Assisting Providers at the Point of Care: A Conversation with a Clinical Decision Support Pioneer

Aug. 23, 2016
Jonathan Teich, M.D., is on the leading edge of health IT, with great passion for clinical decision support systems. He recently spoke with Healthcare Informatics about CDS trends and challenges.

Jonathan Teich, M.D., is a wearer of multiple hats when it comes to healthcare and health IT. Currently, Dr. Teich is a practicing emergency physician at Brigham & Women’s Hospital in Boston and Assistant Professor of Medicine at Harvard. But his roles in medicine don’t stop there. He’s also the chief medical informatics officer (CMIO) at Elsevier, one of the world’s biggest producers of scientific and medical information, where he works with the clinical solutions group to develop products that use sophisticated search technology and "smart content" to deliver evidence-based, actionable information clinicians can use to make decisions at the point of care.

Indeed, Teich is perhaps best known in healthcare circles for his passion around clinical decision support (CDS)—the process of delivering situation-specific, actionable health information to clinicians and patients. Most often, CDS applications—such as computerized alerts and reminders to care providers and patients, clinical guidelines, and condition-specific order sets to name a few—operate as components of comprehensive electronic health record (EHR) systems, although standalone CDS systems are also used.

To this end, Teich was on the team of authors who penned the book, Improving Outcomes with Clinical Decision Support: An Implementer's Guide, which first came out in 2005, and has since been updated periodically. In 2012, the authors published the second edition of the book, which received the Book of the Year Award from the Healthcare Information and Management Systems Society (HIMSS).

Teich believes that CDS, in various forms, “can prevent large amounts of medical errors, prevent adverse events, and can proactively optimize care.” Due to his diverse skillset in the medical field and his wealth of knowledge in health IT from spending more than 20 years in medical informatics, Teich is on the leading edge of the sector. Recently, he spoke with Healthcare Informatics about how providers can better maximize their technology investments, how CDS applications are working to help physicians right now, and what’s needed to improve. Below are excerpts of that interview.

With the huge investment the industry has put into healthcare technology, the attention is increasingly turning to maximizing the value of these systems. Is that a place where most patient care organizations sit today?

I would certainly hope so, and I do think so. If you look at the U.S., we are at the stage in which pretty much everyone has an EHR—all the acute facilities and a large portion of ambulatory facilities—and what we are moving towards now is simply from implementation to implementation for value. Everyone is driven by payment reform and payment changes which had traditionally been based on documenting more things, but now are moving more into quality metrics and also cost of care metrics. Those things are driving payment, which is driving what people are spending money on.

Jonathan Teich, M.D.

What’s currently working, and what’s needed to maximize these investments?

There are several different areas where we need to spend more of our investment dollars to address care quality. We should be using computers’ powers to help analyze and remember everything to help us. We found research from myself and others from 20 years ago, and have reinforced since then, that clinical decision support, in various forms, can prevent large amounts of medical errors, adverse events, and can proactively optimize care. Broadly speaking, we are looking to trying to provide CDS in one form or another. And there are three or four areas that are the peak of that today.

Can you speak more about these areas?

The changes in care have to do with older patients, as there is much more chronic disease. We talk about population health and precision medicine; these are names for understanding what a patient is, what subset they fall into, knowing what the best care is for that subset, and delivering it. The areas I’m referring to are: clinical pathways, order sets and care plans, quick access to answers, analytics, and patient engagement.

Regarding clinical pathways, so much of what we’re doing is long term. We have a hard time following up on the status and needs of a 67-year-old person who has COPD [chronic obstructive pulmonary disease] and heart failure and is doing well, but needs to be kept on the straight and narrow by being monitored and having his or her treatment changed. Some people do this now with labor-intensive care management programs, but we can use computers to filter out the folks who need our attention most urgently. So if I have 150 patients with congestive heart failure, I can find ways to figure out the top six or seven who need my attention now, who are falling off a path and need a change.

A typical clinical pathway algorithm or program will help you set out a goal and an ideal path. It can provide it for you, guide you to tweak it and help you follow it. So it helps you say that if this patient’s breathing and activity is doing well, then the following is true. But if he or she showed up in the ER or had trouble breathing, then [the pathway] could help you change that too. A follower of chronic care is much better than what we’re doing now, and much better for quality and cost.

The pathway is your guiding rails, but what do I have to do when certain things happen? I need to communicate something to the patient, perhaps setting up a program where every day the patient tells me how he or she is doing in terms of breathing or weight, and I make a change in his or her prescriptions (order sets). If I think about the pathway model, patient engagement speaks to being able to understand patients’ needs and preferences, understand how they’re doing, and give them techniques to use at home and with the provider to optimize things. Maybe I ask a patient to keep a diary of his or her weight, or I do it automatically. Maybe I have a line of communication that goes to the patient every week. There is lots of education to do to keep the patient’s and doctor’s plan in line.

Some skeptics will say that clinical decision support tools haven’t realized the promise of applying knowledge in a manner that significantly improves patient care. What are your thoughts about this?

I think in some ways it’s a baby at risk of being thrown out with bathwater. If you look at the dissatisfaction that people are having with EHRs these days, a lot of that is about busy work, extra documentation, and extra process work that people have to do that they don’t feel they need to do. A lot of that is about documentation requirements and all of that stuff that we have to do for every encounter on every patient to keep track of things. I don’t think that people are opposed to CDS as long as it’s not overdone. CDS is quite well accepted, but in a world where people are annoyed by additional busy work, one has to go out of one’s way that CDS is user friendly and not intrusive.

How are CDS tools better fitting into clinical workflows these days?

People don’t want to be taken off their path or stopped when it comes to workflow. It’s okay if I get more information upfront, even more than I needed, but I don’t want to be going along with my care and then being stopped with different alerts four or five times. So we focus more on proactive means of CDS. Take something like order sets, in which you have to write orders anyway if you’re going to do prescriptions and tests. If I can give you information upfront that says here are the best orders for this kind of patient/circumstance, before I have to think about it being a change, that’s hitting my mind at a much more receptive point, and if I can do it in a way that speeds you up, people do accept that. We think that proactive pieces of CDS like those order sets, pathways, and some care plan items, are much more acceptable compared to a stream of alerts all over the place.

Can providers easily pull out what they need from these applications, or are they still going through a wealth of data?

It’s not as easy as it once was since there is so much information. There are answers to what you need out there. I know that as an ER doctor, if I see a patient that has something I’m not too familiar with, there’s an answer to it. I saw a patient recently with a case of arsenic poisoning, which I haven’t seen before, but I knew that there were answers somewhere. The question becomes, how can I get to that answer quickly?

One of the things we are spending time on now is how to optimize the search when I ask for the information, but also how to presume what I’m about to search and give me those things in the most concise, “chunkable” form that you know. Everyone likes those bits at the top of your Google page that shows you the time of your flight or the score of the game. That’s what we are seeking—to be able to give you optimized answers. We know from research that there are about 80 different questions that clinicians ask in any given circumstance. We can start to understand when you want to know those things before you do. It’s almost like a waiter who refills your drink before you’re done.

And the technology is getting better in this regard. The reason why it was hard to find information was because the growth of the huge amounts of information outpaced our ability to distill and filter it out. We spend a lot of time doing that. The technology of being able to break up information and understand information needs is just much better. We’re at the threshold now; we will see a lot more, better information quicker. And we hope to be at the forefront of that.