Moving forward with clinical decision support

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
 - Jonathan M. Teich, MD, PhD
Jonathan M. Teich, MD, PhD

BOSTON—Jonathan Teich, MD, PhD, likes to think of clinical decision support as “making dumb docs smarter at the right time.”

While Teich, CMIO for Elsevier and an emergency room physician at Boston’s Brigham and Women’s Hospital, made this comment with tongue-in-cheek at a session on CDS at the 2014 AMDIS Fall Symposium, the point he was making is that physicians can’t know everything or be expected to process countless pieces of information at the drop of a hat.

“CDS is giving you the right information at the right time,” Teich said. “If I have the right information tools in front of me when someone asks me about sparkplugs I can tell you about feeler gauges and getting the right spacing, but If I don’t have the right information tool I may not even know what a sparkplug is.”

So, CDS is about providing information, and good decision support should work the same way a global positioning system (GPS) in a car works, Teich said. The technology that goes into making GPS work is sophisticated and intricate, but the end result of all that technology is the communication of a simple piece of information—“turn right” or “turn left”—relevant to the time it’s needed.

“Something like this is implementable, easy to use, valuable and cost effective,” Teich said. “That’s why GPS is everywhere, and CDS needs to get to that point.”

Which isn’t to say that we haven’t seen evidence of good CDS. As Teich pointed out, there have been many studies showing the value of CDS, “so we know that if CDS is well designed and well implemented it does wonders.”

At the same time CDS goes through stages in which it seems to lose popularity and acceptance, Teich said. “Clearly some people are getting it right and some aren’t getting it right.”

There are various things that can be done that make the difference between well-functioning and unsuccessful CDS, Teich said, and one of the questions health IT leaders need to ask is how this information is disseminated and shared. Many healthcare organizations are doing some great things with CDS, he pointed out, but “there are 6,000 organizations out there—and those are just the hospitals—and each one of them has to do this over and over and over again. There’s no way to really share information from one place to another.”

Teich asked how CDS can be made more effective at the particular method level, how new content can be added that can be used as the basis for CDS, and how all that can be made applicable across the country.

For example, adding new content to a CDS may seem simple, but as Teich pointed out “you can’t use every single paper, so how do we get the stuff that’s actually actionable?”

Much of his work at Elsevier involves knowledge and filtered reference, Teich said, adding that Elsevier puts out about seven million pages of distinct information every year. “And I’m an emergency room doc and I may have a patient in front of me who has Kartagener Syndrome,’ said Teich. “And I’ll say ‘what the heck is that?’ I know that somewhere in that seven million pages is the right information but how do I get the paragraph I need?”

Reference tools like DynaMed and First Consult are available, “but even those have turned into little encyclopedias, and I find my residents kind of staring at them for five or ten minutes until they get to what they were actually looking for,” Teich said.

Some of what Teich is working on now asks whether all of this medical information can be taken and encoded. For example, if Teich can take information nuggets, like “chest X-ray is the best diagnostic test for pneumonia, but not for a pregnant patient” and put it into a knowledge base, “the next time something changes I don’t have to change the entire reference, I just have to change that particular nugget and then I can make them available as services.”

“Instead of having to write these fixed books or fixed references, I can actually have something that takes from that particular knowledge base,” he added. “So if the question is about pneumonia we will bring up the latest [information] and cobble it together into the right presentation. And this allows me to have a whole bunch of different presentations that are useful at different times.”

How can best CDS practices and methods be shared and disseminated? Teich noted that when a new drug is developed and approved, every hospital in the country has immediate access to the drug and can drop it into its Pyxis machine. On the other hand, Teich said, if he comes up with a new CDS idea or an idea about how to build a CDS intervention or new algorithm or new rule, every hospital will have its own painstaking process of integrating it into their EHR.

“So there should be something like a CDS Pyxis machine,” Teich said. “Let’s say I want to prevent VTEs (venous thromboembolism). We can go over there, look up VTEs, select what someone else has done, download it, integrate it with my EHR, be able to measure its effects for everyone else to use . . . “

Overall, the outlook for the future of CDS is promising, Teich said. “If you look at standardized methods, standardized technologies, new ways of sending things back and forth, and new approaches for selecting the best thing for the right situation, we should be a in a position where we can actually have more hospitals, more organizations, more practices and more patients be able to get the right information they need at the right time.”