Research Notes Cost Effectiveness of Regionalization-Further Results for Heart Surgery Steven A. Finkler A recent study concluded that efficient production of heart surgeries requires a minimum volume of 500 procedures per center per year. A replication of that study, under a different set of circumstances, is reported in this paper. The geographic setting was substantially different: a west coast city in the earlier study versus a midwest suburb in the current study. The type of hospital was also substantially different: prepaid health plan-owned in the earlier study versus nonprofit fee-for-service community hospital, owned by a religious order, in the current study. Physician costs were included in the previous study but were excluded from the current study. Travel distances were included in both studies but because of the geographic setting tended to be greater in the current study. The results of the current study indicate that efficient production could be attained at a volume of 300 heart surgery patients per center per year. The difference from the results of the earlierstudy is attributable to the treatment of physician salaries, the hospital-specific decisions on appropriate capacity and utilization of facilities, and the greater travel distances. Regionalization of health care services has been receiving an increasing amount of attention. This attention has focused on quality (the conflict between the benefits of treating a patient close to home, in a familiar environment, and the possibility of more sophisticated, expert care in a regional center) and cost effectiveness (whether there are potential scale economies to be realized from the centralization of services in a regional The author would like to thank the editor and the referees for their helpful comments. This work was supported by the Wharton School Accounting Department Research Fund. Address communications and requests for reprints to Steven A. Finkler, Assistant Professor of Accounting and Senior Fellow, Leonard Davis Institute of Health Economics, the Wharton School, University of Pennsylvania. 0017-9124/81/1603-0325/$2.50/0 Health Services Research
326 Health Services Research 16:3 (Fall 1981) center). Previous authors have found that quality of care is likely to improve at higher volumes [1, 2]. One way to achieve higher volumes is through regionalization of services. We will not, however, investigate quality of care in this research note; rather, we will address the question of cost effectiveness. Cost effectiveness was discussed recently by the present author with reference to a new method, component enumeration (CE), used to collect micro data for a study in which it was concluded that strong economies of scale exist for heart surgery, and that efficient production requires volumes of 500 patients per location [3]. These conclusions, however, were based on a single case study, necessitated by the lack of micro cost data.' A major problem of the case-study approach is the question of the applicability of the results. This issue was discussed in the previous study [3], where it was concluded that, although the absolute costs might vary with differences such as geographic location, the shape of the long-run average cost curve for heart surgery should not change. Therefore the conclusions regarding appropriate volumes for cost-effective operation should not vary. This paper reproduces elements of the first study to see if that result does in fact hold under substantially different circumstances. The need for this is twofold. First, such additional information will be useful to health policy planners attempting to decide on the appropriate degree of heart surgery regionalization. Second, and more importantly, it will give another example of the CE methodology for collecting cost data for specific product lines. The CE method was developed specifically for the first study, but its intended use is for general health-care-product cost analysis. It is immediately useful for determining appropriate degrees of regionalization for a wide variety of health care services. There are four major differences between the first study and the one discussed here. First, there is a significant geographic change-from a downtown hospital in a large west coast city to a suburban midwest area about 40 miles outside of a major city. Second, the previous study examined costs in a hospital owned and operated by a prepaid health plan, while this investigation looked at a nonprofit fee-for-service community hospital owned by a religious group. Third, the study reported here excluded all physician costs or fees, whereas the previous study included physician costs. (In the previous study, physicians, being employees of the prepaid health plan, were at times treated as a fixed cost. Objections have been raised as to the accuracy of the results, given that the more common instance is fee-for-service.) The fourth and final significant difference is in travel costs. Travel costs act as an offsetting factor against the savings from scale economies that result from regionalization. In the previous study,
Research Notes 327 four hospitals offering heart surgery, all in close proximity to one another (1/4 to 4 miles apart), were examined to see if the travel costs could possibly offset the savings from regionalization. In this study four hospitals were again examined, but because of the less urban geographic setting of the regional center, they were 6 to 15 miles apart; this increased the travel cost in terms of carfare and the opportunity cost of the time spent traveling by patients and their visitors. The aim of this study, then, was to replicate the work done previously on the cost effectiveness of regionalizing heart surgery. The replication was performed using the same methodology, component enumeration. METHOD The basic CE methodology [3] takes a systems approach to identifying, accumulating, and costing economic resource consumption for a given product line. This means that each resource consumed by patients for a product line such as open-heart surgery is identified. This is a monumental task when one considers all of its implications. If the open-heart surgery program at a hospital increases traffic flow in the parking lot to the point where an extra attendant must be hired, the cost of that attendant is a cost of offering open-heart surgery. The philosophy of the CE method is that the product line under examination is a marginal item for the hospital, in the economic sense. Open-heart surgery will or will not be offered at the hospital. CE assesses how much more the total costs would be if open-heart surgery were offered than if it were not. The CE method is applied by first flow-charting patient treatment throughout the length of stay. This ensures inclusion of all direct costs. All other hospital departments are examined independently for indirect or ripple effects, such as the extreme example of an impact on resources used for the parking lot. Once the departments and types of resources consumed are identified, the volume and intensity of resource use must be assessed. This requires analysis of medical records and records of hospital personnel as well as time and motion studies. The CE method avoids problems of cross-subsidization inherent in the use of hospital cost data or charges. For example, a typical hospital accounting approach to finding the cost of surgery is to divide total operating room or surgery department costs by total surgery minutes to get a cost per minute. A relatively long but routine operation may be assigned more cost than a short but highly sophisticated operation that requires far more specialized equipment and facilities. Since the CE method looks specifically at the specialized equipment and facilities used, it avoids the bias introduced by the averaging of costs.
328 Health Services Research 16:3 (Fall 1981) Further, since the CE method views the service as a marginal one, joint costs are appropriately excluded, since the existence of the heart surgery program has no impact on them. This makes the method ideal for regionalization studies. The cost assessed for the heart surgery program by the CE method includes costs that would be saved by regionalization. It does not include overhead allocations, such as a share of the chief administrator's salary, that would not be reduced if open-heart surgery were no longer offered at that hospital. The CE method was applied only to one hospital for this study. While the CE approach provides voluminous data with a substantial amount of detail, gathering this detailed information only for one hospital imposes some severe limitations on the study. A formal econometric model is excluded by the single data point, as is the statistical support that might be derived from formal empirical tests of the model. As in the earlier study, only one hospital was used because of the expense of data collection. The description of CE provided here should give the reader some notion of the herculean effort required to extend this approach first to a large number of hospitals and then to a large number of services or product lines. To do either was beyond the financial resources available at the time of this study. This study and the earlier one, however, may be thought of as pilot projects providing a foundation upon which more extensive and rigorous efforts may be built. RESULTS AND DISCUSSION The principal result of this study is that strong economies of scale do exist in the production of heart surgery. The relatively flat portion of the average cost curve is reached at a volume of approximately 300 procedures per center per year. (By "flat portion" we meant that the additional reduction in cost per patient as patient volume increases is relatively small.) As in the earlier study, the economies of scale depend largely on the spreading of fixed costs of equipment, facilities, and basic program administration. As volume increases, these fixed costs are spread among the patients, thus lowering the cost per patient. The more patients, the less the fixed costs per patient. Eventually, fixed costs become a minor part of total costs and the variable costs, which are the same per patient at any volume, become dominant. In addition to regionalization of facilities, there is the issue of the effects of fewer surgeons performing a greater number of surgeries each. Such "surgeon regionalization" might result in cost savings due to a learning curve effect (i.e., as volume increases there is a familiarity phenomenon that results in increased efficiency), and it certainly has impli-
Research Notes 329 cations for quality. Statistics are not available, however, on how many surgeries are currently performed by each surgeon. It is possible that one surgeon may operate at several low-volume hospitals. That surgeon might then have been performing a high volume of surgeries prior to regionalization. Therefore we cannot begin to ascertain the impact of regionalization on volume per surgeon; nor, in turn, can we determine the ultimate impact of volume per surgeon on costs or quality. The earlier study indicated a minimum volume of 500 cases per center per year for efficient production, as compared with the 300 figure cited here. There are three key elements that account for this difference. The first is increased travel and inconvenience costs for patients, the second is method of payment of physicians, and the third is utilization of facilities and equipment. The earlier study considered travel costs between four hospitals, two of which were several blocks from each other. In the present study the distances between hospitals were considerably greater, and the patients were much more spread out geographically. Both studies considered the cost of additional distances traveled in terms of carfare and the opportunity cost of the extra travel time for patients and their visitors. The bulk of the latter cost is accounted for by an average of four visitors per day for the patients' length of stay. Travel and inconvenience costs for physicians have not been considered, for several reasons. These costs may increase with regionalization, but the net effect is not clear. First, although travel distances for the surgeon would increase with regionalization, this might not increase costs: the surgeon might be able to visit a larger number of patients at one time in a regional center, thus saving separate trips to a number of hospitals; one longer trip could well be less costly than a number of shorter trips. Second, current surgeons might relocate closer to regional centers, and new surgeons would be likely to locate near the centers; assessing the cost of relocation and determining an appropriate period for amortization of that cost is problematical. Third, regionalization may affect the number of cardiac surgeons needed; that is, it may result in fewer cardiac surgeons, each doing a greater number of procedures, which could lead to fewer total trips by physicians, who would see more patients per trip. Clearly, this is a complex issue. The sign of the travel and inconvenience costs for physicians may be positive or negative. This problem, plus the fact that the surgeon is only one individual who sees one or more patients per trip, and who may not make a trip every day during the patients' length of stay while the patients each have about four visitors per day for the entire length of stay, leads to the assumption that the increased travel and inconvenience costs for physicians would not be significant.
330 Health Services Research 16:3 (Fall 1981) The second key element accounting for the difference in the minimum annual volume of cases is the treatment of physician payment. In the earlier study, physicians were treated as salaried employees, which, in fact, they were. Thus cardiac surgeons hired would earn a fixed salary. If volume were insufficient to utilize the surgeons' full available time on heart surgeries, they would perform general thoracic surgery, a task that could have been performed by thoracic surgeons at a lower salary. Therefore the salary differential between a general thoracic surgeon and a cardiac surgeon was treated as a fixed cost to be borne by the heart surgery patients. Given that the aim of a regionalization study is to determine the potential cost savings from centralizing services, the above calculation was proper. However, since cardiac surgeons generally work on a fee-for-service basis, most hospitals will not realize savings from lower physician salary costs if they eliminate heart surgery. In that sense, the earlier study overestimated the volume needed for efficient production by including a cardiac surgeon's salary differential as an additional fixed cost, compared with hospitals working with fee-for-service surgeons. The third major cause of the difference in efficient volumes stems from the utilization of facilities and equipment. Many of the facilities and pieces of equipment associated with a heart surgery program are dedicated to that program; the bypass pump, for example, is used only for open-heart procedures. There is, however, a fair degree of flexibility. While the operating room for heart surgery is often designed in a configuration specifically for heart surgery, and there is sometimes a reluctance to use the room for other types of procedures, it may be used for other procedures as well. In the earlier study, the practice of the hospital was to dedicate operating rooms to heart surgery and to provide a second room when volume exceeded an average of one surgery per day. In the present study, the hospital examined did not dedicate its surgery room to heart surgery, and the practice was to add a second room when the volume exceeded an average of two heart surgery cases per room per day (in both situations a five-day week and 50-week working year were assumed). This difference, in practice, accounts for a part of the difference in the results of the two studies. This brings to light an important consideration in health planning for regionalization of services. Just as two physicians may use a different technique for an operation, different hospitals may have different ideas about the kind of utilization of facilities that constitutes good medical practice. Estimates of costs to be used as a basis for determining an appropriate degree of regionalization must assume particular standards for utilization. The standards chosen will undoubtedly be a matter for debate. The differences in the results of the two studies arise basically from differing hospital situations and decisions. The fact that there is such a
Research Notes 331 substantial difference points up the problem of relying on case-study results without considering their applicability under different circumstances. The validity of case-study results, moreover, is open to question when they are applied to a broad population. Nevertheless, it should be noted that, based on the California statistics cited in the earlier study, a minimum requirement of 300 procedures per center per year instead of 500 means that 88 percent, rather than 97 percent, of the hospitals offering heart surgery in California are economically inefficient. Clearly there is still a need for a rational approach to regionalization. SUMMARY AND DIRECTIONS FOR FUTURE RESEARCH A recent study concluded that efficient production of heart surgeries requires a minimum volume of 500 procedures per center per year; this study was replicated under a different set of circumstances: the geographic setting was substantially different-a west coast city in the earlier study versus a midwest suburb in the current study; the type of hospital was substantially different-prepaid health plan-owned in the earlier study versus nonprofit fee-for-service community hospital, owned by a religious order, in the current study; physician costs were included in the previous study and excluded from the current study; travel distances were included in both studies but because of the geographic setting tended to be greater in the current study. The results of the current study indicate that efficient production could be attained at a volume of 300 heart surgery patients per center per year. The difference from the earlier results is attributable to the treatment of physician salaries, the hospital-specific decisions on appropriate capacity and utilization of facilities, and the greater travel distances. The issue of regionalization is likely to gain more attention as time goes by and the public's call for cost containment in health care intensifies. There are two chief issues: whether it is more cost-effective to regionalize certain services, and what effect regionalization has on quality. While these issues are not independent, care must be exercised by researchers not to overstep their areas of expertise and competence. Economists are prone to confuse quantitative efficiency with quality, and medical researchers are too quick to assume that the hospital cost-finding accounting records will tell what something costs. This study and its predecessor show the potential of cost analysis of hospital product lines for resolving regionalization debates. The need for such analysis is clear. Wisconsin, Oregon, California, and Pennsylvania
332 Health Services Research 16:3 (Fall 1981) are working actively toward legislating regionalization. Other states are likely to follow. The next step for researchers sould be a major undertaking: applying component enumeration or a similar approach to a large number of product lines at a large number of hospitals to gain greater understanding of the economic impact of regionalization prior to regulatory action. In undertaking such an effort, extreme care must be taken not to choose a shortcut based on charges or hospital cost records. There is a great temptation to give lip service to economic costs and the issues of joint and marginal costs while using hospital charges as a matter of expediency. After all, hospital charges are cost-based. But they are only cost-based in the most crude way: first, they are rough, and second, the hospital cost data used for regionalization decisions are substantially biased by averaging (the cross-subsidization issue discussed earlier) and by including a share of joint costs (such as the salary of a chief administrator); thus the cost data that the hospital compiles are not a highly accurate measure of resource consumption. Charges must be high enough to allow the hospital to break even. Blue Cross, Medicare, and Medicaid disallow a variety of costs, such as community service programs, courtesy care, and bad debts. These costs must be made up through charges greater than the average cost. The charges are established under a convention of consensus pricing. If the charge for a specific service, based on hospital-determined costs, is much higher than that of other hospitals in the geographic area, the hospital may adjust the charge substantially downward and adjust other charges upward. For example, if a hospital calculates an average charge of $10,000 per openheart surgery for each of its 50 open-heart surgery patients and finds that other local hospitals with higher volumes and therefore lower fixed costs charge on the average only $7,000 per open-heart surgery patient, it can lower its charge to $7,000 (losing $3,000 x 50 patients, or $150,000) and add 10 cents to the charge for each laboratory test performed to make up the loss. The point is not that rate setting should be revised, but that given the problems inherent in using charge data, or even existing hospital cost data, studies using such data as a proxy for the cost of resources consumed are likely to give highly biased results. Despite the extra cost and effort involved, any attempt to obtain reliable research results will require use of CE or some similar method that assesses resources actually consumed by the service or product. The potential impact of regionalization consists of all the costs saved by the institutions that do not offer a service because of the existence of a regional center. Essentially this requires a summation of all the costs, on a marginal basis, that hospitals would incur to offer the service. The cost of
Research Notes 333 offering the service at the regional center and any travel and inconvenience costs should be subtracted from this summation to find the savings from regionalizing the service. Use of a methodology like CE is the key to measuring the impact of regionalization, as it provides a reasonable assessment of the costs of providing a service. The next step for researchers in this area is to apply CE to a number of product lines at a large enough number of hospitals to be able to draw statistically significant conclusions. NOTE 1. In searching for hospital-wide economies of scale, researchers typically use total costs for the hospital (a proxy for the true opportunity costs) as collected by the accounting information system. Costs for many hospitals can be gathered and compared in this manner. Regionalization, however, typically is discussed only with respect to individual services. For example, should maternity care or hemodialysis be centralized? On the micro level of the individual product line within a hospital, hospital-wide cost data are inadequate, as are accounting costs for individual products. The accounting-cost information has many implicit cross-subsidizations because of the departmental (as opposed to case-mix) orientation of the hospital cost-accounting system. (This problem has been discussed in this journal by Thompson et al. [4].) Thus substantial amounts of detailed information must be collected to ascertain scale economies for a particular product line, necessitating a case-study approach. REFERENCES 1. Luft, H., J. Bunker, and A. Enthoven. Should operations be regionalized? The empirical relation between surgical volume and mortality. New England Journal of Medicine 301(25):1,364, Dec. 20, 1979. 2. Adams, D.F., D.B. Frazer, and H.L. Abrams. The complications of coronary arteriography. Circulation 48(9):609, Sept. 1973. 3. Finkler, S.A. Cost-effectiveness of regionalization: the heart surgery example. Inquiry 16(3):264, Fall 1979. 4. Thompson, J.D., R.F. Averill, and R.B. Felter. Planning budgeting and controlling: One look at the future: Case mix cost accounting. Health Services Research 14(2), Summer 1979.