This transcription is provided by artificial intelligence. We believe in technology but understand that even the smartest robots can sometimes get speech recognition wrong.
Today on This Week Health.
I practiced for over a decade as an ICU pharmacist before I moved into IT and it's not new. Alarm fatigue. Alert fatigue. Anyone who's spent time in an ICU knows it's cacophany of sounds and noises and smells. Not to mention the computer system and the alerts and everything but I think the pandemic has pressed the gas pedal on this for clinicians. There's so many new demands on their time. And anything that can be perceived as a barrier just seems to escalate it.
Thanks for joining us on This Week Health Keynote. My name is Bill Russell. I'm a former CIO for a 16 hospital system and creator of This Week Health, a channel dedicated to keeping health IT staff current and engaged. Special thanks to our Keynote show sponsors Sirius Healthcare, VMware, Transcarent, Press Ganey, Semperis and Veritas for choosing to invest in our mission to develop the next generation of health leaders.
All right. Today we are joined by Anna Dover with First Data Bank and we are gonna have a conversation around clinical decision support and some of the things we're doing around alert, fatigue, clinician burnout, and those kind of things. Anna, welcome to the show.
Thanks for having me.
I'm looking forward to the conversation. This is an area. just coming through the pandemic. We've had a lot of clinician burnout. We now have a fair amount of fatigue in this area, but not all of it is really related to just the pandemic itself.
There's an awful lot of other things that are at work there. And What are you hearing? What are you seeing in the industry around that?
It's funny. I, I practiced for over a decade as an ICU pharmacist before I moved into it. And I think it's funny it's, it's not new alarm fatigue, alert, fatigue anyone who's spent time in an ICU knows it's cacophany of sounds and noises and smells. And it's not to mention the computer system and the alerts and everything, but I think the pandemic has just, it's almost like pressed the gas pedal on this for clinicians. There's so many new demands on their time. And anything that can be perceived as a barrier just seems to escalate it.
What we're hearing from folks. As part of the, kind of the burnout piece, the alert fatigue piece is still there. It's almost like it's exacerbated, but staffing shortages are a major problem and, it's not just from burnout, but it just worsens it. One of the physicians that I know from my past experience shared that due to staffing shortages, they actually had to stop elective procedures in the, or that's challenging, right? that's a bottom line for a health system when you're dealing with that type of shortage.
Yeah. I just got off call where we were talking about that one line item on everybody's P and L right now, which is just absolutely killing them, which is that contract labor line, which in a lot of cases is traveling nurses.
And not only nurses clinicians just in general filling gaps and whatnot. And I've seen where. I think it was peace health they've decided, Hey, we're gonna take one of these hospitals and we're gonna see what the new norm is, and we're not gonna have traveling nurses. And I think what they're trying to determine is what services are they gonna have to shut down or whatever to maintain that facility because they can't continue to maintain that facility. With the shortages that they have and with the costs that have gone up. So dramatically on that specific line.
When I think one of the other things we've heard too is with this use of contract labor to fill these gaps where does training come in and do you assume that there's some baseline understanding of the EHR because they've worked with it before in another health system which ties into, I mean, it all comes back around to decision support is meant to be a stop gap, a safety valve.
So it becomes even more critically important if someone fat fingers, something in a system they're unfamiliar with, it's not one EHR is one EHR, whether it's the same software platform or not.
All right. So let's back up a little bit. Tell us tell us about FDB and your role at FDB.
Sure. So FDB has been around for a long time, over 40 years. We provide. Medication and medical device information for use and software. So most folks think of us as that data behind the scenes. Typically we're program two, we provide information that promotes safe ordering, prescribing, dispensing, barcode administration of medications as well as that clinical screening.
So we can identify if two drugs that a patient's on could cause a problem. That's what folks typically think of us as providing, we also provide advanced clinical decision support. We're really innovating in that space, bringing in patient information that's available in EHRs. We have a medication or medical device database that provides information.
So think about. Implantables in the, or supply chain information. And and we recently launched a new product called Vila, which is an e-prescribing platform to help fill some of the gaps in that space, as well as provide the first e-prescribing network for veterinary medicine. So been around a while and expanded in the space, basically
e-prescribing in veterinary medicine. Wow. Yes. Wow. that's interesting. I I'd never, that's gonna take us in a completely different direction. I'm I'm gonna,
we don't wanna we don't wanna distract from the primary area that I work in, but I think we have a, we first we launched a medication. Database for pet meds. So we've done our med knowledge is our core product. And this was a request from some of our customers in that retail space saying, Hey, can you help provide information for us to dispense pet medications? So it's a growing, it's a growing market and that was pre pandemic. I'm sure we've all heard about folks adopting pets in the pandemic when everyone was home. So you can can see how that, that area's grown as well.
Oh gosh. The number of people who've gotten. So you're doing that same kind of thing, the drug, drug interaction, and that kind of stuff
so we provide the ordering and dispensing information right now for pet meds. We have on our roadmap to provide the clinical screening as well. We just haven't pulled the trigger on that one, but yeah, well, because drug drug interactions are there as well. And as a side note, I've actually had to call the veterinary poison control line myself, where I accidentally put the dog's fleet preventative on the cat and I had to call him it's a cash. Just so you know.
Wow. No, I actually I'm. I'm very aware of that I knew some people who've gone from the medical side over to the veterinary side and from a career standpoint, they said it's an easier place to have work life balance and all sorts of other things.
it was, it was a really interesting conversation to hear how it's, how different it is. So FDB cause I want to talk more about this physician and clinician burnout. And specifically around alert fatigue. I had a, I had an interview with Jake Lancaster with Baptist Memorial out of Memphis, Tennessee.
And we were talking about alerts specifically, and he was saying that they had just finished a massive project where they did away with gosh, I'm gonna get these numbers. I I'm, in fact, I'm not even gonna say the numbers, but a massive amount of alerts. And he said, we got it down to, and when he said the number of alerts, it still sounded a massive number of alerts to me it's it was a significant almost year long project to cut out. Let's say about two thirds of the alerts, but it still left a significant number of alerts that the clinicians had to deal with. is that the case across the board?
Yes. Yes. That's the, that's the short answer. Yes. And I actually, prior to FDB spent about six years of my life optimizing decision support and health system in an EHR around medication warnings, custom decision support, leading the team, designing and building that custom decision support. So when you think about it, it's, that's what we measure with alerts.
When you think about. And EHR, there are all sorts of things that pop up and interrupt in the workflow that may not even be clinical decision support. It could be related to a piece of the workflow that you have to do that isn't actually logged or reviewed. So there's a tremendous number of alerts. I would say when I look at it just. Looking at the numbers because there's no standard metrics out there. But if you just, when I think about my time in clinical practice and the times that I would have to stop a physician from doing something as a pharmacist, from a safety perspective, it's significantly less than the number of alerts that you see out there.
Our clinicians really doing that many things that should stop them in their tracks. I don't think so. And it, it really boils down to. are we able to bring in all the information we need to, to bring in context for whether or not that alert should fire? Right. That's the hard part. That's the kind of the art of decision support I'd say.
And the hard part with alerts is if you fire 'em off too many times, they become white noise, right? So then they're just clicking through things. And the thing that you really wanna stop them and say, whoa, whoa, whoa, what are you doing? It's just not noticed because they just click, they just click through 'em cuz they have to click through so many every day.
Well, and it's a risk benefit calculation every time. And I can tell you, when you look across disciplines, a physician versus a pharmacist versus a nurse, their perception of the need for alert varies dramatically based on the profession. And I would even venture to say. Folks that have grown up in the EHR world are more reliant on alerts because these alerts they've they've had them they may have become a part of their workflow to remind them to go monitor something later. So there's this balancing act of Thinking about how you wanna use the alerts. Is it really a fail safe, or is it a reliance to remind you to do something later? Because there's this piece of it? Well, if we always use it to remind us later, now we're losing the ability to see very often.
So let's talk about pharmacists and the per prescribing of drugs. I, how many of the alerts fall into that category? Do you think a majority of the alerts fall into that category?
As far as volume that are firing to pharmacists
they're firing to pharmacists and well, clearly the pharmacist, it would be almost all.
But what about physicians is most of it around drugs and medications.
A large volume are around medication alerts, but there are a lot of custom alerts that folks put into their software. And very often the request for that decision support is coming from outside of the end user that sees it.
So it could be a quality measure, alert, like identifying a care gap, which has value identifying something that the patient should be on that they're lacking. But then there are also medication alerts and it's a significant volume. It all depends on how the organization has kind of pulled the levers to put that out in the wild.
📍 📍 We'll get back to our show in just a minute, we have a couple of webinars coming up and I don't like webinars. I think they are oversaturated at this point. And I think a lot of them are not all that good. And so that's why I think I'm the perfect person to put together webinars for you. I make sure that we have great topics.
I validate them with CIOs. I make sure we have great guests and I make sure. We actually plan ahead and we actually spend time together before the actual webinar. So it's not just spur of the moment stuff, but we make sure we identify the things that we should talk about in those webinars. And we even collect questions from you ahead of the webinar so that we can make sure to talk about the things that you want to talk about.
So let me tell you a little bit about the two webinars we have coming up. There's a global survey. That we talked about on the today show a thousand cybersecurity professionals found that 30% plan to change professions within two or more years, and cybersecurity threats are growing. And, you know, quite frankly, we need to make sure that we recruit, retain and optimize our staff so that they can be our frontline.
And so the first webinar we're doing is how's your frontline recruit. Retain and optimize your cybersecurity team. And we're gonna talk to experts from Christiana care and Seattle children's and Seuss about their thoughts on this exit of security professionals and what you can do to stay ahead of that.
You can join us August 11th. At, 1:00 PM Eastern time and you can register right on our homepage this week, health.com on the top right hand side, you're gonna have the two upcoming webinars. You go ahead and click on those again. That is August 11th at 1:00 PM Eastern time. The next one, we're going to talk about ransomware, but I've seen a lot of different ransomware, webinars.
I love this one. The topic we came up with is Don. Pay the ransom and rubric is bringing together some great leaders from Thomas Jefferson university in St. Luke's university health system and and rubric themselves. And we're gonna discuss solutions around protecting all of your healthcare data, especially as you're moving to the cloud.
And specifically, we're also gonna talk about epic. Backup in Azure. And what rubric gets doing around that, that webinar is going to be on Thursday, August 18th at 1:00 PM. You can register for both of them. Just go to our homepage this week, health.com upper right hand corner. You're gonna see both of the graphics for those click on the one you wanna attend, fill out the form. And we will see you then now back to our show. 📍
📍 We've talked about volume so talk about the challenges that, that prescribers nurses, pharmacists face over and above just the volume of the alerts.
Yeah. So there are, there are the challenges with alerts and the workflow. There are challenges being able to stay focused on that care of the patient. It can be almost a frenetic workflow, especially when they're getting pulled in a lot of directions. And you can see this repeated from health system to retail pharmacy, where there are just so many new demands. I think the pandemic brought in a different level of critical illness as far as that patient mix in the, in the care that they're providing for patients. So there's a lot packed in there for them to kind of navigate.
Are we hearing. The challenge of alert, fatigue or over alerting were irrelevant alerts. Is that leading to adverse events or outcomes that, that we don't want to have at this point?
So ECR for the last several years, this is something one of the areas that we monitor has listed alert slash alarm fatigue on its top 10, 20 safety concerns in healthcare for the last several years. And. Once the pandemic hit, it fell off, but we didn't solve the problem. We, we just had,
we just had other problems. Yes. Got it.
Yeah. And so I think it just, it compounds and, and it's almost like the spotlight's not quite on it. I think organizations are trying to do more with less. So one of the products that I'm, I manage, we are providing specificity for the patient and our targeted met alerts, but we're doing it in a way to make it easy for the end user or that analyst to deploy and maintain so decision support requires not just an implementation, but maintenance to be good. And it's still important in the midst of all these other things going on, drug shortages, diversion is a major, hot topic.
There's so many things going on firing at these health systems beyond the pandemic, but that have been exacerbated by it.
Interesting. So. As I think about this Let's. Head into the technology a little bit. So how is the technology getting smarter better, faster, or is it still reliant on us? I mean, I heard this whole project from Jake where they essentially went in and looked at every alert and they sat down with a group of people and they went, every alert are, are we getting to a point now where the technology knows, Hey, these alerts aren't really being acted on. And they're smart enough to tell us, Hey, here's a group alerts that aren't being acted on. Is the technology getting better? I guess it's the question?
Yeah. Well, I would. The logging has always been there. The hard part with that type of data is there's so much of it. You talk about that volume of alerts, right? How do you even it's like eating an elephant, you have to do it one bite at a time and, and you need to figure out where you wanna start.
We, as a company have invested pretty heavily in this type of analytics platform. We started with our alert space analytics. It was a feature in this customization tool and we could identify for you the medication alerts, very focused on that. That's our core expertise. Just a top 10 top 10 list, most frequent by volume.
And you could drill into the specific. FTV content causing that alert and decide how you might wanna customize it. We could show you which medications we're causing the alert. So maybe you say, Hmm, at our organization, we're not concerned about propofol being given with this drug and this effect because they're monitored.
And so they can remove it and, and customized we've expanded in that space because just analyzing that data is very challeng. Where we're bringing in some of those custom decision support alerts with med alerts, and it's a more robust backend it's met just for analysis. And we have, we've already built four different dashboard views of different things.
We're working on a fifth and we have on our roadmap about three or four more. So some of it is just, how do you marry the data together in a way that is impactful and thinking about the different levels. There. I think we're getting better at that and realizing what a CMIO may wanna look at is different than what an analyst will wanna look at both valuable.
They likely wanna be able to see what the other one's looking at, but when you surface up changes, so to more directly answer your question. Yes, we have that data we're working on statistical models. We will bring in machine learning models over time likely working for quite a while before we'd expose it.
there's a lot of hype out there with AI and machine learning as a company. We're very. Focused on keeping a clinician in the loop and verifying that it's when we surface something using a machine model, that it's always correct. That's kind of, that's our rule at FTP. It has to be correct. Or we, we don't surface it.
do we ever utilize the clinicians in their workflow to provide feedback? In other words, here's the alert and it says clear. Irrelevant, whatever. And when they click on irrelevant, we now start to see, okay, there's a trend going on here that this is an irrelevant alert. We can, whatever. Are, are we utilizing that?
Yes. So we have, in some of our newer alerts that we've constructed that we're providing the messaging and the acknowledgement reasons or override reasons. We have a button where we put in every single one that says not clinically. So that's a way for us to flag it in that analytics piece to, to pull that out.
I know of folks Ray Chan from Sentara years ago did something similar in his software system. So folks could do that. And I think you're hearing more about EHR vendors putting kind of a thumbs up thumbs down or some other way to provide feedback smiley face frowny face to say whether or not they found value in the alert That requires someone taking the time to respond to it. In my past experience, it tends to be a lower volume because you have to think and process the alert and decide to give the feedback. , but it is very valuable because that means they probably had an emotional response to that alert. If they took the time to tell you that this wasn't useful. So it's still valuable information, even if it's not a high volume,
are we seeing the EHR players want to get into this space that you're talking about?
Yes. I think if you were to look at like the analytics healthcare analytics, the biggest players tend to be the software vendors, the EHR software vendors.
So they, they, compile that data. Right. And so that can be, that can be challenging, but now you have a wide array and there is value in having a vendor that is focused on how do I make this easy for you to gain an insight. Pulling data out is onerous. I'm sure in your past life you had reporting teams and can understand some of the challenges there.
So what does the what does the future look like in this space? I'm curious I mean, we're making progress. And actually I'm, I wanna come back to this, but I'd like to hear what does the future hold in this space, in this clinical decision sports space?
So for us, we continue to move forward and really focusing on alerting when it really matters. We've moved into that precision space where we have pharmaco content. we're working on aggregating information. So thinking more like a physician, right? We have something in our content called a clinical consequence. So we aggregate around this clinical consequence or the, the effect on the patient of these safety checks and then put it all into one view.
So you can see all of the medications and medication combinations that could have that ill effect on the patient. So it's really, how do we, there's so much information out there for clinicians. I think the future holds, compiling it into something that can be informative and legible, because that is one of the great challenges. It can feel phonetic. If you have a, an array of alerts popping at you, whether it's local decision support or from a vendor, right.
Yeah, so I'm not a doctor. So cut me a little slack here as I try to talk about this. So we have all this genetic data and this genetic data is starting to come in. We have the drug, we have drug drug interaction, but we talk about this personalization, and we talk about this ability to say Hey, this drug isn't good for 75% of the population, 82%.
We could actually lower the standard because we can be very specific with the drug to the. 30% of the population that it will have a positive impact on cause we can identify them in some ways through that. And again, I'm way out of my league here, but is that the kind of things we're looking at to be able to prescribe at that level of personalization?
I think so. Yeah. We if you think about behavioral health and psychiatric medications, there's compelling evidence out there. CPIC will tell you there's value in doing genetic testing prior to. Prior to initiating the drug a psychiatric medication in antidepressant or an antipsychotic so that whether or not it will be effective, think about getting that patient therapeutic sooner finding the drug that can help them with their condition faster.
I think that's, that's already, that's already there. Right? It's thinking about. Identifying when it matters and helping get the information out there. Pharmacogenomics has been out there for a long time. I graduated from pharmacy school over 20 years ago and we were talking about it then. And it's been kind of waiting as the evidence there it's there.
So I think, yes. And I think consumers are more aware of the impact of these things. You have, you have at home genetic testing folks can do that it doesn't replace working with a provider. If you're going to use it. in Guiding drug therapy.
The most high performing systems in your head right now describe what separates them, what the really good systems that you deal with the health systems that you deal with, what separates them, what are some of the activities they do? What are some of the things that you, you look at and you go that's, that is what separates them. That's what makes them leaders in the.
From a decision support perspective. I'd say, yeah. Okay. So from that perspective, I'd say they have a process. They have, they have a process that says, we know we need to continue iterating through this. It's not you, you fix it and it's done. So they have ongoing.
So there's, there's a, there's a maintenance and a governance that's sort of set up when you've put it. in To say, Hey, this isn't, this isn't a, we built the house we're done. This is a it's a constant iterative process.
Yes. So there's that, that iteration knowing we need to review and refine in an ongoing basis, something that was working yesterday may not be working today for any number of reasons with the system is always changing. The other thing is. An awareness that there's only so much space for decision support. So again, that idea that something that was put in two years ago may not be providing the value or may not be necessary, or may not be as important today for something new that you wanna put in the system.
So I'd say there's that awareness as well. And that governance piece, I can, I cannot overstate the value of that because as we mentioned before, perception of the importance of different alerts varies by discipline and by specialty. And so having that conversation, there's a natural friction between physicians and pharmacists nurses, even respiratory therapists or other clinicians in the space.
So knowing who you're targeting with that alert, having the conversation, having the right folks at the table, to weigh in on the decision making.
So governance, what is a good makeup of governance? I mean, I assume there's a, a process that is apparent to people of how to escalate it, a challenge and those kind of things, but then there's a group people. What kind of governance processes have you seen and seen and who's a part of them that, that really works.
And so you have to have your CMIO. Absolutely. You have to have physicians at the table. Pharmacy needs to be there. Nursing or nursing informatics needs to be at the table at a minimum. I would say that's for an overarching governance structure that we've seen likely needs some analysts or folks that are doing some of the work with decision support there as. When you get into medication alerts, the high performing ones tend to have physician stakeholders engage with pharmacy.
Now some pharmacy leads the charge and it works well, but for the most part, it's good to have physician input in there. And that tends to require more of a subcommittee because it's a little more digging into the details with medication alerts. There's a lot packed into that type of. Decision support and usually requires because it's more of a direct safety check versus identifying a care gap.
It requires a little more sleuthing to make sure you're not going to unintentionally create a safety issue. So I would say typically you'll see kind of a subgroup that digs into that and can bring their findings up to the larger committee.
Fantastic. One of the things I've been doing at the end of these interviews, and I've been shocked how much information I unearthed with this question, which is what's the question I didn't ask that you're that, that you're surprised we didn't talk about.
think the question that, that you didn't ask is how do we measure, how do we know when something works well or doesn't right now it's very rudimentary. It's that fires a lot. That's overridden a lot. Right. But there's so much more that goes into whether or not it's effective decision support.
And it's almost like we measure what we can. It doesn't mean we're measuring everything we need to. And I hope in the future, that's where we head and we start to kind of unearth more of that and learn more. Of what we need to be measuring.
Yeah. Cuz we have competing goals here. Right. We have safety and then we have clinician burnout. And a lot of times when we're dealing with these alerts, it's a lot of anecdotal. This is, this is why earlier on I was trying to get at what do we what are we tracking? What are we seeing? And you said, yeah, the log files a treasure trove of information, but it's a massive amount of information. And that's how I feel in cybersecurity too. When I describe cybersecurity, people are like, well, don't you see these attacks happening? I'm like, yeah, like 50 of them, every second . And so we have to determine which one of those 50 was real.
And then another second happens and we have 50 more, we need to evaluate and to a certain extent. We can't do that as humans anymore. We have to create algorithms and models that are at least able to go through the 50 and go, Hey, these 40 don't even bother these 10. Yeah, but these two look at these two and I would assume that at some point technology is going to.
Step in here, as you say, not replace the clinician aspect of it. We did, we, that's not what we're going for here, but to assist them and say almost to know when to fire off an alert and when not right now, it's a very hit a gate. Here's your alert kind of thing. And there's some sophistication that if it was a pharmacist on the other side who was firing off all the alerts, they'd go, yeah, fire that one off. No, no, yes, no, no. Are we gonna get to that level of sophistication, I guess is the question
I'm not sure of the place of that type of machine model and alerting yet. There are folks out there working with it. I will share very frankly, we do have some NLP and some things like that. And our And are alerting now, but I told you we've taken a very strong stance of, it has to be 100% correct every time if we're gonna use it. Right. And that's really hard to do there's you need pretty massive data set for that, but I think where we are right now is we're using what we know is available consistently in the EHR to provide context, lab values diagnoses that are likely to be added because we know that there could be challenges with diagnosis, documentation Demora other demographic information about the patient.
We can calculate risk scores. So we're using the data that we have now we're monitoring the way that it performs and we're constantly revising how we do that. The other thing that we're doing is providing options. So in our catalog, you may wanna turn on our QT prolongation use case, which allows you to have a more specific alert.
You may have heard of this one, but it's basically an alert to prevent a potentially lethal heart arrhythmia or heart rhythm. So it's important information, but it's hard to be specific enough. And so we add that specificity, but we know there can be variability in systems. So we have options. We have six options based on your makeup, your clinical decision making, which one would you like to deploy?
So I think it's not a machine, but it is saying an awareness. Of this variability by health system, by software system. And so on to, to bring that context in,
and that is the challenge of software, cuz you're developing it, not for a single system with a single set of tech of applications. But as I said to people, the EHR is just one piece of it. We bring in data from. Hundreds of devices, all hundreds, thousands of devices strewn all over the hospital. We bring in information from tons of software packages we had. As I like to tell people, we had 900 different applications going at any one time.
Well, that's an awful lot of applications and a bunch of them. We're feeding the EHR and you, you have to make sense of all these different things that are coming in there. So it's yeah, it's, it is. It's challenging to be to be in, in your role of product management when the the sandbox is not finite.
Right. I mean, it's kind of fun, too. New challenges every day. Interoperability is the promise is there and we've come a long way. There's a lot to unravel still in there. And I think there are challenges, but you know, for us, we're a content company at our core. So while we have some web service integrations in one vendor, we may have a more traditional integration in another it's it's about the content.
It's been good for us as a company too, really growing in that space. We provide interoperability content. So we're putting it out into the wild. It's interesting. When we start getting it back, what happens in the software? Like to your point, each piece of software, while it can consume the same data from us, what it outputs can vary pretty significantly. So I think that you bring up a great point. Most health systems have many, many, many systems interfacing in, and then you have HIEs and all this information moving with the patient. There's a lot to kind of think about as you're designing and. What is reasonable reasonable to expect on a consistent basis, but also what is a reasonable value to see in that data?
Fantastic. Anna, I wanna thank you for your time. I really wanna thank you for sharing your expertise with us. Thank you very much.
Thank you for having me.
What a great discussion. If you know someone that might benefit from a channel like this, from these kinds of discussions, go ahead and forward them a note. I know if I were a CIO today, I would have every one of my team members listening to a show like this one. It's conference level value every week. They can subscribe on our website thisweekhealth.com or wherever you listen to podcasts. Apple, Google, Overcast, everywhere. Go ahead. Subscribe today. Send a note to someone and have them subscribe as well. We want to thank our Keynote sponsors who are investing in our mission to develop the next generation of health leaders. Those are Sirius Healthcare. VMware, Transcarent, Press Ganey, Semperis and Veritas. Thanks for listening. That's all for now.