for recently published articles with “systematic review” in the title. I decided to just stick with these four for the time being because 1) the search still yielded 121 articles, and b) I wanted to save something for the bigger 101 Days Report.
What I found!
I used a network plot to look at associations between words in each article’s abstract. A bigger circle means the word showed up more frequently. A thicker line means the relationship between two words (circles) showed up more frequently.
The first thing we notice below is that the abstracts are really methods heavy. Most of the terms are about how the study was designed (e.g., RCT, randomized control trial) and analyzed (e.g. confidence interval). There’s not anything in the network plot that suggests topic clusters with the exception of the topic of the decade, COVID 19.
The topic clustering was more successful at pulling out the themes. Below, I plotted frequency of topic on the x-axis, and the key distinguishing terms on the y-axis. We can see specific topics around birthweight and other infant-specific studies. Also, a bunch of stuff published on COVID.
What interesting about the most frequently occurring topic is that it seems to be broadly healthcare-related, even though the term healthcare doesn’t appear in the most distinguishing terms. This is likely due to the fact that most of these 121 articles are health-focused, and that term might have gotten washed out due to the term-frequency value.
I added a pdf of a table that lists the most representative articles for each topic. Click on the link below to download and check out the citations for yourself.