Estimating Observation Unit Profitability with Options Modeling


Christopher Baugh, MD, MBA; e-mail:


Background:  Over the past two decades, the use of observation units to treat such common conditions as chest pain, asthma, and others has greatly increased. These units allow patients to be directed out of emergency department (ED) acute care beds while potentially avoiding inpatient admission. Many studies have demonstrated the clinical effectiveness of care delivered in such a setting compared to the ED or inpatient ward. However, there are limited data published about observation unit finance.

Methods:  Using the economic principles of stock options, opportunity costs, and net present value (NPV), a model that captures the value generated by admitting a patient to an observation unit was derived. In addition, an appendix is included showing how this model can be used to calculate the dollar value of an observation unit admission.

Results:  A model is presented that captures more complexity of observation finance than the simple difference between payments and costs. The calculated estimate in the Appendix suggests that the average value of a single observation unit admission was about $2,908, which is about 40% higher than expected.

Conclusion:  Subtraction of costs from payments may significantly underestimate the financial value of an observation unit admission. However, the positive value generated by an observation unit bed must be considered in the context of other projects available to hospital administrators.

In the current health care finance environment, hospitals are increasingly pressured to operate near or at full capacity to remain fiscally sound. As a result, crowding in the emergency department (ED), in large part due to lack of inpatient bed availability,1 has emerged as one of the most serious problems facing physicians and patients in the ED today. Therefore, opportunities to more efficiently manage patients by directing patients out of the ED, but also away from the inpatient service, have the potential to mitigate crowding. Observation units can fill this need, but the role that they play in the finances of the hospital has not been well described previously. In this article, a unique analysis is presented that captures the complexity of this interaction to better inform hospital decision-makers charged with resource allocation about the economic viability of an observation unit at their institution.

A stock option is created when one purchases the right to buy a stock at a predetermined price during a predetermined period of time. This price is called the strike price. One pays a small fee for this right, which affords time to benefit from more information about the stock. At any time during this observation period, the investor decides either to exercise the option (i.e., buy the stock at the strike price because the market price is now above the strike price) or to allow the option to expire as worthless (i.e., pass on buying the stock because the market price is now below the strike price), thereby incurring a small loss.2 The general economic principle underlying this transaction is that by incurring the risk of a small loss, the opportunity for a large gain is purchased. Figure 1 illustrates this process.

Figure 1.

 Basic stock option framework. At Time 0, the price of the stock is P0. After time passes, the stock could either be in an “up” state, with a higher price (P1), or in a “down” state, with a lower price (P2).

The economic theory supporting options can quite readily be translated to apply to admissions to an observation unit. Buying an option and using an observation unit both afford the decision-maker more time to gather information and make a more accurate decision in the future (see Figure 2). After a stay in the observation unit (usually about 6–12 hours3), the clinician decides to either admit the patient to an inpatient ward or discharge the patient to home. Additional information is gained during the observation unit stay (e.g., vital signs, response to therapy, imaging results, consultant recommendations). The clinician then reassesses the patient, resulting in a second decision point; the clinician, like the investor, decides whether he or she would like to exercise his or her option for full admission to an inpatient ward. This process often results in more appropriate disposition decisions, thus allowing a higher proportion of ED patients to be discharged home and avoiding unnecessary inpatient admissions.

Figure 2.

 Observation unit option framework. C = option price; P0 = value generated when patient is immediately (time [t] = 0) admitted to inpatient ward; P1 = value created by discharging patient from the observation unit to home; P2 = cost associated with admission to hospital ward after observation unit stay. ED = emergency department; CPT = Current Procedural Terminology; Obs = observation; DRG = diagnosis-related group; inpt = inpatient.

Our analysis directly applies stock option theory to observation units, resulting in the derivation of a simplified formula that can more accurately reflect the profitability of an observation unit admission than by basic subtraction of costs from payments. To illustrate the use of this formula, an example is presented in the Appendix.


Study Design

This article is a financial analysis of public data, and institutional review board approval was not obtained.

Study Setting and Population

The study setting for the analysis is a capacity-constrained tertiary care hospital with an observation unit.

Study Protocol

Options have an economic value. This value can be calculated by finding the difference between the current and estimated future values of costs and payments (i.e., the net present value, or NPV). The estimated future value is found by multiplying the probability of achieving that value by the value itself for each future state and then adding them together.

In its most simple form, there are two states: an “up” state, where the object of interest gains value, and a “down” state, where it loses value. With respect to observation units, the up state (P1) is the value created by discharging an observation unit patient to home. These patients generate an extra source of payment, since the Centers for Medicare and Medicaid Services (CMS) model for payment involves a rate higher than the normal ED rate for this extended care. Discharged patients also minimize resource utilization and keep inpatient beds open for other patients, including transfer patients. The down state (P2) is admission to an inpatient ward after a stay in the observation unit. These patients use the same ED and observation unit resources as the patients in the up state, but then go on to use even more resources as an inpatient. Additionally, no ED or extended care payments are made, as a single diagnosis-related group (DRG) payment is applied to the entire encounter. Additionally an inpatient bed is occupied, creating an opportunity cost by blocking admission of other patients, often with higher-paying DRGs. The value (i.e., payments minus costs) generated by a patient who is not readily discharged from the ED at Time 0 (P0) is the value of immediate admission to an inpatient ward. The value of the option (C) is the NPV generated by sending the patient to the observation unit at Time 0 and thus the maximum price one is willing to pay to delay the disposition decision for the patient.

Because observation unit patients are admitted at an institution-specific predictable rate, one can reliably predict the likelihoods of the up and down states. Providers balance the priorities of maximizing both the utilization and the discharge rate of the observation unit, and several sources point to an industry standard discharge rate of about 80% to achieve this balance.3–6 This generalization is used for the purposes of calculating an estimate, but an institution-specific value should be substituted if it varies significantly from this assumption.

Data Analysis

Because the “pricing” in the disposition of patients is generally not accurate, we cannot assume that P0 is simply a reflection of the value of P1 and P2 and the probabilities of reaching those two states, as in stock pricing. As a result, by putting a patient into an observation unit and creating the option, we create value because (0.8*P1 + 0.2*P2) > P0. Quantifying this difference is the key to finding out the true value of utilizing an observation unit, and the remainder of this analysis will focus on this determination.


Further algebra can be used to simplify the above equation and reveals some important findings (see Figure 3). This final formula exposes several important characteristics of observation medicine finance. First, the costs of the ED stay drop out. This makes sense because the ED costs are effectively a “sunk cost” (i.e., costs are incurred prior to making the decision to put the patient into the observation unit). Thus the cost of the ED visit is irrelevant to the determination if the observation unit use creates value. Next, note that we must account for the entire cost of the observation unit stay. This is an obvious finding; to benefit from an observation unit, it must be used. Also note the “−0.8*(DRG payment – cost of inpatient stay)” term in the final formula from Figure 3. This is the term that captures the value of avoiding inpatient hospitalization for those diagnoses where hospital costs exceed DRG payments. It is well known that many DRG payments do not cover hospital costs; as a result, it is unprofitable for hospitals to deliver care to these patients. If these patients can be effectively managed without an inpatient admission, value is created. On the other hand, if the hospital profits from an admission (DRG payment exceeds hospital costs), using an observation unit to divert these patients away from an inpatient ward would potentially destroy value. Finally, the figure “0.8*open-bed value” captures the value created by not tying up an inpatient bed with a patient who could have been discharged to home after an observation unit stay. This value is important to keep in mind, especially for hospitals near capacity, since patients with more favorable payment-to-cost ratios (e.g., many transfer patients) can fill the open inpatient beds created by use of an observation unit. Of note, the multiplier of 0.8 in front of these elements reflects the discharge to home rate, in this case 80% (any other rate can be easily substituted if necessary for an institution-specific calculation).

Figure 3.

 Derivation of final model. C = option price; P0 = value generated when patient is immediately (time [t] = 0) admitted to inpatient ward; P1 = value created by discharging patient from the ED to home; P2 = cost associated with admission to hospital ward after observation unit stay. ED = emergency department; CPT = current procedural terminology; Obs = observation; DRG = diagnosis-related group; inpt = inpatient.


Observation units divert patients requiring a prolonged ED stay away from inpatient care, thus opening up capacity both in the ED and on inpatient wards to care for more severely ill patients, usually ones that require more resource-intensive care (thus resulting in higher DRG payments on average). This creates a higher case-mix index for the hospital, shifting to higher DRG payments. In addition, both ED and extended care payments are captured for those patients treated and discharged home from the observation unit. Another aspect of inpatient reimbursement that further increases the value of the observation admission is that inpatient claims are usually denied at a higher rate than observation claims. This further reduces the revenue associated with inpatient hospitalizations. As a result, the hospital captures more revenue from both the ED and the inpatient wards by treating appropriate patients in the observation setting.

Over the past decade, the overall Medicare payment-to-cost ratio has declined to below zero for many diagnoses, but also remains higher than this ratio for Medicaid.7 This ratio for private insurers remains over zero (if not, all hospitals would go out of business). Thus, the importance of redirecting (when medically appropriate) those patients whose DRG payments cover less of the hospital costs than others toward an observation unit becomes clear. The disparate mix of profits to be had from Medicare DRG payments makes patient selection for treatment in an observation unit rather than an inpatient ward an interesting opportunity for certain hospitals to capture higher profits. Observation medicine has proven an effective tool in treating such disease processes as chest pain, asthma, congestive heart failure, and many others.3 High levels of patient satisfaction, comparable quality of care to inpatient admission, lower costs, and lower lengths of stay for specific patient populations cared for in observation units have also been well documented.8–13

MedPAC has shown that about one-third of Medicare DRG payments result in hospital losses (payment-to-cost ratio less than 1), one-third approximate costs, and about one-third exceed costs.14 However, the current DRG weighting system is based heavily on hospital charges, not costs. As a result, those patients who require more hospital resources that are susceptible to inflated charges end up with more profitable DRG payments. These patients have historically fallen into the category of surgical (e.g., cardiac and orthopedic); as a result, the earlier part of this decade witnessed the rise of the “specialty hospital,” wherein patients with more profitable DRG payments were targeted. New CMS proposals are now targeting hospital costs, not charges, as the metric upon which to base DRG payments, which should result in a reshuffling of DRG payment profitability and likely cause a great majority of payments to more closely approximate hospital costs.14 As a result, shifts in the profitability of diagnoses may impact which patients are most cost-effectively managed in an observation unit in the future.

This analysis shows that if the NPV of an observation unit bed is positive, hospital management should consider either expanding an existing observation unit or starting one if none already exists in the context of other positive NPV projects that are available for funding. Additionally, there is a limit to the profitability of an observation unit. If a hospital sees too few patients to efficiently utilize an observation unit, then suboptimal observation unit utilization, and thus more limited revenue, will not make up for the substantial fixed costs of setting aside a bed for observation. Also, in hospitals that already employ an observation unit, if expansion beyond what is appropriate for that particular facility’s patient volume and case-mix index occurs, then a higher percentage of patients will need admission to an inpatient ward after an observation unit stay, resulting in higher resource use without higher compensation, thus lowering profitability.

This analysis also illustrates that patients treated in an observation unit can create a “win-win” scenario. Patients receive care at least as good as on the inpatient service and are more satisfied.10,11 The hospital is able to use inpatient beds more efficiently and treat more profitable patients. Insurers pay for ED and extended-stay visits instead of inpatient hospitalization, thus saving money. However, the major caveat to this assumption is that the observation unit is being efficiently managed. This means that the number of observation beds is appropriate for the size and health characteristics of the patient population it serves. In addition, the emergency physicians must choose appropriate patients for observation to achieve a discharge rate that minimizes inpatient admissions and maximizes utilization. A discussion of optimal patient selection for observation unit admission will not be presented here, but is a topic worthy of further analysis.15,16


Several proxies were used to determine the final value for an observation unit patient in the Appendix. Each of these proxies does not fully capture the complexity of the inputs and outputs required to determine an exact value. As a result, one must interpret the analysis as a best estimate, not a precise calculation. The assumptions and limitations of the model must be carefully considered before applying this estimate to any specific institution. For example, the inpatient revenue was calculated using Medicare DRG payments. While nearly all U.S. hospitals receive payments from CMS for their Medicare patients, the specific payer mix for any particular hospital may greatly vary. As a result, hospitals receive a range of different reimbursements from patients with Medicare, Medicaid, private insurance, and no insurance for the same types of inpatient hospitalizations (and ED or observation unit stays).

This analysis does not consider the start-up costs of an observation unit in the final calculation. Construction costs vary substantially based on number of beds, room size, local labor and material costs, and many other factors. In addition, accounting for these costs can be done in several ways, such as an amortization table wherein the costs are divided up over a predetermined number of years and counted toward the costs of each individual patient over that time period. Also, many hospitals create observation units by converting existing space for this use, thus minimizing construction costs. Of note, the resources required for observation beds are much less than those required for acute care or inpatient beds (e.g., less space per bed, more patients per nurse). Nevertheless, this cost obviously needs to be included in the NPV analysis of any observation unit project under consideration.

In addition, by creating an observation unit, valuable space is taken away from other potential uses (e.g., more acute care ED beds, more inpatient beds). As a result, the value created by use of an observation unit must be directly compared to the value created by these other uses prior to proclaiming that the optimal use of such a scarce resource is by expanding or creating an observation unit. The values of these alternative uses are not presented in this article.

In addition, patients other than just ED transfers may be turned away from a tertiary care hospital due to lack of inpatient beds, such as direct transfers and patients scheduled for elective surgery. Since these patients also have potential to generate profits for a hospital, they could also be considered an opportunity cost of filling an inpatient bed with a patient who could have been managed an in observation unit and then sent home.

Finally, this framework is intended to be applied to a tertiary care hospital that is capacity-constrained. Since the loss of transfer patient plays a significant role as an opportunity cost, a hospital must regularly receive transfer patients for this entire model to be applicable. In addition, if inpatient space is available for use and empty, then a hospital may find that sending patients to those beds instead of an observation unit better utilizes available resources. However, if bed capacity is constrained, then the efficiency afforded by an observation unit would be a great asset.


By using the analogy of a stock option, a more robust financial analysis of observation unit profitability can be made. Options modeling reveals that the basic subtraction of observation unit costs from revenues may underestimate true profitability by as much as 40% (Appendix). Few analyses of observation unit profitability and how it interacts in the complex financial system of the hospital have been previously performed.


To illustrate an example of how the formula derived in this article can be used, revenue and costs for both observation unit use and inpatient bed use were estimated for a select group of diagnoses. Specifically, the 10 most common conditions treated in observation units17 were investigated and contrasted against the 10 most common conditions seen in emergency department (ED) transfer patients18–20 (Tables 1 and 2, respectively).

Table 1.   Most Common Observation Unit Diagnoses17
  1. Source: Mace et al.17

 1. Chest pain/cardiac
 2. Gastrointestinal/abdominal
 3. Asthma/respiratory
 4. General ailments/medical
 5. Dehydration
 6. Psychiatric/social services
 7. Syncope/near syncope
 8. Congestive heart failure
 9. Head injuries
10. Kidney problems
Table 2.   Most Common Emergency Department Transfer Patient Diagnoses*18–20
  1. Sources: Gray et al.,18 Rourke and Kennard,19 Papson et al.20

  2. COPD = chronic obstructive pulmonary disease; AAA = abdominal aortic aneurysm.

  3. *Excludes ST-segment elevation myocardial infarction, cerebrovascular accident, intracranial hemorrhage, major trauma, and burns.

 1. Post–cardiopulmonary arrest
 2. Overdose
 3. Acute respiratory failure (pneumonia, COPD, asthma)
 4. Abdominal catastrophe (AAA, appendicitis, etc.)
 5. Seizures/epilepsy
 6. Gastrointestinal bleeding
 7. Aortic dissection
 8. Smoke inhalation
 9. Fractures (not major trauma)
10. Sepsis

Those diagnoses that are usually accepted for transfer by tertiary care hospitals at peak capacity are also acknowledged and excluded in this analysis. These most often include ST-segment elevation myocardial infarction (STEMI), major trauma, burns, intracranial hemorrhage (ICH), and stroke (CVA). Because these patients may potentially be transferred regardless of ED or inpatient capacity, the presence of an observation unit will likely not affect their acceptance. In addition, these patients will quite often require an intensive care unit (ICU) bed, which is not the same destination as the patients under consideration for an observation unit admission. However, the next set of patients, those requiring a higher level of care than what their current medical setting can provide, but are also not among those most commonly accepted for transfer noted above, are at risk of being diverted away from the hospital due to ED or inpatient capacity constraints. These patients were factored into the “opportunity cost” analysis.

In this analysis, Medicare diagnosis-related group (DRG) payments were used for these diagnoses as a proxy for revenue. Medicare payments are an ideal proxy because they are accurate, available, and generalizable. After determining the most common conditions for both observation and transfer patients, these conditions were matched with the most likely resulting DRG codes. Using the 2005 Centers for Medicare and Medicaid Services (CMS) Medicare Provider Analysis and Review (MedPar) data, the average national DRG payment amounts were determined ($3,210; see Table 3). Another assumption made in this article was the link between category of complaint and final DRG. In reality, the DRG is paid after a claim is received, and this claim is created after the hospitalization. As a result, the patient’s final diagnosis is known when the DRG is assigned. The patients being transferred or admitted can be categorized by chief complaint, and although this will generally correlate with a DRG code, it is impossible to prospectively link the two together with complete accuracy.

Table 3.   Inpatient Revenue and Cost Summary
ConditionDiagnosis-related Group (DRG)DRG Payment ($)Hospital Cost ($)Average LOS (Days)Net Profit (Loss)Net Profit (Loss) per Hospital-day
  1. GI = gastrointestinal; LOS = length of stay.

  2. *Used in the final calculation.

 Chest pain/cardiac140/1432,410.643,553.302.25(1,142.66)(507.85)
 General ailments/ medical278/4213,192.284,370.154.00(1,177.87)(294.47)
 Psychiatric/social services425/430/4324,726.774,909.136.77(182.36)(26.95)
 Syncope/near syncope1422,772.345,070.732.50(2,298.39)(919.36)
 Congestive heart failure1275,376.356,740.415.10(1,364.06)(267.46)
 Head injuries29/323,035.094,798.482.85(1,763.39)(618.73)
 Kidney problems321/326/3322,551.283,728.883.10(1,177.60)(379.87)
 Average 3,210.25*4,487.55*3.62*(1,277.30)(416.09)
 Post–cardiopulmonary arrest129/1446,283.295,565.644.15717.65172.93
 Acute respiratory failure87/99/101/4759,102.397,494.396.101,608.00263.61
 Abdominal catastrophe154/164/165/170/47814,889.2012,047.348.442,841.86336.71
 GI bleeding1745,216.154,802.674.70413.4887.89
 Aortic dissection104/110/47829,157.6722,866.039.836,291.64639.83
 Smoke inhalation449/4503,168.603,136.222.8532.3811.36
 Fractures (not major trauma)211/219/224/224,531.124,620.583.05(89.46)(29.33)
 Average 8,878.727,583.535.341,295.20166.97*

In addition, the 2005 CMS MedPar data also report the national average hospital charges by DRG. To determine the average cost per DRG, these charges were multiplied by a cost-to-charge ratio (CCR). For this analysis, CCRs of 0.35 and 0.25 were estimated for observation and transfer patients, respectively. The CCR for observation unit patients was based on the assumption that on average, a congestive heart failure (CHF) admission results in a loss of $1,288.21 Given the Medicare DRG payment for CHF and this amount of loss, a CCR of 0.35 was calculated. This value was then generalized to apply to all of the observation unit diagnoses, resulting in an average cost of inpatient stay of $4,488 for these patients (see Table 3). For the transfer patients, a lower CCR, on average, was assumed to be more accurate, given that they are more likely to require services with lower CCR, such as operating room services and anesthesia (see Table 4).

Table 4.   Centers for Medicare and Medicaid Services (CMS) Estimation of Cost-to-charge Ratios by Service
  1. Source: CMS Web site, 2005 data.

Routine days0.56
Intensive days0.50
Labor and delivery0.46
Other services0.38
Inhalation therapy0.20
Operating room0.19

Finally, length of inpatient stay was factored into the final value of the opportunity cost. Mean inpatient hospital length of stay (LOS) was also listed by DRG in the 2005 MedPar data file. For each of the observation and transfer diagnoses in Tables 1 and 2, calculations of the average daily profit (or loss) were made using the CCRs assumed above and are listed in Table 3. Since transfer patients typically stay in the hospital longer than observation patients, the analysis needs to account for this extra resource utilization (i.e., you do not simply “trade” an observation patient for a transfer patient). This adjustment was made by calculating the profit (or loss) per inpatient day for transfer patients. This amount was then multiplied by the average inpatient LOS for an observation unit patient if they are assumed to have been admitted to an inpatient ward instead. As a result, this correction results in a more accurate determination of the true opportunity cost of admitting a patient to an inpatient ward instead of an observation unit. The average inpatient LOS for observation unit diagnoses was 3.62 days, and the average profit per day for transfer patients was $167. As a result, the opportunity cost avoided by utilizing the observation unit and keeping an inpatient bed open was estimated at $604.

To determine the costs and revenues of an observation unit stay, the available literature regarding this topic was surveyed. Recent and useful articles outlining the costs for such common observation diagnoses as chest pain, asthma, head injury, and transient ischemia attack were found8,9,22–25 (Table 5). The value of these costs were adjusted to 2005 dollars and averaged, using this mean as a proxy for average observation unit costs ($1,138). Finally, the revenue collected by the average observation patient was estimated by the average payment at Brigham and Women’s Hospital in 2005 dollars across all payers for all observation unit patients ($3,175).

Table 5.   Costs Associated with Select Observation Diagnoses
DiagnosisCost ($)*Reference
  1. *In 2005 U.S. dollars.

Chest pain/cardiac1,066.21Jagminas8
Head injuries882.93Norlund24
Transient ischemic attack864.00Ross25


Using the formula derived in the main body of this article and inputs described above, the value of a patient treated in an observation unit was determined to be $2,908 (see Figure 4). This value is substantially higher (about 40%) than the basic subtraction of observation unit costs from payments, which equals $2,037. The contributions of the avoidance of losses on inpatient admissions and the opportunity costs saved by keeping inpatient beds open explain a great deal of this discrepancy and are illustrated in Figure 5. Additionally, this analysis can also be applied to smaller, non–tertiary care hospitals, but these opportunity costs would not be of benefit since patients are rarely transferred to these institutions. In these cases, the benefits of an observation unit may still be quite positive, as the more efficient care of patients eligible for a observation will likely yield cost savings.

Figure 4.

 Estimation of observation patient value.

Figure 5.

 Contribution to observation patient value.

This model was subjected to a sensitivity analysis wherein each of the previously stated assumptions was varied across a plausible range while others were held constant to determine the effect on the conclusions. The model did appear to be especially sensitive to the CCR assumed for both observation and transfer patients as well as the rate of discharge to home from the observation unit. Table 6 illustrates the range of values obtained for the final estimated observation admission value as the CCR was varied. This demonstrates that as the CCR rises for transfer patients, the open-bed value term becomes negative, reducing overall value (i.e., transfer patients become unprofitable). Also, as the CCR for observation unit patients decreases, the overall value also decreases, this time because the hospital is foregoing profits by not admitting these patients to an inpatient ward. As a result of this analysis, the most accurate calculation would require an institution-specific CCR for each of the most common diagnoses seen in observation units and in transfer patients declined transfer due to inpatient capacity constraints. In addition, the frequency of each of these diagnoses should be weighted in the calculation, rather than a simple average, which was used in this calculation.

Table 6.   Sensitivity Analysis Estimated Observation Patient Value (or Loss) Varied by Cost-to-charge Ratios (CCR)
Transfer PatientsObservation Unit Patients*0.4
  1. *CCRs of 0.35 and 0.25 were used in the final estimate for observation and transfer patients, respectively.


The rate of discharge to home from the observation unit was also varied, and the final value increased to $3,413 and decreased to $2,402 from the base case of $2,908 when the rate was changed from 80% to 90 and 70%, respectively. This suggests that a higher discharge rate is more profitable; however, industry experience shows that a rate much higher than 80% may be difficult to achieve.3 Finally, the transfer diagnosis of aortic dissection appeared to be an outlier in the calculated amount of profit per day. Removing this diagnosis reduces the open-bed value from $604 to $414.