Short Link: http://ehr.bz/rand
[Nota bene! This post would load too slowly if I included all 200 embedded tweets here. You’ll find them at Rand Tweets.]
If you’re a business these days and you’re not monitoring what people on social media are saying about you and your product or service, you’re making a mistake. The same is true for health IT. I’ve been working in this area for decades. I’ve been on Twitter for four years (longer than more than 99 percent of users). I’ve never heard or seen such jaded, cynical, or even gleeful comments as I saw on Twitter in reaction to the recent New York Time’s article about the RAND report questioning whether EHRs have or can reduce costs. That said, there were insightful and constructive comments too.
Social media is a wonderful way to find and interact with people interested in the same things, and holding the same opinions, as you. This does have the effect of creating echo chambers. For example, most tweeps I follow or who follow me are much more similar to each other and to me than any collection of randomly chosen tweeps. I’m always surprised, when I conduct a systematic search on Twitter, how, well, “different” people are from me and my view of the world. Plus, I find folks who I wouldn’t think might be interested in what I’m interested, but do, in fact, have an opinion. (Eric Topol: OK. Edward Tufte: OK. Bette Midler? Yes. She has an opinion about EHRs.)
I used the following methods for this decidedly informal and un-automated sentiment analysis:
- I used a variety of search terms, both from the title and the article text.
- I looked at about 2000 tweets.
- Exclusion rules reduced that to the about 200 you see below.
- Excluded simple tweets (ie. no sentiment expressed) of article title (sometimes paraphrased, though not if paraphrase implied a positive or negative sentiment) and links.
- Excluded simple retweets (though, presumably, RT sentiment usually reflects sentiment of original tweet).
- Excluded tweets with offensive content
- Excluded tweets mentioning specific companies
- Resulting tweets are representative, not exhaustive. After the tenth “Not surprised” (most frequent sentiment) I stopped collecting.
- That said, the most frequently tweeted sentiments (such as “Not surprised”) do appear more than once.
- Search and collection occurred about 48 hours after publication. (I sampled about 24 hours later and the volume diminished substantially by then).
- Some exceptions to my exclusion rules crept through. (I merged half a dozen searches, resulting in a few duplicates. I deleted those obvious to me at the time, but did not invest much more effort than that.)
- I did not use microblogging sentiment analysis software. That’s why it an “informal sentiment analysis.”
Off the top of my head (told you this was informal) here are frequent or notable reactions:
- I’m not surprised. No one should be surprised. Why is anyone surprised?
- EHRs (specifically) are not at fault. It’s the [scape goat here: vendors, payers, gov] that’s the problem.
- Technology is not at fault. It’s the [scape goat here: culture, luddites] that’s the problem.
- What are the implications for my country’s similar effort?
- EHRs aren’t usable, need better design (with which I personally agree)
- See! I’m not the problem (from physicians)
- Too preliminary to judge. Wait until after meaningful use stage two or three.
- Interoperability/lack of standards/lack of patient engagement/etc. is the real problem.
I’m sure you can come up with another couple of categories.
Anyway, like I said up front, this is an informal sentiment analysis of reactions to the NYT’s report on the RAND study. I thought about generating some sort of bar graph, based on sentiment category or polarity (negative five to positive five?), but I’d just be formalizing my own implicit biases. Instead, browse the tweets and bring to bear your own pattern-matching and -explaining cognitive algorithms.
If you are in EHRs or health IT, I think it’s important for you to read these tweets, even if some of them make you wince or seem so so wrong or just plain weird. There is also some great insight and good suggestion. To the degree that Twitter is a real-time mirror of what society thinks, these tweets are what they (not the health IT tweeps you follow) think about you (generically, in this context).
In fact, I encourage you to start conversations with these tweeps. I have done so and benefited from it. If you’re already on Twitter: favorite, retweet, reply. If you’re not on Twitter, what a great excuse to join! Just head on over to with an email address, choose a Twitter name (you can change it later. I’ve done so twice), and agree, disagree, agree to disagree, and so forth. In fact, I created a Twitter list called and added the people whose tweets are below. Browse and follow whoever seems interesting. I’m sure they’d also be interested in what you have say about what they had to say. Think of it as outreach from health IT to the rest of society. Feel free to mention this blog post (http://ehr.bz/rand) and say sent you.
The reverse holds true too! If I mention you below (or you noticed when I added you to the ) feel free to follow those I follow or follow me. Of course you are free to follow anyone at anytime; you don’t need my permission or invitation. However, a high percentage of my fellow tweeps are (ahem) somewhat, peripherally, implicated involved in the healthcare information ecosystem described by the RAND researchers. It’s a target-rich environment, so to speak. You (a non-health IT member of society) and we (health IT folks) will benefit!
To end on an even more positive note, there are lots of new technologies set to “disrupt” lots of industries, including healthcare. These tech trends include mobile, cloud, workflow (my hobby horse), language (with connections to workflow), social, and data technologies. In fact, many of the tweeps I follow, or follow me, on Twitter are intent on doing this disrupting for the greater good. I look forward to your tweets and agreeing with you, disagreeing with you, and retweeting you.
Engage!
[Nota bene! This post would load too slowly if I included all 200 embedded tweets here. You’ll find them at Rand Tweets.]