(Last Updated On: March 4, 2019)

Effect of Emergency Department and ICU Occupancy on Admission Decisions and Outcomes for Critically Ill Patients

Kusum S. Mathews et al. Critical Care Medicine. May 2018. [paper]

Why I Chose This Study

We have no beds, the MICU is now down to our crash bed” It seems all too frequent that we hear those dreaded words from either our fellow intensive care unit (ICU) colleagues or nursing staff. The critically ill patients keep coming into the emergency department (ED) and the ICU is already at full capacity. With these odds against the care provider teams, how do we respond?

For the ED physicians, do we continue to consult the ICU for admissions even though the bed situation is tight? Does the ED accept the responsibility of ICU patients boarding until a bed in the ICU or until an overflow ICU makes a bed available? Or, do the ED physicians provide more intensive care in the ED until the patient is safe for a step down or floor admission. Inversely, what approach does the MICU take in these at capacity situations? Are more patients rejected simply due to the lack of beds or are patients accepted regardless of where they end up (ED boarding vs other intensive care units)?

What harms do patients face when they are triaged secondary to hospital resource factors? Is there a way that we can avoid these situations and expose the patients to less harm? For me, this topic seems crucial in our daily practice and it affects the care in both in the ED and the ICU.


The volume of ICU admissions from the ED has increased around 50% from 2001-2009. Hospitals struggle with this increase in critically ill patients as more resources and a higher level of care are required. Issues arise when the ICU has no available beds and critical patients are still presenting in the ED. Situations like this can lead to patients being denied an ICU admission where they would regularly be admitted when beds were available, leading to increased hospital mortality.

Other implications with increased ICU admissions include an increase in length of stay in the ED. Critically ill patients have about a 32% increase in ED stay time, be it from boarding till a bed opens up or being denied an ICU admission. These long wait times for admissions can cause further burdens on the ED, such as higher cost, increased usage of resources, and lower adherence to best practices. More importantly, studies have demonstrated that critically ill patients who endure boarding times greater than 6 hours experience a higher risk of inpatient mortality.

Research Question

The purpose of this study was to investigate the effect of ED and ICU capacity strain on ICU admission decisions and to investigate the potential association of prolonged delays in admissions on in hospital morbidity and mortality.

Study Design

This study was a retrospective cohort study completed at a single academic urban tertiary care hospital with a closed 14 bed medical ICU with 91% average occupancy.

In the ICU there was a nurse to patient ratio of 1:2, coverage was in-house pulmonary and critical care boarded physician, a pulm critical care fellow, and house staff. There were also other intensive care units which could serve as overflow for admissions to the medical ICU when it was full. The emergency department had a five bed area that was specifically designed for acutely ill patients, staffed by an ED attending and upper level EM residents. The nurse to patient ratio in this area was 1:3.


The subjects for this study were patients aged 18 years and older with ICU evaluation requested by ED staff from 10/1/13-6/30/15. The ED physicians would request an in-person ICU consult with the final decision of admit or deny made by the ICU attending. If the patient was admitted, they would board in the ED until an ICU bed was available. To note, if the patient was boarding in the ED, the ED physician would continue as the primary team with support from the medical ICU until admitted. If declined by the medical ICU team, the ED physician would admit the patient to the inpatient medicine team, with an ICU consult if needed.


The primary objective was to identify predictors of ICU admissions decisions (accept vs deny), specifically examining the effect of ED and ICU volume on these decisions.

Secondary objectives were to measure the effect of post consult ED boarding time on patient outcome of in-hospital mortality or morbidity, captured by the presence of persistent end organ dysfunction and or death (PMOD+D) at 28 days, adjusting for ICU admission, boarding time, and other patient/hospital related predictors.

Patient characteristics collected:

  • Age
  • Gender
  • Race/ethnicity
  • Insurance
  • Prehospital location (nursing facility/hospital vs home)
  • Severity of illness at the time of the MICU consult (this study used the Mortality Probability Model III (MPM-III) score)
  • Time of consult (day vs nightshift)
  • Primary admission diagnosis
  • Code status (from time of consult to ICU admission and to the hospital discharge)

Hospital related factors:

  • Continuous measurements of ED and inpatient census (specifically ED census counts at the time of consult, the number of patients actively being managed by the ED team and those in the high acuity section).
  • Inpatient census was obtained hourly from each intensive care unit.
  • Both the ED high acuity service and the medical ICU were at about 90% occupancy at average.
  • ED length of stay pre-consult
  • ED boarding time from time of consult until ED departure to either ICU or wards
  • Admission to other ICUs as overflow
  • ICU hospital length of stay


The following variables were analyzed using various statistical tests: For predictors of ICU admission decision, the authors used T testing, Chi-squared testing, analysis of variance and non-parametric testing to look for differences in baseline characteristics and ICU admission decisions. Multivariable logistic regression was used to determine the odds of receiving an ICU accept admission decision by patient and hospital related characteristics. For predictors of persistent organ dysfunction and or death (POD+D), propensity score methods were used to adjust for baseline characteristic imbalances in ICU admissions (defined as the probability of being admitted to the ICU, conditional on measured baseline characteristics). A stepwise logistic regression model was performed using a p-value of 0.2 to determine baseline characteristics that predict ICU admission. Multivariable logistic regression was performed to determine risk factors associated with POD+D with the main variable of interest being ED boarding time. The goodness of fit was assessed with the Hosmer-Lemeshow test.


There were a total of 854 consults for ICU admissions during the study period which represented 43.7% of all ICU consults received.  Of the ED consults, there were 455 patients (53.3%) accepted by the ICU with 57 patients (12.5%) who required overflow admission to another ICU since the medical ICU was full. Patients accepted to the ICU were younger (mean 61 vs 65), came from a non-nursing home facility (12.5% vs 24.8%), and had higher MPM-III scores at time of consult. The patients also had more pulmonary system diagnoses on admission (accept cases of 41.5% vs deny cases of 30.8, p < 0.05).

The medical ICU being at capacity was the only hospital related factor significantly associated with a lower probability of being accepted to the ICU (OR of .55 (.37-.81)). Patient related factors associated with a lower odds ratio of receiving an ICU accept decision were older age and nursing home origin. High severity of illness and having no care limitations at the time of consult were associated with higher odds of being accepted.

Results from the multivariable logistic regression adjusted for ICU triage decision propensity scores demonstrated that long ED boarding time after consult is associated with an increased probability of POD+D. Nursing home origin and higher severity of illness were also associated with higher odds of POD+D.

Ambulances line up outside Mercy Hospital Chicago.


Overall, the study supports that critically ill patients have lower odds of being accepted for medical ICU admission when at full capacity, although there are beds available in other ICUs. The authors also state that longer boarding times in the ED were associated with worse outcomes for critically ill patients. There were only a few patients who were admitted to other ICUs as overflow but there was no significant difference in outcomes. There was, also, no detected effect of ED volume when it came to the ICU teams’ admission decision.

The limitations of this study include the observational study design. There was insufficient electronic medical record documentation specifically about the decision of the ICU to admit or deny admission. Therefore, this study tracked the capacity at the same time the decisions were made. This has the potential to reduce bias as the ICU team may deny fewer admissions if they know they are being studied. But organizing the study in this way could attribute causation to simple coincidence.

Other limitations included the small population study size, the small sample sizes in subgroups, the fact that the severity of illness score was collected at time of consult and was not a dynamic measurement throughout the hospital course, and limited chart documentation about discussions of code status from ICU consult, to ICU admission, to hospital discharge. There was also a lack of data on resource utilization especially on cost, transfers, re-admission, and the difference between ICU admission or denial.

There was also limited data in the measurements of ED crowdedness at the time of ICU consult. Dynamic changes in patient volume and acuity take place during a patient’s ED length of stay and these were not recorded. Also, this study was conducted in a single institution which could have different admission processes and structure compared to other centers.

As for my take on the study, I thought it was interesting that the ICU denial rate was so high. 47% of ICU consults from the ED were declined. Personally, I have never dealt with denial rates that high and I would be concerned if this was happening at our clinical sites. The article doesn’t go into detail about this which makes it difficult to determine why the patients were denied so frequently. With a high block rate as above, there would appear to be some sort of issue in the admission process.

While the article introduced the problematic nature of critically ill patients boarding in the ED they did not go into detail of what factors increase morbidity and mortality. For an example, in the ED nursing care isn’t typically 1:1 or even 1:2 as in an ICU. So, a critically ill patient boarding in the ED can experience less intensive nursing, or other patients under that nurses care could receive less attention. In my own experience, I try to continue intensive management of boarding ICU patients to prevent decompensation, but this can become confusing for nursing staff when two teams of providers are placing orders on the patients (ICU and the ED).

Though still early in my career, I feel confident sending blatantly critically ill patients to the ICU. Intubated patients, those requiring extended periods of time on noninvasive positive pressure ventilation, patients on IV drips, and those requiring frequent lab or neuro checks definitely require ICU care. If a bed isn’t available in the medical ICU, I still feel strongly about sending patients to another ICU. The take-home points for me are that critically ill patients require higher levels of care and delays to receive this care can be harmful. If I feel that a patient requires ICU care, I consult the ICU for an admission even if they have no open beds.

Aaron Case is an EM/IM Resident, class of 2021

EM/IM Sessions are reviewed in journal club style by the current attendings and residents, as well as alumni of the UIC IM/EM program prior to publication. This post was specifically reviewed by Adam Rodos, MD, Assistant Director of the IM/EM residency program. Elspeth Pearce, MD, Editor.