Advancing CVD Management with Digital Triage Tools in the emergency room

CB: I would say in my role at Brigham and Women's Hospital working in the emergency department the three biggest challenges are crowding, staffing and patient throughput. I mention crowding. That's really inpatient crowding, so their inpatient services are full. When those patients are waiting for beds upstairs, they accumulate in the emergency department. And that kind of shunts beds away from newly arriving patients to the emergency department. As a result, we're seeing a large portion of our patients in hallways and there's a big risk in waiting room census being very high. The next is staffing. We're seeing a huge staffing crisis right now, affecting in particular nurses, but also, medical assistance, transporters, techs of all kinds. And as a result, we have fewer staff to take care of more patients and that leads to staff burnout, which then exacerbates the staffing problem. And the last is throughput. As you can imagine, in a setting with fewer staff and fewer resources, the importance of getting patients safely through their visit in the emergency department, with all the treatments and diagnostics we do, is even more crucial today than ever before. So we're really struggling with finding the right timing and throughput of testing, in order to get people through their journey as safely and efficiently as possible. 

EG: As the head of the Chest Pain Unit, I see a crowding of patients. There is a shortage of bed capacities in the hospital and therefore we need to triage patients into safe discharge and select patients who are ideal candidates for admission. The major challenges in digital healthcare is the poor familiarization of physicians with this kind of digital devices and the very heterogeneous landscape. For example availability of electronic health records in hospitals even in Germany. 

CB: The two main challenges for digital health right now are translating the protocols that we have written down really through traditional methods. These could be kind of written on whiteboards on laminated posters around the department getting them translated into our electronic medical record environment. And so making that switch from essentially paper to electronic has been very challenging because that integration is harder than you might assume because the clinical experience of interacting with these protocols is somewhat challenging to get right and as a result exposed to a lot of digital tools that are well intentioned but really not working very well. The other main challenge that we have is a lack of bandwidth among the digital health infrastructure because there are so many competing demands on their time. And as a result, there's usually big lags between when we identify something that needs to be improved and when it's actually implemented. For example we have lots of quality and safety problems that need to be fixed with a digital tool solution. We have times when the digital tool goes down and needs to be put back up from an IS downtime or just a server going down or something like that. So the idea of actually thinking ahead for improvement to bring something new into that environment you're competing with these other things on the list and as a result it's very difficult to get things done. 

EG: Electronic solutions can overcome some limitations. They can for example improve guideline assurance and improve the quality of care for patients. And they even can reduce the length of hospital stay therefore increase patient satisfaction and reduce also costs for staff for treating patients and monitoring. The requirements for implementation of new digital tools to impact practice are really based on three pillars. I would say first, it needs to be clinically correct. The underlying recommendation needs to actually resonate with the clinician and feel valuable to the clinician. Secondly it needs to be well integrated into the medical record. It has to be able to speak with a record that the user can't go out of the record in order to access the tool. And then third it needs to be a business case for the leadership to embrace bringing that tool into a particular environment. And so if any one of those three pillars is missing, then the tool is going to have a really uphill battle in order to get adoption. And so you really need to consider each of those three when you're looking to integrate. Digital tools can impact patient care in the emergency department really by giving a better classification of patients by shrinking that observation or gray zone and more accurately classifying patients who are either ruling in or need further hospital based diagnostics or ruling out and being able to be safely discharged home. I think right now there is some confusion around that kind of observation or gray zone. And so if we can shrink that as much as possible through these tools, I think that'll be a great benefit. I think also giving a bit of confidence to the physicians that they can stop testing the low risk patients and safely get them home without causing harm to those patients will help reduce unnecessary testing and hospital resource use. The current triage of chest pain patients is done using an algorithm that is part of the ESC recommendations. So patients have to have a pre-test probability for the disease, for acute coronary syndrome. Then if this is true then they will have an EKG. And if the EKG is normal, they will have biomarker testing with high sensitivity troponins. And the triage algorithm currently is proposing a baseline value and subsequent testing after 60 minutes or after two hours the so called ESC 0/1h, 0/2h algorithm And with this, the patients are triaged into a rule-out zone, a rule-in zone, and into an observational zone if neither category fits. Digital tools could improve this triage that has some limitations. The most important limitation of the current triage is that the time between blood sampling is not, correctly considered. So it's used as a default, but not actually the elapsed time between blood samples. And second, the numbers of required blood samples are often not fulfilled. So they are protocol violations in the number of blood tests. Sometimes the second blood draw is omitted, and sometimes there are excessive blood draws beyond the diagnostic set. The main trends that we're seeing right now in digital healthcare are risk prediction and machine learning or artificial intelligence being integrated into digital tools. Right now there's a lot of effort put into not just calculate, simple risk scores but actually use machine learning to give a percentage, perhaps of risk or need for hospitalization. And having that information earlier in the emergency department stay might help with kind of anticipating that disposition later in their stay, as well as trying to make a safe disposition of keeping a patient you might have otherwise discharged home. And then what we're seeing more and more right now is a lot of excitement over what AI could bring as a complement to the clinician in order to make him or her more efficient and safer in their practice of emergency medicine, to take them away from being at the computer so much and actually put them more at the bedside where they want to be spending their time. I think that we have to improve health care delivery for the sake of the patient. And every effort that we are doing is beneficial for a patient. And I believe that implementation of clinical decision tools using algorithms that are based on knowledge on guidelines or even, importing artificial intelligence into these systems will be very helpful in the future.