Can this patient go to the observation unit?

Author: Matthew Wheatley, MD (Assistant Professor of Emergency Medicine, Medical Director of CDU, Emory University School of Medicine) // Edited by: Alex Koyfman, MD (@EMHighAK, EM Attending Physician, UTSW / Parkland Memorial Hospital) and Brit Long, MD (@long_brit, EM Chief Resident at SAUSHEC, USAF)

Screen Shot 2016-05-18 at 9.06.26 PM

For those of us who work in an ED with an observation unit, this is a question we get constantly from admitting teams. It can be frustrating to spend time working a patient up and reassessing them over a few hours in the course of a busy ED shift, having put in the thought that the patient would be best served on an inpatient team, only to have the admitting physician undercut your judgment, possibly without even seeing the patient.

So how can you decide? What tools or evidence is out there to help a busy ED provider decide who will do best in an observation unit vs on the floor? Let’s discuss some “do’s” and “don’ts” for patient selection for observation units.


DO: Know the existing admission criteria to your unit.

Not all obs units are the same. This is usually a function of the nursing level of service they are able to provide. Some units take patients who can’t walk at baseline; some will observe patients who are awaiting sobriety or on a psychiatric hold. It is a good idea to know what your unit can handle – the general inclusion and exclusion criteria. Also you should know how to access the information real-time during a shift – whether in a protocol binder, or available electronically.

What if your observation unit doesn’t have formal inclusion/exclusion criteria? No problem. It is a good idea to have them as it helps to select patients who will do well in observation. It may be worth meeting with the Observation Unit medical director or ED medical director to discuss formalizing something. Look at for an example of some general inclusion/exclusion criteria.


DO: Consider the Obs Unit for patients with a single condition that requires limited work-up.

The Center for Medicare and Medicaid Services (CMS) defines observation services as outpatient care rendered to determine the need for admission.1 This definition provides a framework for observation patient selection. Basically, it says observation is good for patients who are too sick to go home and need more therapy or testing to see if they need a full inpatient stay.

In the clinical environment, this can be an asthmatic patient who is not unstable, but needs a few more neb treatments prior to discharge; the patient with chest pain who needs provocative testing; the patient with a TIA who needs MRI and risk-factor modification.

Ross et al published a paper listing best practices for ED obs units.2 Two of the principles involve patient selection: focused patient care goals and limited duration and intensity of service. The paper encourages units to aim for an average length of stay of 15-18 hours with 70-80% of patients being discharged home. To put this another way, does the patient you are considering for the observation unit have a 70-80% probability of being discharged home in 15-18 hours? If yes, they are likely appropriate for observation.  On the other hand, the asthmatic with a history of multiple admissions for 3 or 4 days or the neuro patient who needs an MRI and multiple consults may be better suited on an inpatient team.


DO: Consider using the observation unit for patients off protocol.

 This may seem counterintuitive, but the observation unit is there to decompress the ED and take patients who would normally stay in the ED for prolonged work-ups or therapy.

First, a word about Protocol-driven Observation Units. A recent paper published in Health Affairs detailed the operational advantages of obs units where the care was delivered via protocols (Type 1 units).3 Protocols allow for more efficient, streamlined care that can be easily managed by a nurse or Advanced Practice Provider. Look at for examples.

That being said, not all patients fit cleanly into existing protocols. It may even be advantageous to think of these ‘protocols’ more like ‘guidelines.’ It is for this reason that my unit has a generic orderset. Provided the patients meet the general inclusions for the unit and are 70-80% likely to be discharged in 15-18 hours, they can come to the unit.

If you see common conditions admitted to your unit under the generic protocol, it may be an opportunity to create a formal protocol. It is through this process that we have created transfusion protocols for our observation units.


DON’T: Use the observation unit for prolonged work-ups.

Observation units are the ideal place for the active management of stable patients with straightforward problems. However, patients requiring multi-step evaluations are best served in other locations. An example would be a patient with headaches and neurologic symptoms who needs an MRI, EEG, LP, neurology, and ophthalmology consultations.

First step in these cases is to consider why the patient needs the work-up in the first place. If this is to facilitate an outpatient work-up, then it really doesn’t meet the criteria for observation (care provided to determine the need for admission). In this case, the patient’s disposition is set, so you’re really not performing observation services.

For patients in whom the disposition is unclear, or who have difficulty with outpatient follow-up, observation units are seen as an attractive option. However, consider the fact that each additional test and consultation adds time to the visit and will likely stretch beyond the 15-18 hours recommended. Such patients should be admitted if they truly can’t get their care as outpatients.


DON’T: Place patients in the obs unit with unclear endpoints.

There are 2 groups of patients that fall under this rubric: cases in which the ED physician does not know what is going on and patients with acute on chronic pain. An example of the former category is an elderly patient who has declined from his or her baseline but has a negative work-up in the ED. Again, observation units function best when they are allowed to actively manage patients to specific endpoints. Patients such as this can be difficult to “sell” to an admitting team, yet they are high risk for ED discharge. Unfortunately, they usually are admitted from the observation unit in high percentages. In addition, inpatient units often have better access to services such as Physical/Occupational Therapy, Social Services and psychiatry than observation units do. Bottom line: if you are unable to state the goals of observation in a single sentence, the patient should be admitted.

The second group of patients would be those with chronic pain/nausea/dizziness and similarly, they do poorly in observation. This can be another group of patients who are difficult to “sell” as an admission; however, if they are not able to go home after 3-4 hours of active management in the ED, then and additional 15 in the observation unit rarely improves their chances of discharge. This is particularly true of sickle cell vaso-occlusive crises. Looking at a patient’s historical performance is usually predictive of how long their admissions usually last.

The abdominal pain protocol is useful in cases where you are trying to rule out or in a specific condition. In a patient with RLQ tenderness, a period of observation and serial exams can be a great clinical tool in ruling in or out appendicitis. Similarly, patients with cholelithiasis can go to the obs unit for HIDA scan.


DON’T: Use the obs unit for patients who really should be admitted.

This may seem obvious, but there are many patients where an admitting/consulting service will want to use the observation unit as a stop gap. One such patient population are patients who will need sub-acute rehab or Skilled Nursing Facility (SNF) placement, such as elderly patients with falls and pain. Again, it’s tempting to place these patients in observation because they aren’t medically ill. First, in most cases, it takes greater than 15-18 hours to control these patients’ pain and get them back to functional status. Secondly, if the patient ends up qualifying for SNF placement, CMS will only cover the first 30 days of SNF care if they have had 3 days of inpatient hospital care (obs days do not count toward this).

The second group this applies to are patients who will need invasive procedures: GU stents, cardiac caths, going to the OR. It is not appropriate for a consultant to ask you to hold onto a patient until they can take them to the OR. This would be antithetical to the CMS definition of observation as they are not receiving anything to determine the need for inpatient admission.

Observation Units are excellent clinical tools that allow for efficient treatment and diagnosis of patients who are too sick to go home but not sick enough for admission. Appropriate patient selection is essential for them to operate as designed; however, this can be a gray area, prone to provider risk tolerance. Use of published data, best practices, and online resources can help reduce uncertainty as to which patients will be appropriate for obs units.


References / Further Reading:

  1. July 2009 Update of the Hospital Outpatient Prospective Payment System (OPPS), CMS Manual System, Pub 100-02 Medicare Beneifit Policy, Transmittal 10, Change Request 6492. Department of Health and Human Services (DHHS), Centers for Medicare & Medicaid Services (CMS). 2009.
  2. Ross MA, Graff L, Suri P, O’Malley R, Ojo A, Bohan S, Clark C. State of the Art: Emergency Department Observation Units. Critical pathways in cardiology 2012;11:128-38.
  3. Ross MA, Hockenberry JM, et al. “Protocol-Driven Emergency Department Observation Units Offer Savings, Shorter Stays, And Reduced Admissions.”  Health Aff 2013;32:2149-56.

2 thoughts on “Can this patient go to the observation unit?”

Leave a Reply

Your email address will not be published.