Share to Digg Share to Facebook Share to Twitter Email this Print this

Gaining Focus

Managing Hospital Beds

Research by Thomas Best, Donald D. Eisenstein, David O. Meltzer and Burhaneddin Sandıkçı

Thomas Best is a PhD student at the University of Chicago Booth School of Business.

Donald D. Eisenstein is professor of operations management at the University of Chicago Booth School of Business.

David O. Meltzer, MD, is associate professor in the department of medicine, and associated faculty member in the department of economics and the Harris School of Public Policy at the University of Chicago.

Burhaneddin Sandıkçı is assistant professor of operations management at the University of Chicago Booth School of Business.
Gaining Focus

Using dedicated wings allows a hospital to achieve institutional objectives under strained bed capacity.

O

ver the last two decades, the demand for hospital services in the United States has been on the rise. Inpatient admissions have increased by 14 percent and emergency department (ED) visits by 40 percent, while the number of staffed hospital beds has dropped by 15 percent.

The pressure for medical services under limited bed capacity is especially felt by urban teaching hospitals such as the University of Chicago Medical Center (UCMC), which is located in a mostly poor and underserved area. UCMC struggles to serve a community that has limited options other than going to the emergency room for routine medical care. Many such patients, when admitted to the hospital, receive general medicine (GM) services, one of the 25 patient care services offered by UCMC. At times, patients admitted into the GM service occupied nearly one-third of the 295 adult care beds available in the hospital. Furthermore, due to the 1986 Emergency Medical Treatment and Labor Act, patients who present themselves to the ED and require admittance must be admitted if the hospital has the capacity and capability to treat them.

This is not just an issue of revenues for the hospital— it certainly is a financial burden—but more importantly it is a strain on UCMC's ability to care for some of the most complex and critically ill patients throughout the country. UCMC is a premier research and teaching hospital with expertise in such areas as advanced cancer and cardiology treatments. For some of these patients, UCMC may be one of few options for the care they require. Furthermore, the mission of UCMC as a research and teaching hospital calls for a level of inquiry into complex medical cases, and its accreditation as a teaching hospital requires it.

UCMC faces difficult decisions on how to best utilize its limited capacity, which has spurred passionate debate. The hospital has a commitment to the local community in which it resides but also to the most critically ill and complex patients. At the same, it must continue to be a financially viable premier research and teaching institution.

To help address some of these issues, UCMC received special dispensation from the state of Illinois in 2006 to divide its medical and surgical adult beds into mini-hospitals or wings. A wing is defined by a fixed set of specific patient diagnoses and a fixed allocation of beds. In practical terms, each wing will house a subset of UCMC's 25 patient care services. A particular patient can only be admitted if a bed is available in the wing that provides the medical service the patient requires.

UCMC formed four wings: general medicine, cancer treatment, cardiology, and multi-specialty care for the remaining medical services. A number of beds is specified for each wing, but the beds need not be physically isolated from another wing. However, the number of patients admitted to a wing at any given time must not exceed the agreed-upon limit as stipulated in the agreement with the state. The GM wing was initially allocated 69 beds, but at the time of this writing, it is allocated 35 beds. Such changes to the formation of wings can, in practice, only be made about once per year.

The effective reduction in the number of available GM beds was implemented along with an agreement with Mercy Hospital and Medical Center, a community hospital located near UCMC. Under the agreement, patients presenting themselves to the ED who need admittance for routine care are given the option of being admitted to Mercy, where they can typically get a bed immediately. Furthermore, patients transferred to Mercy would still receive care from UCMC physicians who would begin to divide their clinical time at Mercy. Mercy benefits from this arrangement because its cost structure is much lower—it can make a profit on the same patient that UCMC cannot. UCMC has a significantly higher cost structure due to its advanced technology, teaching, and research costs. Proponents of the arrangement view this as a win-win for both the hospitals and patients. Opponents of the arrangement view this as "patient-dumping."

Analyzing such wing formation and hospital collaboration spawned the interest of researchers Thomas Best, a Chicago Booth PhD student, and University of Chicago professors Donald D. Eisenstein, David O. Meltzer, and Burhaneddin Sandıkçı in a recent study titled, "Efficient Management of Strained Inpatient Bed Capacity."

The interest of the researchers spans a number of issues. One is to understand how a hospital like UCMC should form wings; that is, how many wings should be formed, which medical services should be assigned to each wing, and how many beds should be allocated to each wing. The researchers also sought to understand if forming wings resulted in any advantages due to "operational focus." It is well understood that when focusing an operation, including medical services, greater efficiency and improved quality can result. For example, a medical facility dedicated to cancer treatment can attain improvements both in efficiency and quality by focusing its resources on a narrow set of treatments. Forming wings at UCMC is one potential way to attain some operational focus.

Finally, this initial work leads to the more general issue of how to plan care across a community of hospitals. Forming collaborations across health care facilities to take advantage of focused care in some places and multi-specialty care in others, can improve patient care and access to medical services as well as overall efficiencies.

Operational Focus

After UCMC formed wings, evidence emerged that new efficiencies were being realized. In particular, the average length-of-stay (LOS) of patients admitted to the hospital dropped significantly after the wings were formed, an effect which could not be attributed to a change in patient mix or improvements in technology.

The researchers discussed the data with the physician who acted as the patient flow director for one of the wings. He described how his job evolved after the wings were formed. He started daily meetings with case managers, nurses, and doctors within the wing. They would discuss each patient and ask the question "Why is Mr. Jones still in the hospital?" Case workers could focus on the types of discharge needs typical for the wing in question. And very importantly, the stakeholders of a wing began to take ownership of their beds. As Eisenstein explained, "It was now clear to all that turning over a bed resulted in capacity for their patients."

The researchers also collected data that showed the turnover of beds was greatest when the wing was almost full. As occupancy in the wing neared 100 percent, the average LOS decreased.

How to Form Wings

The authors formulated a model that can help hospitals decide how many wings to form, the number of beds in each wing, and the type of services that should be assigned to each wing. The model seeks to maximize a function of bed utilization. That is, given a formation of wings, each wing is evaluated in terms of this function, which has two components: the average number of beds occupied in the wing and the average value of a patient-day within the wing. The average number of beds occupied incorporates a model that considers the random arrival of patients for service.

The model is very flexible in terms of setting a weighted value of a patient admitted to a wing. It can set the weight in terms of profitability. Alternatively, it can set a weight in terms of the efficiency of care. In other words, how equipped is this facility to care for this type of patient? Or, it can set a weight in terms of the complexity of care to address how well a patient matches the core competency of the facility. The initial test runs of the model, as applied to UCMC, use a weight that is a combination of all of these attributes. For each type of patient, a national measure called the DRG (Diagnostic Related Group) relative weight measures the cost and complexity of care of patients. The larger the value, the more difficult or complex a patient is to care for, and thus typically more profitable.

To estimate the average LOS for a patient, the researchers used a function that decreased the estimated LOS for a proposed wing as demand approached and exceeded the available capacity. The model uses both UCMC data as well as national databases to estimate the relative demand for each type of patient. The model considers over 500 different types of patients to estimate their relative demand and value. These types are then aggregated into 25 different medical services. For a hospital the size of UCMC, there are numerous ways to form wings, so the researchers used advanced optimization techniques that were able to find solutions that they could prove were effectively optimal.

An example of the model's output using data for UCMC is as follows. If the total demand for hospital beds per day is just 67 percent of the bed capacity, then the optimal solution is to keep all 25 medical services together in one wing. That is, do not ask for special dispensation to subdivide the hospital. The reason is that when demand for beds is relatively low, then it is best to pool the demand variances of the various medical services in one large pool of bed capacity. The higher-weighted services are in little danger of being crowded out by lower-weighted services.

But as the total demand for beds increases, it begins to become more advantageous for the hospital to form wings. There will be days when the hospital experiences more demand than capacity and days when it has slack capacity. Consider, for example, when demand is on average equal to the hospital's capacity. In this case, the best solution is to form two wings: one for the top 18 higher-weighted services and one for the next six lower-weighted services. The medical service with the lowest-weighted value is allocated no beds. This indicates that the hospital is not well suited for this type of medical service and may wish to explore facilities that are better suited to meet the demand for this service. However, it is also possible to add constraints to the model that sets a minimum number of beds available for each service.

More wings are formed and more medical services are allocated zero beds when overall demand is 33 percent higher on average than the bed capacity of the hospital. The researchers believe that UCMC is close to this case. Its collaboration with Mercy, which has spare bed capacity, is critical to manage demand when it exceeds the capacity of the hospital. Furthermore, the model predicts that forming wings can actually increase the total number of patients a facility is able to accommodate. "This is because our model incorporates reduced length-of-stays as demand increases and wings narrow their focus of care," remarks Sandıkçı.

If other hospitals manage their beds in a similar way, then one can imagine a network of hospitals forming partnerships across a community. Each one would focus on the medical services that they do best. This "global" approach may be a better solution to a hospital's limited capacity than expecting a hospital like UCMC to be everything to everybody. "It is impractical to think that every facility can function as a general hospital," says Eisenstein. "We should take advantage of focused care where appropriate."

"Efficient Management of Strained Inpatient Bed Capacity." Thomas Best, Donald D. Eisenstein, David O. Meltzer and Burhaneddin Sandıkçı.