Novel emergency medicine chest pain triage algorithm for clinical optimization

Guest speaker: Prof. Dr. Med. Evangelos Giannitsis MD, FESC - Department of Internal Medicine III Medical Clinic and Polyclinic, University Hospital of Heidelberg, Germany

*This transcript was generated using an AI-based transcription tool and may contain errors. It reflects the spoken content of a live webinar and has been lightly edited for clarity.

I allude again to the principal problem in emergency departments, that is overcrowding. Not only overcrowding, but as you can see on the right side of the panel, the number of hospital admissions for chest pain is increasing, but the number of beds is decreasing.

And this is a figure that shows in different countries how over the years, the number of hospital beds has been reduced for different reasons. Sometimes to save money, the number of beds is constantly less. And the situation where we are working in emergency departments is becoming more and more crowded.

So you can see here the situation in the Chest Pain Unit in Heidelberg University, where we plotted three grades of crowding that depended on the number of patients that were simultaneously in the ED and the number of physicians and how long they had time to spend on the patient. And you see that most of the time was in heavy crowding or at least in middle crowding. So this is a relevant level of crowding throughout the year.

And we know that crowding is associated with adverse outcomes. So every hour delay in the stay in the emergency department increases death. That is seen here, that six hours of length in the ED will almost double the risk of death. And depending on the time, you can see that on the right side. Also, the risk of readmission to hospital will increase by 20 to 60%.

And therefore, I think that one of the great solutions after the introduction of high-sensitivity troponin assays is acceleration of the process of triage and diagnosis. Disposition of patients can be done much faster. And to rule-out, where you can consider immediate or early discharge; rule-in for admission and maybe for catheterization; and then hopefully a small observe zone of hopefully 25% that requires further observation and further testing, diagnostic testing. And troponin testing has received a Class I B recommendation for the first protocol. This is the ESC 0/1- and 0/2-hour algorithm. While the 0/3-hour algorithm is the second choice. One thing that needs to be mentioned is also shown on the right side of the guideline recommendations, that the result of troponin should be available within 60 minutes after the blood draw. So this is the turnaround time. It has to be very quick.

A good sign, a good finding of these algorithms is that they are safe, they are effective, and they are safe. So this has been shown by many, many observational trials, by randomized controlled trials. And in the meta-analysis that I present here, you can see that negative predictive values with these fast algorithms are well above 99%. And mortality rates for 30 days and for one year are extremely low.

Now, many physicians struggle with implementation of the 0/1-hour algorithm because they think they do not have the infrastructure to perform fast blood testing, and they are not able to accomplish the 30-minute or the 60-minute turnaround time. A second obstacle is that there's some confusion. First, physicians need to know which assay they use, and then they have to decide which protocol they want to use: the 0/1- or 0/2-hour algorithm. Because, as was mentioned before, troponin is assay-specific. The interpretation of serial changes is specific and also depends on the timing of the second blood draw. So there is a little bit of confusion here.

And in real life—this is a British study on the real-world implementation of the 0/1-hour algorithm that was tested in two consecutive years. So the first was the immediate testing phase. And then there was a subsequent phase a year later. And I have to go briefly around this slide. The key finding is that even in low-risk patients, where you do not need a second troponin measurement, there are second troponin measurements. Whereas in the middle, in the intermediate risk group, where you need serial testing, that is not commonly performed; between 30 and 60% of these patients receive a second troponin. And even for patients who have high risk of being qualified for rule-in, a subsequent troponin measurement. And all this is, first, is guideline violation, is protocol violation, and delays the stay of patients, that the disposition is delayed.

What about the times? How successful can the 0- and 1-hour algorithm be implemented? You can see here the two consecutive years, and what is shown on the left part of the figure is the 60-minute time interval plus/minus ten minutes. And you can see that throughout the risk categories, this goal was never achieved. And even in the second year, when they had some experience, the learning curve was higher, they never achieved the 60-minute time interval. On the middle, in the middle panel, you can see that in a different graphical display overall and throughout the year. So never reached the 60-minute time interval.

And I can show you now from my own exploration from our data. So on the left side, you see data from my Chest Pain Unit in the published study where we showed the time to the second blood draw within the risk categories. And you see, this is not a line exactly at 60 minutes. This is a right-skewed distribution, covering longer time intervals. On the right side, you can see a supervised observational study, one of the studies that served as the guidelines to recommend Class I B, and you can see here, students or staff had a stop clock and measured exactly the 60-minute interval for the next blood draw. So here, precision was due to supervision by people.

And what does this mean for the interpretation of concentration changes with troponin? So first, this is an example showing applying the different protocols: 0/1-hour, 0/2-, 0/3-hour protocol. So a change value of 3.7 in Case 1 would qualify for the observe zone. If you do not measure within 60 minutes but have a later measurement that is within the two-hour sampling interval, this patient is ruled out. If you have this second case, higher concentration, here it is 6.4. You are in the observe zone. But if you measure at three hours, because the time point of measurement was not at 120 minutes, but it was at 150 minutes, then you are ruled out. So it's a big difference and makes different classification depending on the truly elapsed time. And that is relevant here. You can see here in our analysis that not only the risk categories changed substantially, but also the death rate was different according to proper classification of the elapsed time to interpretation of the change.

Moreover, doing a guided protocol, a supervised protocol by a CDSS tool would improve the number of blood draws that they actually take. And so you could—we saw that a lot of blood draws were taken in excess, too many. Not as per protocol. And in many cases, 14% of cases, there were too few bloods taken. So an ideal CDSS tool would cover important gaps and unmet needs. It would ensure protocol adherence. Could also consider standard operating procedures of local hospitals or national requirements. Could improve guideline adherence and also documentation, that could improve correct diagnosis with effects on outcomes management and reimbursement. Could shorten the time to disposition, facilitate the timing of scheduled blood draws, could reduce the observation time, and so on.

So for such a tool that we designed, only few things needed to be input. So first, the time of the last episode of chest pain in order to see whether the 0/1-hour algorithm is possible. The first troponin value at the time of blood draw, a second troponin at the time. And if you utilize national... if using sex-specific cutoffs. The readout was the annotation of triage category and the annotation of the guidelines that was applying. And then the algorithm also tells you if a next blood draw is required and at which exact time point. And if you do not reach the exact time point, it interprets the result according to the elapsed time point.

So this is a real-time visualization of this, of this tool. So, input is the time of the day. The time when the chest pain onset was. Then the time when the patient had his first blood draw. Then this is real time. Then the troponin results should be awaited. So in this case, it was seven. And then, there's an annotation. So you require a second blood draw because the troponin was above the LoD. And this means this prompts the serial troponin measurement. Then there's another time point of blood draw, and then the second blood draw result that is nine in this case. And then there's again an interpretation of this, with an annotation to which guideline it refers. And tells you this is ruled out, and so forth.

So, and this tool is now embedded in the algorithms due to this tool provided by Roche that covers a lot of topics in chronic disease and acute disease, and in prediction of disease. It is incorporating patient data, physician data, and lab information. And so we are here in the center, we are doing a tool to improve the diagnosis. This tool that I introduced has now received the CE mark; it is an approved tool that can be used. It is commercially available and an ongoing prospective trial is conducted with a scheduled enrollment of 1,500 patients, split into 600 cases and 600 controls. And the outcomes are triage efficiency, adherence to ESC guidelines, length of CPU stay, and 30-day MACE. And so we are expecting the results within the next 3 to 6 months. And thank you very much for your attention.