Hi, I'm Naveen Muthu. I'm a general pediatrician with my clinical practice in pediatric hospital medicine and I also am director of our Human Systems Informatics Lab in the Department of Biomedical and Health Informatics at Children's Hospital of Philadelphia and lead our Clinical Decision Support Committee at the hospital.
WHAT ARE SPECIFIC EXAMPLES OF HOW YOUR WORK HELPED CLINICIANS AND/OR PATIENTS?
Let's talk about something specific from the pandemic. One of the things that was really kind of miraculous during the pandemic was the volume at which new scientific evidence was generated. We initially thought that there might be a certain set of therapies that might be helpful or not helpful, and then very quickly a number of institutions started conducting research studies, granted more on the adult side than on the pediatric side. But based on that, our understanding of the people that benefit the most from treatment from certain medicines and the medicines that help the most (like here are the people that really should get steroids for treating their illness) really evolved pretty rapidly. Normally, when we build a piece of decision support, it is months of work to sort of understand the evidence, translate that into what you might want to build into the electronic health record, and go through a process of testing it to see if it works the way you intended it to. We really didn't have the luxury of doing that because we really wanted to make sure that patients were getting evidence-based care as rapidly as possible. We had this very flexible creative process during the pandemic where we were directly working with our infectious disease experts. They were letting us know as evidence was evolving like “Hey, this recommendation that we have in the electronic health record might be out of date.” There was an expert committee that would suggest updated recommendations, and we would figure out how to get that translated into an order set or whatever else it was that clinicians were actually using to make their decisions during patient care. I think just how facile and rapid we were in keeping up with that in our patient care was really something that I was pretty amazed by, and it spoke a lot to what happens when you don't have a number of competing priorities but you sort of have one priority that everybody's aligned on.
WHAT QUALITY IMPROVEMENT PROCESSES HELPED YOU QUICKLY IMPROVE DURING COVID?
I think just the iteration because there is the possibility that any piece of decision support introduces new hazards in terms of safety and equity. In normal times, I think we are appropriately cautious about introducing any new piece of technology, but really during COVID, we embraced this idea of a plan-do-study-act (PDSA) cycle, and really being able to build things iteratively. So, if we were building decision support for vaccinations, and we hadn't quite figured out how to work through the second dose workflow yet in the inpatient setting, we were still able to roll out something more focused to support the first dose vaccination and then retain a manual process for the second dose scheduling that we then figured out how to solve and then built in to the the technology to have a more automated workflow for that. Just having that sort of iterative approach to solving problems was definitely something that was really valuable during the pandemic.
HOW SHOULD QUALITY IMPROVEMENT, INFORMATICS, AND IT TEAMS INTEGRATE TO DRIVE BETTER CARE?
That's a very interesting question. One of the really interesting things that we've been thinking a lot about is what is the overall structure of governance for those teams, and, if there's a match or a mismatch, how does that affect the ability of those teams to work together. For example, let's say that your quality improvement teams are really pushing a more federated system where local clinical teams like individual divisions, say the cardiology division or the pulmonology division, are really taking ownership of their quality improvement work, but the IT team and the EHR analysts within that team are not organized in a similar way. They have a more central intake process, and they only have one queue that everything goes into. That mismatch makes it such that those local clinical teams can't really take ownership because they don't have control of those resources on the IT side.
One of the things to think about is what is the structure and governance of each of those domains: Is it centralized, or is it decentralized? If it's decentralized, how is it functionally decentralized? And then, do those things align with each other? Because if there's a mismatch, then I think it's really hard on either side of that mismatch to get work done. One group is expecting central priorities, and the other group is expecting being able to respond to local needs, and that becomes a challenge.
WHAT ARE THE BEST APPROACHES FOR SHOWING ROI (RETURN ON INVESTMENT) TO ORGANIZATIONAL LEADERSHIP?
This gets back to what I was saying about the evaluation and like what tools do we use the reason that I am excited about working with the Phrase team. The reason that I think that approach matters is because ultimately it gets back to the thing that everybody cares about is did a patient get the outcomes that they wanted. If we're talking about return on investment, it's really about tying whatever it is that you're doing to all the steps in the process to be able to demonstrate that the patient got the outcomes that they wanted. If it's “we’re building some pieces of decision support”, it’s okay, does decision support change care processes? And did those care processes improve what care patients are actually getting and care about? That's what it comes down to, and that's the return on investment that anybody cares about.
WHAT SOFTWARE OR OTHER TOOLS DO YOU USE THE MOST TO COMPLETE YOUR JOB?
I would say in terms of tools you heard me say the thing that I am interested in is that as we're sort of putting this decision support technology out there, we think that's going to change care delivery processes in some way, and ultimately, because of that change in whatever the care delivery process is, that patients are going to have better outcomes. For us, the software or the tools that we use is really all about doing that evaluation to say, “hey, did the decision support get used?” Phrase is our best way of looking at that information and to then understand, based on that use of decision support, did the use of that decision support lead to a change in the actual care process. Then, as a result of that, did the patients get better? So, we sort of have a broad set of analytics tools that as we get farther and farther out of that chain we start using. But, that's really the set of questions that we're sort of most interested in and the tools that we're most interested in are the ones that can help us answer those questions.
HOW DO YOU ENVISION YOUR ROLE EVOLVING OVER THE NEXT 5-10 YEARS?
I don't know about 10 years. I would say that over the next five years, I think the way that we've started thinking about our clinical decision support program is in three different ways: one is that we've been thinking about the overall use of clinical decision support. We really don't have a great handle on all of the different designs or patterns or functionalities that we're using for clinical decision support. I think we have a lot of focus on alerts and a lot of focus on order sets, but the hope is that, as electronic health record technologies evolve, we're able to take advantage of more functionalities, more designs, and more approaches to decision support. So, really getting a handle on what's out there, what fraction of clinical care is being supported by clinical decision support, and being able to get some sense of how we grow that is a priority for us. The other priority for us is that historically I mentioned alerts are particularly expensive in terms of the cost of interrupting the person that you're delivering decision support to as way of delivering decision support. Given the burden that clinicians are facing from all of the things that we've heard about in health care, we don't want clinical decision support to be one more thing that's contributing to it. We have a focus on shifting away from these burdensome interruptive forms to other forms of decision support that are just as effective. The last piece is thinking about safety and equity, and how do we ensure that any decision support that we're introducing into the system is not creating new hazards from either a safety perspective or an equity perspective, and what are the things that we can do before implementing, and then also how do we surveil for things after implementation to make sure that that's actually the case. Those are our three priorities moving forward for our decision support program.