When a new surgical method is shown to produce outcomes similar to those of a more established procedure, finance leaders should estimate the opportunity cost associated with the new approach to decide whether it merits adoption.
At a Glance
- Although robotic hysterectomy does not produce significantly better outcomes than laparoscopic hysterectomy, hospitals may feel pressure from patients and clinicians to use the robotic procedure.
- Hospitals that opt for robotic hysterectomy over laparoscopic hysterectomy face not only higher variable costs, but also an opportunity cost in the form of lost surgical capacity.
- Estimating the opportunity cost of performing robotic hysterectomy provides crucial data for hospital executives in deciding whether to invest in the procedure.
Hospital executives often face pressure from physicians to invest in technologies that promise to deliver new benefits to patients. The physicians might even argue that a new technology could improve the hospital's profitability. However, physicians' supporting data may be limited, and the executives may end up being swayed by the perception that the hospital will lose patients to competitors if it does not make the investment.
Robotically assisted hysterectomy (RH) provides an apt example. With robotic assistance already well-established by procedures such as prostate surgery, pressure may be rising to use the robotic approach in other common procedures, including hysterectomy.
What advice can finance leaders offer hospital executives regarding this decision, taking into account that the procedure long has been performed laparoscopically—without the assistance of a robot—in many institutions?
In 2013, Jason D. Wright, MD, and colleagues published a study based on more than 250,000 hysterectomies for benign gynecologic disease. The researchers found that outcomes achieved using RH were not significantly better than those achieved using laparoscopic hysterectomy (LH), and the variable cost of performing RH was about $ 1,200 higher than that of LH.a The study also documented the rapid growth in the number of RH procedures, particularly in hospitals that already performed other robotic surgeries.
In an editorial accompanying the study, Joel S. Weissman, PhD, and Michael Zinner, MD, highlighted the role of direct-to-consumer advertising, which, in their opinion, may "only fuel unnecessary utilization" of robotically assisted surgery. (Their opinion recently gained support from a study on the impact of marketing language on patient preferences for robotically assisted surgery.b) They also pointed out that payments for laparoscopic surgery are the same with or without robotic assistance. Thus, they asserted, neither patients nor physicians have any readily apparent or tangible incentive to use the less expensive method.
Weissman and Zinner proposed policies to redress the balance in instances when a new technology is no more effective and is more expensive than the current standard. If the demand for RH is patient-driven, they suggest a larger copayment—in effect, a premium—for this procedure. Based on the data in the Wright study and the fact that payments for RH and LH are the same, the initial estimate of this premium would be the difference in variable cost between the two procedures: about $1,200. If hospitals and surgeons are driving the demand, the authors proposed asking them to justify use of the more expensive technology, presumably after making them aware of the premium.
However, a premium equal to the difference in variable costs is too low because it fails to account for the longer time required to perform an RH procedure. Opting for RH imposes an opportunity cost on the hospital because the facility's surgical capacity decreases.
Here, we show how to calculate an appropriate premium for RH that reflects the opportunity cost by taking into account differences in surgical times.
The Concept of the Calculation
Our calculation of the premium for RH procedures is based on the assumption that the hospital has already purchased a surgical robot for other procedures and is extending its application to hysterectomies, many of which are performed laparoscopically.
For the sake of convenience, we assume that the hospital has dedicated one operating room (OR) to either LH or RH. We estimate the number of LH or RH procedures that can be performed in one month, then calculate the difference in the monthly cash flows generated by each option. The premium is determined by dividing the difference in monthly cash flows by the number of RH procedures performed per month. (We ignore the initial investment because it represents a sunk cost.) The difference in the hospital's monthly cash flows is given by the following formula (where DRG refers to the payment received for either procedure):
[Number of LH procedures performed per month × (DRG – variable cost for LH)] – [Number of RH procedures performed per month × (DRG – variable cost for RH)]
To calculate this quantity, the values for the DRG, variable costs, and the number of procedures of each type performed monthly are needed. If the number of RH procedures equals the number of LH procedures, the difference in monthly cash flows reduces to the monthly difference in variable costs. Dividing the result by the number of procedures, we arrive at the difference in variable costs per procedure, which we refer to as the "reference premium." Although Wright, et al., suggest the reference premium is $1,200, the difference in variable costs varies by hospital, reflecting hospital practices and vendor contract terms.
Now, as OR managers know—and as OR case volumes demonstrate—RH procedures take longer than LH procedures, meaning fewer RH procedures can be performed in a given time interval. It is evident that, as the number of possible LH procedures that can be performed relative to RH procedures increases, the premium grows. Further analysis of the expression also reveals that the premium increases as the ratio of variable costs for RH to DRG decreases.
Performing the Calculation
When assessing the financial impact of replacing LH with RH, the biggest challenge lies in estimating the number of LH and RH procedures that can be performed monthly. To estimate these numbers, we built a simulation model. The model assumes that the hospital dedicates one OR daily to either LH or RH. The OR uses a nominal 10-hour workday with up to one hour of overtime, allowing for 11 hours of use per day. (See sidebar below.)
The input data for the computer simulation were based on the operative times of expert surgeons plus a 30-minute turnover time. The simulation found that in 22 days—about one working month—the hospital can perform either 59 LH procedures or 43 RH procedures. This difference equates to a ratio of approximately 1.4 LH procedures for every RH procedure that can be performed.
Finance leaders who have access to simulation can build their own models using local parameters to estimate the number of procedures of each type that can be performed in a month. Alternatively, they can make educated guesses based on prior laparoscopic-to-robotic conversions, taking into account reasonable estimates of the added time when LH is replaced by RH.
We used our simulation results to select the range of ratios of procedures (number of LH procedures to number of RH procedures) that can be performed in the available OR time for use in the accompanying exhibits. The figures show the ratio of the actual premium to the reference premium (with the difference between those two values representing the opportunity cost) as a function of DRG and for a range of ratios of RH variable cost to DRG. Each exhibit is based on a different reference premium: $800, $1,200, and $1,600.
As shown, the ratio of the actual premium to the reference premium—with the reference premium equaling the difference in variable costs—grows as the reference premium decreases and as the ratio of the number of LH to RH procedures increases. This finding was expected: As the reference premium decreases, the actual premium is dominated by the opportunity cost associated with the displacement of LH by RH.
Also, the greater the number of potential LH procedures that cannot be performed due to scheduled RH procedures, the greater the opportunity cost associated with performing RH and, thus, the larger the premium required to recoup that cost. Furthermore, as the ratio of RH variable cost to DRG decreases, the difference in cash flows increases and so does the actual premium relative to the reference premium.
Consider a situation characterized by a DRG of $8,000 and a reference premium of $1,200, as shown in the exhibit above. When the ratio of LH to RH procedures is 1.4, as in our simulation, the ratio of the premiums ranges from 3.8 when the variable cost of RH is 10 percent of DRG to 3.0 when the variable cost is 40 percent of DRG.
In this scenario, RH costs the hospital between $3,600 and $4,560 more than LH—or roughly half the DRG payment—to perform a case for which LH would achieve an equivalent outcome. This is the amount of the premium the hospital theoretically would have to demand to break even.
Putting the Results in Perspective
Consider the annual financial impact on a hospital of performing RH in the dedicated OR to the complete exclusion of LH. Let us assume:
- DRG is $8,000
- RH variable costs are 20 percent of DRG
- 1.4 LH procedures can be performed for every RH procedure
Under these assumptions, the annual forgone cash flow ranges from $1.9 million to $2.5 million (12 months × 43 RH procedures per month × ratio of premiums [from the exhibits above and below] × reference premium). This amount is comparable to the investment associated with a robotic platform. If RH does not fully displace LH—which is a more realistic assumption—the estimate of the actual premium would be reduced using different case-count scenarios. (If the hospital were to fully dedicate a robot to RH, of course, the capital expenditure would have to be included in the analysis, and the actual premium would rise.)
The decision to promote the increased use of RH relative to LH should be weighed by comparing the forgone cash flow with potential benefits that may be identified and quantified. For the example above, such added benefits would need to be at least half of the DRG for each minimally invasive hysterectomy to make robotic replacement clearly advantageous.
Increasing OR hours would be a possible reaction to our observation that fewer RH procedures can be performed in the time allocated for the current volume of LH procedures. However, because 10 hours is a common length for an OR day, adding time would mean dedicating time on other days or in additional ORs. Such maneuvers increase variable costs, and our calculation of the actual premium versus the reference premium is easily replicated taking into account these new conditions.
Another point involves claims made for length of stay when robotically assisted procedures are considered. Replacement of a major open surgical procedure by a minimally invasive approach, whether laparoscopic or robotically assisted, is likely to reduce length of stay, potentially reducing the cost of hospitalization and opening new capacity under the right conditions.c However, many of the newly proposed robotic procedures—e.g., hysterectomy, myomectomy, cholecystectomy, fundoplication—cannibalize procedures that already are performed laparoscopically. Hence there is no length-of-stay benefit to be gained, and our modeling approach becomes dispositive.
Passing the actual premium to the patient, as suggested by Weissman and Zinner, would consume 7 to 9 percent of median annual income for a family of four in the United States, and we predict most families would decline this option if presented with the outcome data. Certainly the decision to stick with LH may be easier in a capitated environment than in a fee-for-service system, where such a strategy could result in losing hysterectomies to competitors.
Findings of our analysis could be used to obtain buy-in from physicians on a strategy to promote preferential use of LH versus RH. Many institutions or practices provide a productivity-based incentive to surgeons who are employed by the hospital, or the surgeons themselves collect the professional fee for each case. In either scenario, the funds flow could be arranged to allow surgeons to share in gains when the hospital's net margin increases by taking the most cost-effective approach to cases.
Although our analysis is based on data for hysterectomies, the method could be applied to other surgeries in which distinct procedures achieve essentially identical outcomes. It particularly is useful if one of the options introduces new technology that requires a large capital investment.
Vikram Tiwari, PhD, MBA, is assistant professor, anesthesiology and biomedical informatics, School of Medicine, Vanderbilt University, Nashville, Tenn., and director, surgical business analytics, perioperative services, Vanderbilt University Medical Center, Nashville, Tenn.
Dan C. Krupka, PhD, is managing principal, Twin Peaks Group, LLC, Lexington, Mass.
Warren S. Sandberg, MD, PhD, is professor and chair, Department of Anesthesiology, School of Medicine, Vanderbilt University, Nashville, Tenn., and a member of HFMA's Tennessee Chapter.
a. Wright, J.D., Ananth C.V., Lewin, S.N., et al., "Robotically Assisted Vs. Laparoscopic Hysterectomy Among Women with Benign Gynecologic Disease," Journal of the American Medical Association, February 2013.
b. Weissman, J.S., and Zinner, M., "Comparative Effectiveness Research on Robotic Surgery," JAMA, February 2013; and Dixon, P.R., Grant, R.C., and Urbach, R.R., "The Impact of Marketing Language on Patient Preference for Robot-Assisted Surgery," Surgical Innovation, June 2014.
c. Krupka, D.C., Sandberg, W.S., and Weeks, W.B., “The Impact on Hospitals of Reducing Surgical Complications Suggests that Many Will Need Savings Programs with Payers,” Health Affairs, November 2012.
Using Simulation to Understand and Improve Operational Processes
Our objective is to estimate how many laparoscopic hysterectomy (LH) procedures and robotically assisted hysterectomy (RH) procedures can be performed in a dedicated operating room (OR) in one month. If the operative times for LH and RH never changed, this would be a trivial problem. However, we know that in our hospital each procedure is characterized by a distribution of times, which need to be taken into account to arrive at a credible estimate. The problem is compounded by our practice of allowing the OR day to stretch from 10 to up to 11 hours. To account for this complexity, which cannot be handled with mathematical formulas, we used computer simulation.
A computer simulation is a technique commonly used to reproduce the behavior of a physical system, such as production at a manufacturing plant, traffic patterns on a busy street, or the flow of patients in an emergency department. The objective is to understand the relationship between the system's inputs and outputs and, in turn, design better processes.
Simulation modeling allows managers to move away from averages-based thinking to decision making based on probabilities, which is much more realistic. In a simulation analysis, an entity of interest—a patient, for example—is run through the process repeatedly, with each run based on randomly drawn times from the probability distribution of times characterizing the process.
Using simulation of healthcare delivery systems to understand operational processes has been widely studied. Using simulation to inform financial decision making in hospitals is not as common but could be beneficial in annual budget preparations, premium negotiations with payers, and capacity planning (Schoenmeyr, T., Dunn, P.F., Gamarnik, D., et al., "A Model for Understanding the Impacts of Demand and Capacity on Waiting Time to Enter a Congested Recovery Room," Anesthesiology, June 2009).
Many commercially available software packages can be used to perform simulation for highly complex situations. For our relatively simple case, which involved only one OR and one surgical procedure, we were able to use MS-Excel.
The program has a built-in "histogram" utility that can be used to create the data distribution needed as the input for the simulation model. A combination of the "vlookup" and "rand" functions then can be used to randomly generate samples from the input variables' distributions. The simulation model can be run multiple times (pressing the F9 button on the keyboard generates a series of new random numbers), with the output collected for each simulation then subjected to standard statistical analysis.
In our scheduling data, patients spent an average of 180 minutes in the OR for LH and 345 minutes for RH for CPT code 58552 (which includes patients in the finance department's billed procedure codes of 68.31, 68.41, 68.51, 68.61, and 68.71). If we used these averages to examine the differences in throughput between LH and RH and used the same scheduling logic as in the simulation—10 hours of staffed OR time, 30 minutes of turnaround time, and a maximum of one hour of overtime—we would conclude that, on average, we can schedule only one RH procedure per day, compared with three LH procedures per day.
This conclusion would underestimate RH throughput and overestimate LH throughput. The simulation, on the other hand, factors in the variability associated with these procedures (standard deviation of 63 minutes for LH and 116 minutes for RH), and its output suggests that two LH cases per day will be performed in the single OR 38 percent of the time and three will be performed 68 percent of the time (with an average of 2.61 cases per day). For RH, one case per day will be performed in the single OR 25 percent of the time, two cases will be performed 58 percent of the time, and three cases will be performed 15 percent of the time (with an average of 1.98 cases per day).
Extrapolating this difference to a full month of 22 working days gives us monthly throughputs of 59 cases for LH and 43 cases for RH, for a ratio of 1.4. As seen in the exhibits in our article, this outcome factors into computations of the differences in monthly cash flows between the two procedures.
Publication Date: Friday, August 01, 2014