I first heard of the concept of an emergency department integrated intensive care unit (or ED-ICU) from Scott Weingart on EMCrit. In my mind, this job sounds perfect. You get to manage the first few hours of undifferentiated critical illness. These hours are, in my mind, the most interesting in all of medicine – involving empiric treatment, frequent reassessment and adjustments in therapy, a broad differential diagnosis, the cognitive exercise involved in finding the right path, and of course multiple procedures. Considering the association between prolonged ED boarding of ICU patients and poor clinical outcomes, (Carr 2007; Chalfin 2007; Mathews 2018) it has been hypothesized that ED-ICUs might also lead to better patient care. This is the first paper directly looking at the impact of an ED-ICU on patient outcomes…
Gunnerson KJ, Bassin BS, Havey RA, et al. Association of an Emergency Department-Based Intensive Care Unit With Survival and Inpatient Intensive Care Unit Admissions. JAMA network open. 2019; 2(7):e197584. PMID: 31339545 [free full text]
This is a retrospective, before and after trial looking at the impacts of the implementation of a new ED-ICU at the University of Michigan.
Any adult patient coming through the ED.
A period of 897 days after the ED-ICU was opened. The ED-ICU consists of 5 resuscitation or trauma bays and 9 patient rooms. It is staffed 24 hours a day by a dedicated emergency physician, as well as residents and/or physician assistants. The nursing ratio is 2 patients per nurse. All patients are assessed and treated in the main emergency department before being transferred to the ED-ICU as deemed necessary by the treating team.
A period of 897 days before the ED-ICU was opened.
The primary outcome was 30-day mortality among all ED patients before and after ED-ICU implementation.
There were a total of 349,310 patients treated in the emergency department during the study period. The mean age was 48.5 years and 54% were women. The mean daily census of the ED-ICU was 6.9 patients. Overall, more patients were seen after the ED-ICU was implemented, and they were somewhat older and sicker.
The mean time to ICU level care (ED-ICU or ICU) was 5.3 hours in the before group and 3.4 hours in the after group. The number of patients receiving “ICU level care” within 6 hours was 78% in the after group, as compared to 58% in the before group. (It is not surprising that, with the existence of an ICU in the ED, patients got there faster.)
In an unadjusted analysis, there was no change in all cause mortality (1.97% vs 1.98%). However, in the adjusted analysis (these patients weren’t randomized, and were therefore different at baseline) mortality was decreased (2.1% vs 1.8%; OR 0.85; 95% CI 0.8-0.9). This would translate to 1 fewer death for every 333 ED visits.
Interestingly, there was a slight increase in the adjusted mortality that occurred in the ED (0.08% vs 0.11%; OR 1.36; 95% CI 1.09-1.71). The authors conjecture that the presence of the ED-ICU allowed physicians more time to discuss goals of care, and led to comfort care measures being instituted earlier. I think this is an essential part of all emergency care, but admit it requires a significant amount of time that isn’t always available.
Overall admissions to the formal ICU went down a little (3.2% vs 2.8%). The number of ICU admissions less than 24 hours duration also decreased (12.5% vs 9.1%). There was no difference in the number of patients transferred to the ICU from the ward within 24 hours of being admitted.
Obviously, this trial is of interest to people working in the massive tertiary centers that already have ED-ICUs or are considering instituting one, but what do these results mean for the rest of us – the vast majority of us who work in emergency departments where this will never be a viable option?
I am going to assume the difference seen here is real. That is a huge assumption. There was no change in unadjusted mortality. It is a single center study, so the results may not be generalizable. It is based on a chart review, so the data is likely to be imperfect. It is a before and after trial design, which introduces many possible biases, most of which have been discussed at length on this blog before. (The two groups are not exactly alike, and the results relied on adjustments, which are far from perfect. Care may have just been improving with time. Most importantly, the trial design isn’t blinded and the authors clearly weren’t disinterested in the results.) But for now I am going to ignore those many biases and just consider the implications if we assume the results are true.
If ED ICUs actually save lives, what does that say about normal emergency departments? Will the care I provide – working in hospitals that don’t have such a resource – always be inferior? These results could be disheartening to the vast majority of us who will never see the inside of an ED ICU. However, instead of considering the unit as a whole, I think it is important to break down the individual aspects of the care being provided, think about which are truly contributing to the benefit, and see if we can replicate those in a standard emergency department setting.
One obvious benefit that many ICUs have over emergency departments is nursing ratios. Although tight budgets and (poor) managerial decisions are increasingly leading to higher patient to nurse ratios in ICU settings, really sick patients in critical care almost invariably receive 1 to 1 nursing (with other nurses helping out). In the Michigan ED-ICU, it was a completely manageable 2 patients per nurse. Obviously, our resuscitations also start with multiple nurses in the room, but there is a critical period, after the majority of the action has finished, where we all wander away to look after other patients, leaving the resus nurse to manage the still critically ill patient. Unlike ICU settings, the emergency nurse is frequently also responsible for multiple other patients. Funding and staffing emergency departments to ensure we can provide 1 on 1 nursing for critically ill patients throughout their stay makes sense. Managers don’t like this, because critically ill patients come in intermittent boluses, and it means that you will appear to be overstaffed on the quieter days. However, there are always things that can get accomplished on those days, and a well functioning department needs to be staffed for the busiest days not the quietest.
There are a number of routine practices in most ICUs that aren’t routinely provided in the emergency department, any of which could be contributing to the benefit seen here. Examples include routinely elevating the head of the bed, starting chlorhexidine washes, ensuring appropriate (light) sedation, focusing on delirium prevention, and being meticulous about lung protective ventilation strategies. Some of these interventions are simple, and we should probably just institute them without waiting for further evidence. Others are more time intensive, and given the existing demands on emergency department staff, could have unintended consequences. (Time spent on these interventions takes time away from other tasks.) Therefore, it would be nice to see studies that look at outcomes after instituting individual interventions in the ED. (I am told that there will be a paper out soon from ACEP and SCCM outlining what they consider to be best practices. I will update the blog when I see that publication.)
Another difference to consider is that after the implementation of the ED-ICU, physicians and staff are gaining a great deal of experience not traditionally available to emergency providers. Outcomes might simply be improved because physicians are seeing more sick patients and performing more procedures. Skill improves with experience. There may also be benefits from more frequent interactions with the main ICU. This degree of experience with critically ill patients will never be available to most emergency clinicians working in the community, but we might want to consider its implications. Perhaps we need to create formal partnerships that allow practicing community physicians to gain more experience in the ICU. Maybe we need to rely more heavily on simulation training to ensure that all emergency departments are comfortable managing critically ill patients, even when they are seen infrequently.
However, the biggest difference between an emergency department and an ED-ICU might simply be the amount of time a doctor spends at the bedside. (When I asked Scott Weingart about this study, his guess was that the benefit seen was simply from doctors and nurses being at the bedside for longer.) Most emergency doctors don’t have a lot of spare time. We barely have time to use the bathroom or have a snack on most shifts, so asking us to spend more time with patients isn’t easy. However, knowing that there could be a mortality benefit from spending more time at the bedside of critically ill patients is valuable information. So few of our interventions save lives. If I can make an impact on mortality by spending more time in the room – taking my time with a more thorough exam, multiple re-examinations, thinking about the differential diagnosis, and ensuring I am not missing anything – that is something I want to do. It will, of course, depend on the day. Making a sprained ankle and a viral cough wait 30 minutes longer to reduce the risk of mortality in a critically ill patient is a no brainer. On the other hand, that extra time may not be worthwhile if it means leaving multiple patients with potentially life threatening conditions unassessed in the waiting room. It will have to be a risk assessment based on the day.
Of course, time is a valuable resource, and although I decided to gloss over the problems with this study, this trial certainly doesn’t prove that there is a mortality benefit. The difference could easily be explained by bias or artifact in the trial design. Therefore, we need to be cautious before rushing to institute any massive changes based on the results of this trial. There are some simple changes that are no brainers. Intubated patients should almost all have the head of the bed elevated. A good sedation package is just good care. We should probably all consult with our ICUs to see what interventions they think are important, and I will add the ACEP/SCCM checklist when it is published. If an intervention is going to be started as soon as the patient gets upstairs, it certainly makes sense to start it downstairs, assuming we have the resources available to us. More importantly, we should just get these patients up to the ICU as soon as possible. (The concept of a critically ill patient having to wait 6 hours for an ICU bed is somewhat foreign to me.) If it isn’t possible to transfer the patient upstairs, another solution is to have the ICU come to us. (Sometimes the bed isn’t available yet, but the ICU nurse can physically come to the ED to assume care for the patients and start ICU protocols).
A lot of this discussion might sound a little ambitious for many emergency departments. However, most of us got involved in emergency medicine because we wanted to look after critically ill patients. We want to make a difference. Whether or not the results are valid and generalizable, I think we should all think carefully about this paper’s implications for our own departments.
This before and after study demonstrated a reduction in adjusted mortality after instituting an emergency department ICU model of care. All emergency departments should consider what aspects of ICU level care they might be able to adopt in order to improve the outcomes of their critically ill patients.
EMCrit Wee – First Study of the Benefits of an EDICU
REBEL EM: Impact of ED-ICUs on Mortality and ICU Admissions
Carr BG, Kaye AJ, Wiebe DJ, Gracias VH, Schwab CW, Reilly PM. Emergency department length of stay: a major risk factor for pneumonia in intubated blunt trauma patients. The Journal of trauma. 2007; 63(1):9-12. [pubmed]
Chalfin DB, Trzeciak S, Likourezos A, Baumann BM, Dellinger RP, . Impact of delayed transfer of critically ill patients from the emergency department to the intensive care unit. Critical care medicine. 2007; 35(6):1477-83. [pubmed]
Gunnerson KJ, Bassin BS, Havey RA, et al. Association of an Emergency Department-Based Intensive Care Unit With Survival and Inpatient Intensive Care Unit Admissions. JAMA network open. 2019; 2(7):e197584. [pubmed]
Mathews KS, Durst MS, Vargas-Torres C, Olson AD, Mazumdar M, Richardson LD. Effect of Emergency Department and ICU Occupancy on Admission Decisions and Outcomes for Critically Ill Patients. Critical care medicine. 2018; 46(5):720-727. [pubmed]
Morgenstern, J. What can we learn from the ED ICU model?, First10EM, December 9, 2019. Available at:
One thought on “What can we learn from the ED ICU model?”