Tariq Ahmad, MD, MPH, discusses leveraging the AEMR for heart success.
Good afternoon thank you so much for inviting me. My name is Tariq Ahmed and I am the chief of heart failure at the Yale School of Medicine today. We're gonna be talking about whether the H. R. Can help clinicians do better. This is a million billion dollar question that I'm sure every cardiologist thinks about these are my disclosures. Now I don't think anyone's going to argue with the fact that clinicians are human beings and human beings are unpredictable. We've been studying human behavior for thousands of years. There's disciplines from psychology to behavioral economics that try to understand why human beings do what they do. And I think the reason why I wanted to start with this slide is that it's important to think about clinicians being human beings when we're designing E. H. R. Based prompts. If we are to get them to do what we want them to do which is the best care for patients. Now some of you might have heard of nudges. Nudges are a behavioral economic construct that describes things that happen in the background that allow that we use to get people to do what we want to do on the left side to make sure that people dispose of their cigarette butts. It's to get people to buy the large box of popcorn to wash their hands and even go to urologists. Now I think we need to use nudges effectively in the electronic health record ecosystem. How do we do right now? I think we get a solid F. Grade. This is a E. H. R. Based alert that I got last week and it does not accomplish anything other than annoy the clinician who's looking at it. So now this is not news. We've shown in multiple multiple studies that there is no association with just E. HR use and quality of care. This is a great study that showed this to be the case in heart failure but this extends way beyond heart failure. Is it benign? No there have been studies that suggest that the could actually hurt patients and clinicians. This is an example of the epic base thr that was launched without any clear study that in subsequent analysis was shown not to be very sensitive nor specific. So you can imagine that this may have lead to inappropriate treatment for lots of patients On the right side. As an example of the dramatic increase in these alerts during COVID-19 and that of course led to patient to physician burnout as well. So these are not benign. They may hurt patients and clinicians as well. Now we showed what information alone can do in a really interesting study that I think personally I'm biased was was was quite fascinating. So for decades clinicians have heart failure. Clinicians and researchers have said and believed that just by providing clinicians prognostic information about heart failure patients that those clinicians will make better choices that will lead to better outcomes for patients. The reveal HF was a pragmatic randomized controlled trial of more than 3000 patients in our health care system where we randomize patients to their clinician getting information on their one-year prognosis versus usual care, which is nothing. And it was just information. We didn't tell clinicians what to do. And the results were fascinating. It showed that there was an absolute no impact on clinical outcomes, including one year mortality. Readmission data in vision mortality. And then also fascinating was that there was no impact on clinical decision making in terms of what medications were used. Whether or not these patients who were very high risk were referred for palliative care. Um so when we ask clinicians, well, why do you think this didn't make any difference? They said you just give me the information, you didn't tell me what to do with it. So that's an important message because if you just tell clinicians some information without putting it into context or telling them what to do with it, it surprises them. They don't really take it into account and they'll just probably click out of the alert because they're trying to do something else. Can this information alone approach cause harm. We found this in another study, we did the A. K. I alert trial that was randomized again to just giving clinicians information about their patient having acute kidney injury versus usual care, which was nothing. And even though the overall study was negative, it showed that at one of our non teaching hospitals diverse an association with death. So this might just be spurious. But it might be a real thing because if you give information to clinicians who may not be as experienced or know what to do with that information then they may make choices that could hurt patients which it seems like in this case it might have done so. Actionable information is the answer. Here are two studies that demonstrate that on the right hand side is an example of our study which is the prompt HF clinical trial. The prompt HF clinical trial was presented at A C. C. As a late breaker and published simultaneously in Jack earlier this year. And the way that we designed this study was in the outpatient setting. We randomized clinicians to getting detailed information about their individual patient in front of them when they were in clinic seeing that patients saying this is their ejection fraction, this is their creatinine and these are the meds that they should be on and the ones that they're not on. And then we made it very easy for the clinician to just press one button and have access to all the medications that that patients should be on. And essentially put it made it as easy as possible for the clinicians to to do what we thought was the right thing to do. Larry allen in colorado published this really interesting study called Epic HF. He took a different approach. He give the information to the to the patients here and that actionable information was brought to the clinicians to see if that would modify uh the prescription of medications or not. Now both of us found that the use of guidelines for actor medical therapy went up quite a bit on the right side. You see the overall results from the prompt HF clinical trial. There was a statistically significant increase in use of G. D. M. T. after after 30 days, which we was we were quite excited about. Um and on the left side you see the overall results from epic HF. And that also led to an increase in God directed medical therapies. But as one of my mentors uh texted me soon after I was very excited and had presented these results that what are you gonna do with the rest of the 70% of patients who did not have appropriate up penetration. And that's an important question because these passive alerts that are informational still do not completely solve the problem which is to get um get clinicians to get the patients on all the appropriate therapies that they should be on. So the findings that we have from so far are that passive alerts even if they're informational can make a big difference. They can get us to our goal which is the best evidence based therapy for heart failure patients in my case. But we have to push across this nudge continuum so we can keep testing which nudges are going to get us to the best evidence based therapy. And what I believe very strongly is that is that this needs to be in a randomized controlled fashion with the electronic health record. Epic. In our case, randomized controlled trials are very easy to do. And these should be done as part of usual clinical care. So that at the end of these daily experiments that we're doing that we can now modify our behavior in order to get to that goal which is the best evidence based therapies for heart fire patients. So I have several take home messages. Number one Nudges are key to changing human behavior. Human behavior is very complex. You need to use nudges to be able to modify human behavior currently used. E H R based nudges are highly ineffective. These are not studied. These are launched into the HR ecosystem and for the most part they contribute to clinician burnout and they actually may even hurt patients. Information alone is probably completely worthless or even harmful. Actionable information helps. It's better than nothing. But just a little bit, there's a long way to go beyond just this actionable information. The way that we do it right now. More effective. E HR Nudges need to be designed All HR interventions should be studied in a randomized controlled fashion because that's the way to generate information for the future. So with that thank you very much. Please email me or tweet at me with any questions that you may have. And I am happy to take any questions. Thank you so much.