Designing for Health: Protecting Clinician Attention in the Digital Age with Hasan Ahmad, DO

Intro: 

Hello and welcome to Designing for Health. I’m Nordic’s Chief Medical Officer, Dr. Craig Joseph.  

My guest today is Hasan Ahmad, a hospitalist, clinical informaticist, and Vice President and Associate CMIO at Parkview Health. We talk about something almost every physician knows well: the steady accumulation of EHR interruptions that were probably created with good intentions but somehow never go away.  

Hasan shares how Parkview took a data-driven approach to clinical decision support and eliminated more than 100,000 interruptive alerts every month, roughly a quarter of its alert burden. We explore why de-implementing technology can be harder than implementing it, why every decision-support tool needs an owner and an exit strategy, and how poorly designed results routing can bury clinicians in unnecessary work. Ultimately, this is a conversation about treating clinician attention as the scarce and valuable resource that it is, and designing our systems accordingly  

Let’s plug in. 

Craig Joseph, MD, FAAP, FAMIA:  

Well. Welcome Doctor Hasan Ahmad to the podcast. How are you today, sir? 

Hasan Ahmad, DO: 

Doing well. How are you? Thanks for having me on. 

Craig Joseph, MD, FAAP, FAMIA:  

It is my pleasure. Where do we find you today? Not specifically your address. We’re not looking for security problems. But what city and state are you in right now? 

Hasan Ahmad, DO: 

I am in Fort Wayne, Indiana. That’s the home base for Parkview Health. 

Craig Joseph, MD, FAAP, FAMIA:  

And what do you do at Parkview Health? 

Hasan Ahmad, DO: 

So, I am the associate chief medical information officer at Parkview. Parkview is a 16-hospital health care system across northeast Indiana, northwest Ohio. I oversee mainly our inpatient informatics and clinical decision support. However, I’m our only associate. So, I kind of like, get into a lot of different things, you know, other than other than those things. 

Craig Joseph, MD, FAAP, FAMIA:  

Duties as assigned. I think. Yeah. Yeah. I’m familiar. I’m familiar with that role, sir. Oh, you’re a physician. Well, maybe that’s not so obvious, but you are a physician. So. And I and I happen to know your fellowship trained, which is a minority of the people I talk to because there weren’t a lot of fellowships. Those weren’t options for a lot of people until in the last, what, 5 or 10 years or so. What kind of physician are you? 

Hasan Ahmad, DO: 

Yeah. So, I work as a hospitalist. I’ve been doing that for my whole career. I first started when I first got internal medicine residency. I did nights as a nocturnal. And. Yeah, so I have, you know, people say about the whole, like, how’s the HR doing it? 2 a.m. I’ve lived that. I still do shifts like admitting shifts right now at Parkview. Yes. You’re right that I did Informatics Fellowship I before I came to Parkview, I had done my Clinical Informatics fellowship at the University of Washington in Seattle, and so I joined Parkview after that. 

Craig Joseph, MD, FAAP, FAMIA:  

Now, just out of pure personal curiosity, nocturnal. So, are those people different than the rest of us? Like, what makes someone interested in doing shifts? Only at night so you can be nocturnal? As a hospitalist, I think you can be a nocturnal as an emergency medicine doctor, but someone who mostly works the night shift. Why? As someone? Because to me, like, that is the most horrible thing you could possibly do to me, but I are there. Is it a different kind of work? Is it, is it? What appeals to you? Or was it just. Oh, no, that was the shift I got. I wasn’t really looking for them. 

Hasan Ahmad, DO: 

Yeah, I’ve done nights, I’ve done days, I’ve done everything. So yeah. First coming out I worked as a nocturnal in like a level two trauma center in Tennessee. And you know, it’s kind of different because it depends on where you’re working. Like it’s mostly admissions and putting out fires. And those admissions could come from the emergency department. If it’s a place with an open intensive care unit, you could also be doing admissions for patients who are coming from outside hospitals. And so and while you’re also putting out fires for patients who are already admitted. And so your attention is like very divided in the tasks that you need to complete during your shift. Whereas in the day, like you have set number of patients that you’re supposed to see and you have X amount of hours to see them determine their plan, and it’s a little bit different. 

And there’s more in day. There’s more stuff about like case management, discharge planning. Sure. Just absent from the night shift. There’s some of that like when they come in. But really that’s your goal is to admit the patient and determine their initial plan and move on because you have a lot of things to do during the night shift. 

Craig Joseph, MD, FAAP, FAMIA:  

I’ve just always been fascinated by that. And I know that you’re popular with your colleagues, right? Because most people are trying to avoid the nights and the weekends. 

Hasan Ahmad, DO: 

I did it for a couple of years now. I mainly do like swing shifts for admitting patients. So to keep my head still in the game and see how things are working in our HR system. But I don’t do many nights anymore. But I know how it is. 

Craig Joseph, MD, FAAP, FAMIA:  

Yeah, you have felt the pain, so now you can imagine how to do what your day job is right now, which is mostly spending a lot of time in the electronic health record specifically for inpatient and clinical decision support. But as you mentioned, duties elder Wise describe. But let’s talk about clinical decision support. So, I know that you’re a customer of epic. I think that one of the reasons I wanted I was so excited to have you on the pod is to kind of talk about some of your clinical decision support or and I think one of the things that you started working on right away. Correct me if I’m wrong, was what I used to. What we used to call a best practice advisory is now called opus, at least in the Epic system. Our practice advisories, right? Isn’t that what the O stands for? It is. Yeah. So were people complaining or is this something that you kind of came into and said, “oh, there’s a lot of work that we could we could do”. So how did that. How did that play out? 

Hasan Ahmad, DO: 

It was a combination of both. You know, it was me going to see patients and then getting a bunch of pop ups while I’m trying to think about what I’m going to do for this patient and then saying, you know, I think we have a problem here. You know, I’ve always had an interest in clinical decision support, and so we didn’t have very good governance over it. And we’ll get into all this. First of all, you need to figure out that that’s a problem that you have to solve. And then kind of look into how you how you get there. 

Craig Joseph, MD, FAAP, FAMIA:  

So, I’m always fascinated by the fact that it’s, you know, if you kind of grow up in the system like you were there from Go Live and had been live for 5 or 10 years or something like that, you kind of get used to the way it works, and it takes sometimes a fresh set of eyes to kind of come in and go, oh, wait, I’m just trying to admit a patient, why are you asking me for things that are relevant for outpatient or things that aren’t relevant right now in the middle of the night? And sometimes people kind of just used to this, but, you know, kind of takes a new person or someone kind of coming in who’s not invested in the build and the decisions that were made to say like, this doesn’t seem like it’s optimal. So, you encountered it actually taking care of patients? 

Hasan Ahmad, DO: 

Yeah. And, you know, when we got into it, I thought of like, why do we need to take this on now to get this fixed? Right? There were about three things. One is we found out the scale of the problem. This was not just small annoyances. This was an enterprise attention problem. And so, we found there were 12,000 interruptions, interrupted alerts per day. Each one took an average of six seconds of someone’s time. That equals about 20 hours of clinician attention consumed daily across the system. And many of these are overridden about 90% of the time. Second is the speed. We found a lot of dismissals that were under two seconds. Nobody’s reading anything in under two seconds. So good alerts for being swept away with the noise. 

We were conditioning clinicians to dismiss everything to click before they read. And the third is ownership. Like alerts were firing daily where nobody could name the owner. Why does this exist? Who asked for it? Is it still a problem? No. Silence. And I really like to compare it to pharmacy. So, if we have a med medication and there’s no prescriber and there’s no indication pharmacy would shut that down in like five minutes. But we were fine letting software continue to cause these interruptions with no owner. So we had to reframe the issue from like an ear cleanup issue to more about patient safety, workforce sustainability and having trust in the clinical system. 

Craig Joseph, MD, FAAP, FAMIA:  

You said the G word: governance. So oftentimes as you kind of learned and you just told us some of these alerts or clinical decision support tools are put in place at Go Live because, well, we think they’re going to be a good idea or some physician or someone came and said, hey, this is a problem and this is the way to solve this problem. And it might have been actually functioning quite well, but it’s not functioning well. Now. What was the governance when you when you started this project at a high level? And then what were some of the changes that you needed to put into place to make sure that you, you kind of solve the problem? A and B made sure that in the future you’re going to be constantly looking for things that weren’t a problem when you started them but become a problem over the years. 

Hasan Ahmad, DO: 

There’s a few things. First of all, I chair our Enterprise Clinical Decision Support Committee. I was trained in clinical support during my fellowship and have had interest in that ever since. And so, the first thing we really had to do was understand that alerts get born and nothing ever dies. There’s a committee to approve something, and it’s nobody’s job to retire from something. And we needed to come to consensus that that’s what’s happening. 

Craig Joseph, MD, FAAP, FAMIA:  

Wait, let me just I want to stop you right there, because I don’t think I’ve ever heard it so succinctly put that there’s a committee to create this stuff. But there’s no job. No one’s job to make it die, to retire it. Yes. Yeah. That’s yeah. That kind of just stops you right there. So, and that’s true not only of clinical decision support but like. 

Hasan Ahmad, DO: 

A lot of things. 

Craig Joseph, MD, FAAP, FAMIA:  

Right. Lots of things in medicine. Hey. Well, when I did my residency 30 years ago because I’m incredibly old, this is how we did it. And yeah, well, we may not do it that way anymore, but that’s the way I was trained. And it takes someone to kind of shake you by the shoulders like, no, we don’t do it that way anymore. All right. So that’s a great point. I’m sorry to take you off, but I wanted to just put an exclamation point. 

Hasan Ahmad, DO: 

Yeah. This is a very important point is that and there’s asymmetry in the psychology. So, nobody’s ever in trouble for over firing. And whenever you try to remove something, every everybody vividly imagines the bad outcome. That’s going to happen the week after you turn something off. And so, everyone feels like adding is careful and removing feels risky. And so even when the data is saying the opposite. So, this is like one of the main points that we had to really understand and acknowledge before we could make progress. You know, there’s other there’s other things too, like the fact that building an alert is just a form, but when you are trying to turn something off, it’s usually some type of extended negotiation to turn things off. 

Sometimes it’s like the owner who requested it left like five years ago, and there’s nobody left to say yes to keeping an alert. And so, we really needed to have a back door for these alerts. We needed to have named owners there for the outcome of the alert. Who is supposed to check if the workflow evidence or requester has changed? Who has the authority to retire an alert? Otherwise, people just say like, leave the alert alone because it’s the safest organizational choice, even when it’s not safe. And so we needed to give clinical decision support the same lifecycle that we give that pharmacy gives the formulary, which is you have front door, you do validation, you do monitoring and you do deliberate removal. And so most you know, most organizations just have a front door. They don’t have back door. And that’s part of the problem that why organizations have so many alerts. And you have this problem of alert fatigue in the organization. It’s because of that. 

Craig Joseph, MD, FAAP, FAMIA:  

So, it sounds like we should just put pharmacists in charge of that stuff. You know, like that’s a joke I just threw out there as a joke. 

Hasan Ahmad, DO: 

Pharmacy alerts are the most frequent type of alert in our system. And, you know, that is one of the reasons in our Enterprise Clinical Distance Support Committee that we brought in pharmacy and we brought in medication safety pharmacy leadership because we even though in our initial alert tuning that we did, we found that, you know, making changes in pharmacy alerts especially tough. And we need to have the right people there to, you know, be part of making the decisions for pharmacy alerts. So, we had to include them also in our enterprise clinical to support committee so that everybody is not making decisions in silos. We’re all on the same page looking at the alert and seeing all the same data when making decisions. 

Craig Joseph, MD, FAAP, FAMIA:  

Are you talking about drug interaction checking? 

Hasan Ahmad, DO: 

Yeah, I mean there’s those. And then there’s also open related to pharmacy. There are multiple ways people can get interrupted from a pharmacy standpoint. Usually, you know, you could have a vendor that you get your medication alerts from, and then you could have, you know, custom designed to open in your health system that, you know, your pharmacy analysts had developed over time. And so, we need to think about both when we’re talking about interruptions. 

Craig Joseph, MD, FAAP, FAMIA:  

So how does one start? You came in, you kind of saw as a practicing physician that there were a lot of alerts. You started poking around and saw that, hey, there doesn’t seem to be a way to retire or kind of kill any of these things. And a lot of them who a lot of these alerts and other clinical decisions, support tools that had champions for good reason at the time are gone, and there’s no one to champion them or to comment on them. 

So, you know, how do you kind of get your arms around, okay, how many do we have? How am I going to go through? Or how are we as a governance group going to go through every alert? Or did you do that? And to determine like, hey, how are we going to get back to the best number and quality of a decision support that we have? How does one even begin? 

Hasan Ahmad, DO: 

So usually, you know, you have somebody who’s requesting something. The burden is really distributed like nobody is nobody in the organization is when they’re bringing an alert there; nobody is seeing that there’s 12,000 alerts per day. They’re only focused on that one alert that’s tied to the thing that they’re caring about. And the problem is that the data is in dashboards. So what do you have to do first is you need to make an inventory. I mean, it’s boring, but you have to do it. Inventory makes you look bad. That’s why nobody wants to start doing it. Nobody’s going to like, you know, get excited about a spreadsheet with 500 plus alerts, and you have to assign owners to things. 

You have to determine who owns accountability for these alerts. And some people would rather not know, but we decided we’d rather know. And because you can’t govern, we can’t see. You need to have an inventory and then you can’t prioritize what you have not inventoried. So, you have to start with an inventory. And so, every alert got an owner determining what’s the purpose and looking at the numbers. And then we had to prioritize by volume action rate risk and the value. And you know the numbers are very important because they are a way to use data to deepen the removal. If, for example, if I have an alert that’s firing 90,000 times in six months and it has a 1% action rate, like the conversations over like nobody has to win an argument, the number is already there. So that’s why you really need to do like data analysis of all of these alerts and bring the data as your reviewing them. 

Craig Joseph, MD, FAAP, FAMIA:  

Did you just kind of create a custom made spreadsheet or did you use a tool? I think there are third party vendors that sell these kinds of services or tools, or you can get it from your EHR vendor. How did you do this? Or did you just kind of, I’m going to make a report. It’s going to have exactly the columns I want, the way that I want and work from that. As a start for your inventory. 

Hasan Ahmad, DO: 

We didn’t use any third party. We found everything inside our EHR. So first of all, we had to determine what are all the things that we care about in doing the review and where is that data located. So, a lot of times this was in one of those dashboards that the EHR vendor epic has, and it just takes somebody to go in and look at the dashboard, but we don’t look at it. And so, you know, we had to determine basically a way to say this is how you do a review of an OPA. And these are all the metrics that we need to go in and look at when we’re doing that review. And here’s where you find each one. It some of them are in the stash board. Some of them are in, you know, the place in epic where they build these opus some. 

That’s where some of the information is. Such as the criteria that are used to build these things. Like somebody has to go in and look at that criteria and make sure that it’s still that’s working. Right. Otherwise that’s one of the things. So that’s where you need to figure out what really matters in the review process. And then you find everything in the HR. You don’t need to use a third party for that. 

Craig Joseph, MD, FAAP, FAMIA:  

So, it sounds like what you’re saying is, hey, to do to really evaluate each individual alert; you kind of need two pieces of information. One is technical, right? Was it built right? Is it showing to the right people? But to answer that question, you need the clinical or operational background as well. Right? I want to know if the criteria are, the reason for whom this alerts going to pop. I want to know if it’s the right, the right group of doctors or nurses. Well, if you’re not a clinician or an operations person at your hospital, you mean you may not know that answer? And so, did you find did you have people who had both of those skills, like did you have doctors and nurses and therapists who understood the behind the scenes for epic, or did you combine, you know, folks who had some high level understanding of epic, but combine them with an analyst, or was it all of the above? 

Hasan Ahmad, DO: 

No, unfortunately, we don’t have a lot of those people, and I think most organizations probably don’t have a lot of those people. 

Craig Joseph, MD, FAAP, FAMIA:  

I think you’re right, I was hoping I was hoping. 

Hasan Ahmad, DO: 

And you know, you have to turn in who owns it from the clinical side. And that can be hard if like, you have not figured that out before, like you’re going into an alert, and you have to determine who’s going to do the clinical review for this alert to say, is it still something that we need? Is it still following guidance? Is it following some guidance that was there ten years ago and nobody ever checked to, you know, whether it’s still needed like somebody needs to do that? And how do you determine that? We found that. Go and look at like who is the who are the most frequent people that get this alert and talk to them like that’s one way to do it is, you know, you have to talk to the people who are actually using the thing. 

Craig Joseph, MD, FAAP, FAMIA:  

Talk to the users. Oh, I don’t know if I would want to do that. Sometimes they’re unruly. 

Hasan Ahmad, DO: 

Yeah. If you talk to somebody who’s not using it, they’re not they’re not going to give you a great response. 

Craig Joseph, MD, FAAP, FAMIA:  

I know, but they’re usually happier because they’re not using it. So. 

Hasan Ahmad, DO: 

Oh yes, that is the case. 

Craig Joseph, MD, FAAP, FAMIA:  

All right. So, you so you had a technical owner and a clinical owner and those two had to kind of work together, I presume, to kind of come up with the with the answers to your questions of, hey, is this thing effective? And do we still need it? 

Hasan Ahmad, DO: 

You basically have to do the technical review, which is mainly the build and how it’s working in the air, and then the clinical review. And then you need to, you know, bring those two things together and determine, okay, so what do you need to do with this alert. Do you need to keep it as it is. Be modify it, see sunset or get rid of it? Or is this alert doing something very similar to another alert. And we just need to like merge the two. 

Craig Joseph, MD, FAAP, FAMIA:  

Yeah. 

Hasan Ahmad, DO: 

Like those are the four decisions that somebody has to make. And then if it’s modification what modifications are needed that’s it. But you need to do that initial process to get to that decision. That’s like the upfront work that is needed in this type of cleanup. 

Craig Joseph, MD, FAAP, FAMIA:  

And so, at a high level, what did you find? Did you do you have any kind of numbers that you can share with us? Did 10% of your alerts go away or 1% or, you know, at a high level, kind of. What were some of your findings? 

Hasan Ahmad, DO: 

We so far were able to get rid of 100,000 interrupt alerts a month. That was 25% of our total interrupt alert burden, and we’re not done yet. So those were 100,000 alerts per month that nobody had to see, but they were getting shown to people and interrupting, you know, their cognitive thought processes. I mean, it’s not like we turn things off and we have 100,000 gone in a day. This is like multiple months of work that got us to that point. And so that’s, you know, I think of it also as keeping people’s thought processes. You know, I gave you the example of, you know, I’m embedding a patient, I get a bunch of pop ups. Could be I have that patients picture. I’m admitting a patient at 2 a.m. 

I have that patients picture in my head. I have, you know, their kidney function, their list of medications, and what I’m thinking about doing. And then right before I’m about to put an order in, I get a pop up, like, I’m not going to remember all the things that I was thinking about before I got the pop up. So it’s like, how do we selectively remove interruptions from people so that they can preserve their train of thought? Could be like they were in the process of finding something subtle with the patient. And because we interrupted their thought process, that’s gone. So we need to be very selective when we are interrupting somebody. I like to be almost be stingy about interrupting somebody, because attention is like the most expensive thing in in the health system from, from a clinician standpoint. And so, we need to only interrupt people for things that matter. 

Craig Joseph, MD, FAAP, FAMIA:  

So, about a quarter of the alerts of the actual kind of pop ups went away. Was that from a was that a small number of actual kind of alerts that just popped a lot, or was it a lot of alerts that just popped a, you know, a moderate number of times? Were there any easy wins where you’re like, oh, we just knocked out a huge number by taking out these two opens? 

Hasan Ahmad, DO: 

I wish it was. I mean, and some of them were not just like taking the alert out something. It was modifications and the alert, like, you know, getting it to selected users in front of a broad amount of users. Some of it was modifying the criteria of when an alert is shown, because there’s so many things that can go wrong with alerts, there’s like, what is the action that you’re supposed to do in the alert? Is that visible to the user when they’re, you know, when they’re showing the alert? Is this the right problem that still need to be needs to be solved? This is the right way to solve that problem. Like there’s so many questions that go into these things, which is part of like the design that it’s a design issue and it’s that design is what impacts the build. 

It could be what leads you to make modifications elsewhere in the air, like the ultimate problem in which of it which is why you have to get to the ultimate problem of these things, like what was the purpose of this thing? Because it could be that you had an order set problem and somebody put in a request for an alert because something wasn’t in an order set when the patient was getting emitted. And so, you really need to dig deep and like, figure out what was the initial purpose of this thing and, and figure out are their ways that are worked better for the workflow and are non-interactive that that solve this problem like the alert is not the problem, the solution probably there. 

Craig Joseph, MD, FAAP, FAMIA:  

So, you know, all of this is kind of under the, I think the umbrella of DMP. Right. Taking the opposite of implementation. And as you alluded to earlier, it’s not as simple as well. We’ll just get rid of it because everyone comes. And you didn’t say this, but I’m going to extrapolate. Hey you can’t get rid of that. People will die. 

Hasan Ahmad, DO: 

Yeah, right. 

Craig Joseph, MD, FAAP, FAMIA:  

Like and maybe people don’t come out that strong sometimes they do. But my experience, it’s like, hey, you know, that’s dangerous for you to get rid of that. So, any tricks or hints on, you know, how do you respond to that clinician or operations person who saw an example one time where while the stock from to do this thing. And so, we put an alert and you want to get rid of it now. And how can you promise me that this these other doctors aren’t going to make that same mistake? And of course, you can’t promise that. But any kind of any hints for how to deal with some of these D implementation problems from a change management perspective. 

Hasan Ahmad, DO: 

Developmental is one of the most important things that we’ve done here at Parkview to earn back some of the clinician trust in the usability of the system. And I would first say that, you know, safety and usability are not opposing goals. If you have unusable safety tools, eventually that’s going to stop being safe. And so, you have to ask for questions. You have to ask, is this a meaningful risk? Can the person act? Is this the right person in the right moment? Or is this being shown to everybody, regardless of whether it’s the right person? And does this measurably change behavior and outcomes? If something is overridden 95% of the time, it’s protecting nobody. It’s just clicking. People just click through it. 

And then the habit continues to learn that matters. So that was the first thing you do sometimes encounter resistance. And the way that I like to approach things is to first look at what are the outcomes that we’re trying to measure by this thing. And then, you know, if you can’t get to this solution, can we at least try to change something in one of the areas? Can we pilot this in one of the areas, like remove it or making modifications? Sure. And see what changes and measure what the outcome of that is like. If we have some safety metric attached to what we’re changing, then I mean, who wouldn’t be down for that? We’re like, we’re measuring what you care about. Like, so that’s one way to do it. And, you know, even though we got rid of 100,000 alerts, we didn’t have any safety event tied to this. So that is one of the ways that helps us to still continue this work, is that a lot of this stuff just was not needed. 

Craig Joseph, MD, FAAP, FAMIA:  

Trust but verify. Sounds like, hey, I’m going to take this little thing out. We’re going to just make sure that your concerns are not validated because we’re hoping not to see any problems and we don’t. And then you can take a little bit more and take a little bit more. Let me pivot as I see our time is starting to run down. So we’ve talked a lot about open and clinical decision support. I feel like the ugly stepbrother of that talk is results in basket. If you ask people about if you give them two things to complain about, the electronic health record probably in the top five will be both of these rights? One is pop ups that are ill timed or reminding me to do something I was just about to do and getting all kinds of things in my in basket that requires my attention. 

So that seemed to be another problem that you encountered. And what have you done or what are you doing? Is it the same kind of approach from a governance perspective, like, okay, can we take a 30,000ft view and see what some of the problems are and then try to just chip away at them? Is that what you did or doing? 

Hasan Ahmad, DO: 

We did a big project on improving our results, routing in the system, and I can talk a little bit about that. So, you know, there’s a difference between pop ups and results. Routing results. Routing is causing people to be buried slowly in their inboxes. And nobody sees why it’s happening. So. And there could be two failure modes. One is too many people are getting a specific result and then nobody wants nobody knows who’s going to own it. It causes duplicate review, diffusion of responsibility, and results can land where nobody felt responsible. The other thing, and you know, if a result is coming in three inboxes, it can be just as onerous if a result is coming in no inboxes. And so, yeah, the result logic is invisible to the clinician. They see the output that it’s in the in in basket, but they don’t see why. 

And it’s easy to mislabel this. We have a, you know, inbox volume problem. You know we’re getting all these things in our in basket. Or people could label it as a training problem. But the real issue is, you know, a chain of ownership decisions that were buried in configuration. And people built these things over time, like one reasonable rule at a time. But nobody owned the whole picture of how results get routed. And so each rule is perfectly logical. But some of them is chaos. Like everything is working as designed, but the system as a whole is not. So we realized that this is not really an IT problem. The rules are working as they’re designed, but there needs to be governance over this. Like people need to know for result comes in. How and who do we want this result to go to, who can act on this result. And so, we had to create a governance infrastructure around that with clinical leadership to make those decisions, to adjust the rules that were there that determine who a result goes in the system. 

Craig Joseph, MD, FAAP, FAMIA:  

And so, in the olden days, on paper, and let me be clear, we’re talking about labs and imaging studies. And we’re mostly talking about outpatient. Is that accurate? And Ed, maybe. 

Hasan Ahmad, DO: 

Because inpatient mostly people just see results in like their results review. They don’t need to get a bunch of things in there in basket. So yeah, it’s mostly military said. 

Craig Joseph, MD, FAAP, FAMIA:  

So, in the olden days, if I’m in my clinic and I ordered a, you know, a CBC or a chest X-ray, that result would either get faxed back to me or come back on paper in some way and get attached to a chart. And there would only be that copy. Right? And it was my responsibility, since I ordered it, to look at it and decide what I needed to do, if anything about it, and call the patient and or, you know, that kind of a thing. And if I was out, then someone on vacation, some another colleague of mine would be responsible for covering for me. But with the advent of the electronic health record, we’ve been we have the ability to send that result to multiple people. So it might be that, well, the cardiologist ordered a cholesterol test. That’s good. The cardiologist is interested in that. 

But then maybe the system as a whole has made a decision. Well, maybe the PCB would probably want to know about that. There might be others who might want to know about that, who we can identify, even though they never ordered the lab and may not know what to do with the response. And so, to your point, then, that one lab result could go to multiple doctors, and most of us don’t know why we got it. We just know that that we got it. And many of us are kind of anal retentive. And so, we’re like, okay, well, I must now do something with this result because it was sent to me even though I didn’t order it. And I’m not sure what I’m supposed to do with it seems fine to me. 

Maybe I should tell the patient if it’s fine in that kind of context. Then the way you kind of went at it is just with a governance of, hey, we’re going to make some rules that sounds like, right, like, hey, if you if you ordered it, it’s yours. 

Hasan Ahmad, DO: 

Yeah. I mean, ownership is the biggest thing. Like, can we get to a single owner who’s going to own this result, who can actually act on this thing? If too many people are getting the same result, that’s how you end up missing one. So, and it’s not carelessness, it’s just arithmetic. Like I can you can only focus your attention on so many results, and you have to figure out which one you actually have to deal with. And I’d rather have one clear owner with inbox quiet enough to actually see it. And so, we had to develop like to develop clear ownership. We brought clinical leadership from the ambulatory side, from the inpatient side and from the emergency department side. And we kind of went through like, hey, this is a the potassium result that’s coming in. 

And let’s walk through the flow of where this this potassium result goes. And we had to just walk through and determine like, what are all the edge cases of where things can get screwed up in the system. And let’s configure it for how the work is actually done or should be done, so that the right people get the result. And in in baskets are quieter. But we are maintaining the safety of the right person getting the result. 

Craig Joseph, MD, FAAP, FAMIA:  

Where are you in this journey? Are you just beginning or are you near the end? Like where do you think you are? 

Hasan Ahmad, DO: 

So, we have done a lot like we completely redid our whole main routing scheme. Routing scheme is basically all the rules that determine where a result goes. We found cardiology had their own routing scheme, but it wasn’t working. And so, they completely got rid of it because it wasn’t working. They were getting results like that were going to the wrong place, and it was because they were going through their own routing scheme, which wasn’t working. 

They could just go through like the main, main way results get routed. So, we fixed all the routing, the ways results get rooted in the system. What we’re currently working on is and so that was the result of that routing workgroup. What we’re currently working on with results routing is that what we found was that when a provider gets added to the care team, staff were selecting something that says like receive all results on the button and what that was doing. I mean, they may have had a reasonable intent, thinking, oh, I’m just going to keep the clinician informed. 

Craig Joseph, MD, FAAP, FAMIA:  

Yeah. Of course. 

Hasan Ahmad, DO: 

Continuity of care, you know, going to want to know about this. Let’s like send them all the results. So, what it does is route results for tests even if the clinician never ordered them. And then the clinician has no idea why these are appearing. 

Craig Joseph, MD, FAAP, FAMIA:  

All of a sudden. 

Hasan Ahmad, DO: 

And so, this setting that was intended to keep someone informed was silently like redefining who’s responsible. And so, it was one click, that fused communication, routing, implied accountability. And the system never distinguished between like informing somebody or making them responsible. And so that’s what we’re currently dealing with. And this is you know, mainly the fix here is a technical and governance fix. They had a behavioral component that they were selecting. Yes. For a question that the default was no. So, it wasn’t a default problem. The default was no. And they were changing it. So, we had to also train them. They’re not seeing the full scope of the choice of their selection. But this was a design problem, a deeper design problem. 

Staff are not accountable for the results. They have the ability to opt clinicians in. So, we had to start asking like training to compensate for this permission problem and what we whatever ended up doing now is removing the ability for those rules to change the setting so that we’re aligning authority with accountability. 

Craig Joseph, MD, FAAP, FAMIA:  

That’s funny, because I was just thinking, oh, I could solve this problem for you. Make that button go away. 

Hasan Ahmad, DO: 

Exactly. That’s what we’re doing. 

Craig Joseph, MD, FAAP, FAMIA:  

It’s hard to imagine some time where I would want someone who’s not me to opt me in for getting lab results. 

Hasan Ahmad, DO: 

We’re in the process. So, we found out it’s like a security setting. We just take that security setting away. But then what we also have to do is go back to all those records where it’s marked yes, and change it to no. Yeah. And so that’s the big cleanup that needs to happen. And you know, all our ambulatory clinical leadership is all for that. Everybody’s for that. It’s going to reduce all the noise that people are getting in their basket. 

Craig Joseph, MD, FAAP, FAMIA:  

All right. Well, we could talk for hours more and I would be interested in doing that. But we have run out of time. We always ask the same question though, at the end of our of the podcast, which is what is something in your life that is so well designed, it brings you joy whenever you use it. So, is there something like that? 

Hasan Ahmad, DO: 

Yeah, it’s going to sound funny. My Toyota Forerunner is like the thing that’s so well designed that, you know, every control does one thing work with gloves on, keeps on running. It’s serviceable. It’s built. Assuming the owner is going to maintain it. It welcomes user modifications. You know, maintenance and modification are not a failure. Yeah. We’re part of the strategy. And this is you know, which is everything I believe about clinical systems in truck form. 

Craig Joseph, MD, FAAP, FAMIA:  

So, I love it. It doesn’t. Yeah. It doesn’t have to be high tech. And yeah, all those things that you mentioned are definitely product features. Whether it’s an expensive product like a truck or, or, you know, much smaller for your kitchen, that makes a lot of sense. Oh, I love that. 

Hasan Ahmad, DO: 

Yeah, we can apply a lot of things from, you know, different fields like the automotive field to the EHR because, you know, strategies are pretty similar. 

Craig Joseph, MD, FAAP, FAMIA:  

Well, Doctor Ahmad, thank you so much for your time. This has been a great conversation. We’ll definitely get some contact information from you. If people have specific questions, we’ll put those in the show notes. Appreciate your time and thanks for sharing some of your learnings with us. 

Hasan Ahmad, DO: 

Yeah, thank you very much. Thanks for having me. 

Outro: 

Thanks for tuning in. We hope you enjoyed today’s episode. For more on Hasan follow him on LinkedIn and be sure to review the show notes for more content from Dr. Ahmad.   

Check back for more episodes of Designing for Health wherever you listen to podcasts or on NordicGlobal.com. We’ll see you again next time on Designing for Health.  

Links:  

https://www.linkedin.com/in/hasanbahmad/ 

Interruptive Alerts Are a Losing Bet: 

https://www.linkedin.com/posts/hasanbahmad_clinicaldecisionsupport-cds-cmio-share-7502456177872744448-gHc9/  

The AI ‘kill switch’ is becoming a design requirement at health systems: 

https://www.beckershospitalreview.com/healthcare-information-technology/ai/the-ai-kill-switch-is-becoming-a-design-requirement-at-health-systems/