Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
PROBLEMS IN ADOPTION AND USE Significant empirical evidence suggests that the diffusion of new equipment-embodied medical technology often diverges from socially optimal paths. Decisions by health care providers and practition- ers to adopt and use equipment and its related procedures in pa- tient care do not appear to be based on a comparison of their benefits and costs to society. Some examples will illustrate this point: â¢ Fetal monitors have diffused widely into the practice of obstetrical care, but the benefits are not yet clear. â¢* â¢ Computed tomographic scanners have been adopted by more than 500 hospitals in the past 3 years, well in advance of the collection of information on their effectiveness as diagnostic procedures. â¢ Gastric freezing was diffused widely as a surgical technique used for the treatment of duodenal ulcers and was largely abandoned before definitive evaluation of its benefits could be published (Appendix D). â¢ Medical information systems have followed a fitful process of diffusion, encountering significant barriers except in the few research centers currently funded to demonstrate and develop such systems (Appendix E). â¢ Rehabilitative technologies have not developed in accordance with predictions of their great potential benefits to recipients?7 The committee has attempted to analyze the causes of such prob- lems. In particular, it has been interested in identifying sys- tematic tendencies of the health care delivery system to encourage 23
24 or inhibit the diffusion of particular classes of equipment- embodied technology. It finds that, in general, equipment- embodied technology in hospital, clinical, and ancillary services is subject to strong pressure for adoption and use with few coun- tervaling forces. Conversely, coordinative equipment-embodied technologies, particularly those relying on coordination among providers within the health care system to achieve full realiza- tion of benefits, face strong pressures against adoption and use. In fact, when equipment-embodied technology must be applied across institutions or requires the integration of services to fully realize its potential benefits, it suffers from barriers to its adoption and use. Preventive technology, particularly mass screening, and coordinative technology fall into this category. The remainder of this chapter summarizes the evidence that exists to support these findings. The committee is acutely aware of the limitations of the avail- able evidence. Although several empirical studies of the diffu- sion of hospital technology have provided valuable insight into the characteristics of innovative individuals and organizations, the impact on diffusion of the inost important policy-related factorsâthe reimbursement system, the malpractice system, and the organization of medical care deliveryâhas generally not been studied. Consequently, the committee has had to rely on empirical evidence and its own judgment in assessing the impact that the factors have had on the adoption and use of equipment-embodied technology. In making its analysis, the committee has been sen- sitive to the need for additional research and has identified areas where opportunities exist for better understanding of the diffusion process. These research recommendations are italicized throughout the text. THEORIES OF HOSPITAL BEHAVIOR The decision to adopt equipment-embodied technology is based on resource allocation. It requires the allocation of scarce capital resources to the purchase of equipment, and it further implies an allocation of the productive resources of the health care organiza- tion to the provision of certain services. Because the hospital is the major repository of equipment-embodied technology, the way in which hospitals allocate resources is of interest here. Are there definable attributes of hospital behavior that would explain why equipment-embodied technology as a whole, or certain kinds of equipment-embodied technology, would be adopted more or less readily than other productive resources? There are two sets of theories of hospital decision making that bear on the question of equipment adoption. The first assumes that
25 the hospital, while organizationally complex, possesses defined institutional objectives on whose basis its behavior can be pre- dicted. The objectives of hospitals have alternatively been postulated as:* maximization of quantity of services produced; maximization of quality of services produced; maximization of a weighted function of the quantity and quality of services pro- duced;89 maximization of the relative prestige of the hospital in the community;69 and maximization of the joint incomes of the physician staff.72 With the exception of the last hypothesis, which is the most divergent and novel articulation of hospital objectives, the alternatives all can be expressed as special cases of the third objective, the maximization of a weighted combination of the quality and quantity of hospital services produced. If quality and quantity of services are in fact the attributes of interest to the hospital, then the way in which quality is per- ceived, not only by the hospitals' decision makers, but also by the patients who will generate demand for services, has a major influence on the behavior of the hospital. Feldstein28 has theo- rized that hospital decision makers perceive quality of services as a function of the amount of labor and nonlabor inputs devoted to the production of medical care. A hospital that delivers care with greater service intensity would thus be perceived as a higher quality institution. However, perceived quality may be correlated as much with the level of sophistication of those inputs as with their aggregate amounts.46 If patients themselves identify hos- pital quality with the availability of capital-intensive equipment and systems, or specialized labor, then hospitals wishing to max- imize the quantity of services provided would respond by emphasiz- ing these inputs relative to others in order to increase the demand for the hospital's services. Thus, the willingness of hospitals to adopt new technology may rest on the degree to which patients and hospital decision makers equate hospital quality with the availability of this technology. Empirical research on percep- tions of hospital quality by health professionals and patients is lacking; however, at least one practitioner has observed that *These objectives have been postulated for private, nonprofit hos- pitals, which constitute the majority (56 percent) of the approxi- mately 6,000 nonfederal, short-term hospitals in the United States. The remainder are government owned (3l percent) or private, for- profit hospitals (l3 percent), which are likely to have different objectives from those of voluntary hospitals. However, of the hospitals with 300 or more beds, 8l percent are private, nonprofit institutions. It is legitimate to concentrate on this ownership form since, to a large extent, standards of hospital care will be dictated by this subset of hospitals.
26 professionals may be "... charmed by blinking lights and cathode- ray tubes into adopting a technology that will decrease the fre- quency of contact between the human monitor and the patient." This theory would also imply that technology in clinical settings, with its high use and visibility in patient care, would be favored over administrative technology. A second set of theories of hospital behavior rests on the as- sumption that hospitals are complex organizations where decisions cannot be predicted from a single objective. Instead, decisions to adopt new technology are assumed to be a function of the struc- ture of the hospital organization and the relative dominance of competing interest groups. Hospital decisions are made on the basis of intra-organizational politics rather than on the basis of a single goal. According to Tushman,116 the politics of orga- nizations is the "structure and process of the use of power to af- fect goals, directions, and major parameters." The political view of hospital behavior sees it "not wholly determined by environ- mental conditions or constrained to cooperative or goal-oriented behavior but as a complex of cross-cutting strategic decisions and exchanges at all levels of the organization."116 Using this per- spective, one is led to seek out the organizational subgroups that affect and are affected by decisions regarding the adoption of new medical technology. A leading theory of hospital behavior following this line of reasoning has been described as the "physician dominance" theory.53 According to this argument, the hospital is in effect run by the attending physician staff, because it is physicians who direct the patients to the hospital. Therefore, to understand the behavior of hospitals, it is necessary to examine the goals of different kinds of physicians and to examine the coalition behavior of these groups. The physician dominance theory has been criticized by Greer^ and others.93 According to Greer, the hospital adminis- tration and hospital boards may have as much or more power than any individual physician or group of physicians. It may also be true that hospital decisions about hospital technology result more often from the coincidence of goals among various groups than the dominance of particular groups. In the absence of a budget con- straint, the hospital administrator can meet all demands over time. What, then, can one infer from this body of theory with respect to hospitals' adoption of new equipment-embodied technology? Al- though there is little available in the way of deductive infer- ences, the theories do suggest a number of reasons why clinical equipment-embodied technology is valued in the hospital. It is an input in hospitals' production of both quantity and perceived quality of services; some equipment increases the productivity of attending physicians and some contributes to the prestige of the institution and its affiliated physicians.
27 EMPIRICAL EVIDENCE ON THE DIFFUSION OF EQUIPMENT-EMBODIED TECHNOLOGY Because theory provides only broad predictions about factors affecting the adoption of equipment-embodied technology, it is useful to examine empirical studies of diffusion both for cor- roboration of theory and for a better understanding of the nature of the diffusion process. Three central questions are of interest: (i) What attributes of equipment-embodied technologies influence their rate and pattern of adoption? (ii) What attributes of the potential adopters of equipment-embodied technology influence the process? (iii) What factors in the environment in which potential adopters operate influence the process? Each of the questions is discussed in turn. Attributes of the Technology One would expect certain attributes of a technology or innovation to influence the speed of diffusion. Tanon and Rogers113 have suggested that the following are likely to affect the speed of diffusion: â¢ Relative advantages over existing technology â¢ Compatibility with existing values of the institution â¢ Complexityâthe degree to which it is easy or difficult to understand and use â¢ Relative ease in experimenting or trying out the innovation on a limited basis â¢ Observabilityâthe degree to which the results of innovation are visible to others. No systematic study of the impact of these or other attributes on the speed of diffusion of equipment-embodied medical technology has been attempted. However, there is some indirect evidence to suggest that at least some of these characteristics are important determinants. The relative advantages of a new technology, either in improving patient outcomes or reducing patient risks, has been shown to be a strong impetus to diffusion. Greer44 has documented the reluctance of ophthalmologists in one community to adopt a new equipment-embodied cataract removal procedure, phakoemulsification, because of its added patient risk in the absence of clear thera- peutic benefit, in spite of significant savings in patient recov- ery time and opportunities for physicians to augment their incomes. Fineberg's study of gastric freezing (Appendix D) demonstrates the sensitivity of physicians to the pain and risk of a new procedure in the absence of unequivocal evidence of effectiveness. This
28 procedure was largely abandoned several years before the first definitive study of its effectiveness in treating duodenal ulcers. Computed tomographic scanning has rapidly replaced more dangerous, painful, and invasive procedures for detecting brain lesions. The more unpleasant the existing alternative pro- i90 cedures, the faster the substitution process has occurred. Warner136 showed that the diffusion of new drugs for treatment of a desperate condition occurs extremely rapidly compared to drugs used for less serious illness. Studies of the diffusion of innovations in other industries have demonstrated the importance of certain innovation character- istics. The industrial literature78,102 shows that the size of the investment required for adoption of the technology (relative to firm size) is important in determining the speed of diffusion. Presumably, the larger the required investment, the more risk the firm faces and the greater the problems of capital availability. Hospitals, too, face capital constraints, but they are likely to be resolved differently from those of industry. In hospitals the risk associated with capital investment is greatly reduced by the predominant cost or charge-based methods of hospital reim- bursement. The ability of an innovation to reduce costs and increase profitability has been shown to be important in studies of agri- cultural innovation and industrial innovation. In hospitals these financial advantages are likely to be less important than considerations of effectiveness and safety. Indeed, in a study of equipment adoption decisions in l5 Boston hospitals, Cromwell19 found that criteria such as "improvements in patient care," "life saving capability," and "patient safety" were much more important to hospital administrators than were the financial or cost- saving attributes of equipment. Characteristics of Adopters Of course, the potential adopter's perceptions about technology depend upon his own characteristics, and, for this and a variety of other reasons, such characteristics play a role in the diffu- sion of technology. A number of studies of the diffusion of hospital clinical tech- nologies have attempted to isolate attributes of the hospital re- sponsible for rapid or slow responses to new technology. Gordon et al.^Â° showed that the adoption of accepted medical technology (respiratory therapy equipment) is related to the degree of visi- bility of consequences and the degree of decentralization of re- source allocation decisions within the hospital. The study also
29 revealed that hospitals with highly trained medical staffs tend to be more innovative than others. As expected, hospital size and medical school affiliation have been shown to be important determinants of early adoption of new clinical equipment-embodied technology. , In a study of nu- clear medicine facilities, Rapoport9Â® showed that the existence of a high percentage of hospitals with medical school affiliation in a state slowed down the diffusion of a new technology in unaf- filiated hospitals. The leadership role played by teaching hos- pitals in the adoption of new technology is hypothesized to have had a dampening effect on competition among nonaffiliated hos- pitals. In a study on the diffusion of innovative health care services in hospitals, Kaluzny et al. concluded that larger hospitals, particularly in urban areas, adopt innovations earlier and more quickly than smaller hospitals. Cromwell's study indi- cates that bed size influences adoption only in nonteaching hos- pitals.19 Other studies suggest that the more comprehensive a hospital's services are, the more likely the hospital is to be highly innova- tive. Rapoport98 noted that hospitals that adopted nuclear med- icine facilities early tended to be ones already well equipped with specialized services. Similarly, Cromwell19 found in an intrastate study of hospital diffusion that the range and number of other complex services (for example, intensive care unit, ra- dium therapy, cardiac catheter lab) offered by a hospital is positively correlated with adoption. The profit or nonprofit status of a hospital should also de- termine its adoption behavior. Theoretically, profit-making hos- pitals should be slower in adopting new clinical technology, particularly cost-raising technology. However, Cromwell's study showed no consistent tendency for profit-making hospitals to adopt new equipment more slowly.19 The organizational factors responsible for patterns of diffusion of clearly disproven equipment-embodied technology have not been studied in detail. Thus, we have no information on whether the organizational factors identified in the literature as important determinants of adoption of accepted technology are also those that encourage the adoption of poor technology. Perhaps hospitals affiliated with medical schools are better at discriminating be- tween effective and ineffective technologies than unaffiliated hospitals. Unfortunately, there is no information to test this hypothesis. The reason is obvious. The few data sources available to support research on the diffusion of equipment-embodied tech- nology do not compile data on ephemeral technology. Therefore, the study of failed technology would entail retrospective primary data collection, a costly and perhaps even infeasible undertaking.
30 Environmental Factors Hospitals encounter various constraints on their operations from numerous external sources. The nature of these constraints can be expected to be a major determinant of the patterns of diffusion for new hospital equipment. The impact of hospital competition on the adoption of equipment-embodied technology has been studied from several van- tage points. In his study of the diffusion of nuclear medical facilities, Rapoport98 attempted to measure the impact of inter- hospital competition on adoption, using the proportion of a state's population residing in urban areas as the measure of the competi- tive environment. He hypothesized that more urban environments would experience more competition among hospitals. The variable was found to be significant in explainina statewide rates of adoption of nuclear medicine. Rapoport9Â° also hypothesized that states with relatively few physicians might see higher rates of equipment adoption due to competitive forces, because hospitals in these states would have to compete for scarce physician staff. However, physicians also generate demand for services, so the net impact of physician availability is not clear. Indeed, Rapoport found that physician availability is not significantly related to statewide diffusion rates for nuclear medicine. Cromwell19 also studied the impact of the number of physicians in a state on the number of hospitals possessing an equipment-intensive facility. The total number of physicians per capita was found to be posi- tively related to the number of hospitals adopting a technology, and the ratio of specialists to nonspecialists was also important for most services. In fact, the more specialized the equipment, the stronger was this relationship. Thus, it appears that the demand-generating role that physicians play is more important than competition among hospitals for physician staff in affecting adoption rates. Certainly, regulation is likely to affect diffusion of equipment-embodied technology. Some inferences can be drawn from related situations. For example, public utilities subject to rate-of-return regulation appear to have little incentive to in- novate, although they have a greater incentive to invest heavily in capital assets. Indeed, the major incentive seems to come from the delays inherent in such regulationâthat is, the "regulatory lag."7 The major regulatory policies directly affecting adoption de- cisions by hospitals are the state certificate of need (CON) and capital expenditure approval programs. These laws mandate review and approval of large capital expenditures (generally in excess of $l00,000) by local and state health planning agencies, with various sanctions applied to a hospital that goes ahead with an
3l expenditure without this approval. A priori, one would expect this law to slow down the diffusion of expensive equipment. However, the evidence suggests just the reverse. The introduc- tion of CON may have increased diffusion of expensive technology, especially in the early years of each state's program.53 Crom- well et al.19 tested whether the existence of a CON law has a significant effect on the adoption of equipment-intensive clin- ical services. CON was found to be significantly and negatively related to rates of adoption of x-ray, cobalt, and radium ther- apy services, but it was not a significant explanatory variable for other services, including intensive care, open-heart surgery, and diagnostic nuclear medicineâthree services for which it should have been affected. Other forms of health care regulation have also been posited to affect the diffusion of technology. Institutional licensure, accreditation, and certification programs dictate standards of hospital construction and operation. State licensure programs focus largely on fire and life safety, water sanitation, minimum service standards, and guidelines for staffing and staff quali- fications (Appendix E); accreditation by the Joint Commission on the Accreditation of Hospitals (JCAH) also concentrates on these aspects of hospital operation. The net effect of these regulations is unclear, since considerable capital funds can be tied up in meeting licensure and accreditation requirements, thereby reducing availability of capital for equipment and facil- ities acquisition. Needleman and Lewin (Appendix F) conclude that the effect of facility licensure and certification programs on hospital adoption decisions is not well understood and should be investigated further. Perhaps the most important factor bearing on the adoption be- havior of hospitals is unique to the health care industryâthe system of third-party payment. That the methods of paying for health care services would influence the patterns of adoption of equipment-embodied technology seems obvious. Yet, there is very little empirical study of the impact of methods of financing and reimbursement on the adoption of new technology by hospitals. This is the result largely of data limitations and the ubiquitous nature of prevailing reimbursement systems, limiting the oppor- tunity for comparative studies. Only one study has attempted to measure the effect of a change in health care financing on the adoption of equipment. Russell*04 investigated the rates of diffusion of three kinds of equipment- embodied technologyâintensive care units, nuclear medicine, and electroencephalographyâbefore and after the inception of Medicare. The results of the study are equivocal. The increase in funding implied by the introduction of Medicare did appear to speed up the adoption of some services in hospitals within specific size
32 categories, but the result was not uniform across all the tech- nologies studied or across all hospital size categories. A major limitation of this study was the inability to control for changes in technology that independently affect the diffusion process. Com- parative empirical studies in this area are warranted. To what extent does charge- or cost-based retrospective reimbursement of hospitals lead to adoption behavior that is different from such behavior under prospective budgeting or formula rate-setting? How has the diffusion of new technology in other countries with different methods of financing health care differed from experi- ence in the United States? These comparative studies can be augmented by comparison of adoption behavior over time as changes in reimbursement methods are introduced within the United States. Lessons from the Empirical Studies The empirical studies of diffusion, though selective, reveal a pattern to the diffusion process for clinical and ancillary hos- pital technology. At least for the equipment and equipment- intensive services studied, earlier adopters are large hospitals with decentralized organizations and hospitals affiliated with medical schools. There is strong indirect evidence that competi- tive factors also play a role in enhancing the diffusion process. Direct regulation of the process of diffusion has not had much effect. Although there has been little systematic study of the attri- butes of technology itself that affect diffusion, perceived medi- cal promise appears to dominate financial or cost-saving attributes in hospitals' priority setting. But the studies shed no light on how well hospitals discriminate among different kinds of clin- ical technology in this regard. If, for example, there is a systematic tendency for hospitals to over- or under-value particu- lar categories of equipment-embodied technology (see Figure l) relative to one another, studies have not revealed it. Biases against adoption are difficult to detect. Technologies that have not diffused are not highly visible and do not lend themselves to empirical study. Consequently, it is difficult to identify par- ticular procedures or equipment that are fully developed and ready for widespread use that have not been diffused. At best, one can identify instances of promising demonstrations or individual applications that have not been pursued much beyond the develop- ment phase. Little empirical evidence is available on the impact of the system of financing and delivering health services on technology diffusion. In particular, the impact of the system of paying for health care services on the adoption of different kinds of health care technology has not been studied. Nor has there been
33 a systematic analysis of the impact of the medical injury com- pensation system (malpractice) on the adoption or use of new equipment-embodied technology. The effect of the organization of health care delivery on the adoption of different kinds of equipment-embodied technology also remains unexplored. The lack of appropriate control groups and data has hindered this kind of study. Nevertheless, the importance of all three factors in determining the patterns of diffusion of different kinds of technology suggests a closer look at each of them. The remainder of this chapter is devoted to an assessment of the role that these factors play in diffusion based on the indirect and fragmentary evidence that exists and upon the combined judg- ment of the committee. THE IMPACT OF THE HEALTH CARE FINANCING SYSTEM ON THE ADOPTION AND USE OF EQUIPMENT-EMBODIED TECHNOLOGY Four aspects of the present system for financing health care must be analyzed. These are: â¢ Methods for reimbursing hospitals for routine services â¢ Methods for reimbursing hospitals for ancillary services â¢ Methods for reimbursing physicians â¢ Limits to third-party reimbursement. Each is discussed below. Methods for Reimbursing Hospitals for Routine Services As it is presently structured, the health care financing system provides hospitals with strong positive incentives to adopt and use certain kinds of equipment-embodied technology. Payment for hospital services is almost totally covered by third parties (insurance companies, unions, and governments that reimburse hospitals for the provision of covered services to their members) Typically, these payers reimburse hospitals on the basis of charges or costs and pass the expenses on to consumers through periodic premium payments or taxes. For the most part, reim- bursement for routine hospital services is restrospective.* *Several innovative reimbursement programs have been developed over the past 5 years, some on an experimental and some on a permanent basis. These new programs have been based on a pros- pective payment concept, where hospital payment rates or budgets are determined in advance for a specified period. Often, these programs contain incentives to the hospital to introduce cost- saving procedures or technologies.
34 That is, the level of payment is based on actual costs incurred. If hospital costs increase, they are reflected in higher rates of hospital reimbursement. Given this "pass-through" capacity, third-party hospital reimbursement provides incentives to hospi- tals to push the adoption of equipment-embodied technology to the limit of the availability of capital. If hospitals seek to maximize a combination of quantity and quality, then third-party charges or retrospective cost reimbursements would lead hospitals to increase both the quantity and quality of services beyond a socially efficient level.89 This tendency would be checked only by limitations on funds available for investment in any period or by limitations on patient demand for hospital care. Some evidence is available indicating that hospitals do, in- deed, tend to push capital expenditures to the limits of capital availability. Ginsburg39 has shown that the trade-off between capital spending for general bed capacity and specialized equip- ment depends on how crowded the hospital is. If occupancy rates are low, the money will be spent for capital equipment; if oc- cupancy rates are high, pressures for new additions to the hos- pital will mount and capital funds will be channelled in that direction. Salkever and Bice106 have shown that when capital ex- penditures for new bed capacity have been limited through regu- latory action, capital spending has merely shifted to new equipment; in this case, total capital spending is unaffected. How is capital availability determined? Capital funds for the acquisition of new equipment and facilities come from a variety of sources whose relative importance has been shifting over time in a clear direction. These sources include philanthropy, pub- lic bonds, federal subsidies, and debt financing. The proportion of hospital capital spending made possible through debt financing has increased dramatically over the past l0 years, as reflected in the following figures on construction capital: l969 l973 l975 l977 Percentage of total construction funds from debt financing 32%* 54.3%t 56.8%t approx. 67%* Sources: *Iglehart, John K. "Stemming Hospital GrowthâThe Flip Side of Carter's Cost Control Plan." National Journal, June 4, l977:850. t"AHA Research Capsule No. 24: Sources of Funding for Construction." Hospitals 5l:59. This increase in the importance of debt is itself a reflection of the retrospective cost-based system of reimbursement. Since
35 coverage of hospital costs is guaranteed, lenders incur very little risk in making loans to nonprofit voluntary hospitals. Hospitals use debt to augment their sources of capital financ- ing. There are, of course, limits to the absolute dollars avail- able to an institution in any single time period. Bellinger54 has tested the hypothesis that the hospital follows a gradual adjustment process in its investment plans. This is consistent with the investment behavior of most businesses and with the natural conservatism of lenders. In an attempt to deal with the problem of insufficient risk associated with hospital capital expenditures, Congress amended the Social Security Act in l972 to restrict Medicare reimburse- ment for capital costs to those capital expenditures approved by a designated state health planning agency. P.L. 92-603, Section ll22, was intended to increase the financial risk to the hospital (and therefore to prospective lenders) associated with capital expenditures that are not in the public interest. The law has not been effective to date because it does not re- strict funding of the noncapital costs associated with the ser- vice, nor does it restrict hospitals from using endowment and philanthropic funds to cover unapproved capital expenses. Most important, however, up to this time few expenditures for capital equipment have been denied by health planning agencies. Methods for Reimbursing Hospitals for Ancillary Services Certain hospital services denoted as ancillary services are billed separately from the routine daily rate in hospitals. These commonly include laboratory, radiology, anesthesiology, pharmacy, and certain special therapeutic procedures. Depending upon the particular hospital reimbursement program, these ser- vices can generate substantial revenue surpluses for the hos- pital, which can be applied against losses to other insufficiently reimbursed services. As new ancillary procedures are introduced in a hospital, reimbursement is usually guaranteed. Thus, the incentive to adopt new ancillary technology is even stronger than for other services, because it often expands the pool of funds available for capital or operating expenditures. Some hospitals are reimbursed by Blue Cross, Medicare, and other payers on the basis of a predetermined per diem rate that includes ancillary services. Although these per diem reimburse- ment methods differ in their particulars, they all reimburse on the basis of a fixed amount per day rather than on the basis of the volume of services consumed. One would therefore expect these systems to create less incentive for the adoption and use of new equipment in the ancillary services. This hypothesis has not been tested.
36 Methods for Reimbursing Physicians As the primary gatekeeper for the use of clinical and ancillary equipment-embodied technology* and as an important participant in the operation of the hospital, the physician is a key deter- minant of decisions bearing on the adoption and use of medical technology. Consequently, the methods of paying for physicians' services must have a significant effect on the kinds of equipment- embodied technology that will be adopted and used. Several theo- ries of physician behavior have been advanced. The simplest and perhaps the most questionable is that the physician chooses to provide the number and kind of services that will maximize his income subject to legal and moral constraints. A more elab- orate theory is that the physician seeks to reach a "target" level of income subject to constraints on leisure and prestige.25 Income, prestige, and leisure are reasonable and expected goals of any professional; the ethical goal, delivery of quality medi- cal care to patients, must also be considered in an examination of physician behavior. The third-party fee-for-service system of physician reimburse- ment, which rewards physicians on the basis of the number of pa- tient visits or procedures performed, should have a significant impact on physicians' decisions to use health care services, especially in the absence of significant perceived financial or medical risks to the patient.t To what extent does this tendency to overuse health services in general translate into a special problem for equipment-embodied technology? If the ability to perform a procedure depends upon the avail- ability of equipment, then it is incumbent upon the physician desiring to perform the procedure to see that the equipment is adopted by a hospital in which he has staff privileges. The equipment becomes the physician's "tools of the trade." To the extent that the fee-for-service system links the physician's in- come to his ability to perform the procedure, it strengthens the *Most diagnostic and therapeutic services must be ordered by the physician or under the physician's direction. The patient may in- fluence use by seeking out a particular service or by refusing ser- vices. However, the power of the patient to influence decisions is limited, and it is not clear that the patient should be en- couraged to take this role. The more technologically sophisticated a procedure or instrument, the less likely is the patient to be able to adequately assess its relevance to a particular clinical situation. The importance of medical risk in affecting physician decisions is described on page 27 above.
37 imperative that the hospital supply the physicians with these tools. The proliferation of open-heart surgery units in the United States, to the point where the quality of care delivered in these units has been called into question,63 has allegedly resulted from this phenomenon. As teaching hospitals train cardiovascular surgeons and then close their doors to the gradu- ates of their residency programs, these physicians must find a hospital either with an existing capability or with the willing- ness to establish such a capability in order to make a living in the field for which they were trained. In the view of this com- mittee, this example illustrates the combined impact that fee- for-service, interhospital competition for prestige and patients, and the system of graduate medical education has on the rate of diffusion of clinical equipment-embodied technology. Once a physician has access to the equipment necessary to per- form a procedure, then the criteria he invokes to determine the necessity or appropriateness of use are likely to be sensitive to the procedure's income-generating potential in some instances (again, taking into account the risk factors). However, the magnitude of this effect on the use of equipment-embodied pro- cedures by physicians is unknown. Other factors, such as defen- sive medicine, scientific curiosity, and commitment to high- quality care in the absence of financial barriers to patients may be equally or more important than is the fee-for-service sys- tem. Indeed, the use of many diagnostic procedures, particularly those performed in hospitals, may not be strongly influenced by the fee-for-service system, since for the most part the physician who orders a test is not paid for performing it. However, the admitting physician does receive a fee for in-hospital visits that might be justified by ordering tests. In some cases, physician income is directly tied to the per- formance of tests. Some examples are: â¢ Laboratory tests ordered by physicians in private practiceâ in some locations and under some third-party payment mechanisms the ordering physician may profit from the test. â¢ CT brain scans performed by neurologists in private practice. Often a patient is referred to the neurologist for a neurolog- ical workup; it is the specialist's decision whether to order a CT examination. If the specialist also performs the examina- tion, the clear incentive to overuse exists. â¢ Some diagnostic surgery, such as gastroendoscopy. â¢ Diagnostic x-ray procedures in physicians' offices on a nonreferral basis. Hospital-based physicians (for example, radiologists, pathol- ogists, and anesthesiologists) are paid by hospitals for services
38 performed in a number of ways, including salary, fee-for-service, and percentage of gross revenues of their respective departments. Except in the case of salaried physicians, involving approximately 25 percent of pathologists and l0 percent of radiologists,18 their income varies directly with the number and type of services per- formed. Therefore, it is in the financial interest of the hospital-based physician to adopt technologies that will maximize volume, regardless of expense incurred. Labor-saving technology will benefit the hospital-based physician when the labor being saved is his own. Thus, technology that standardizes results, im- proves reliability, increases effectiveness, or shortens the physician's time is clearly in his economic interest. Administrative hospital technologies, including medical infor- mation systems and hospital communication systems, are in a per- verse position with respect to physician interests. Often the establishment of a new administrative technology involves signifi- cant time and inconvenience for a hospital's medical staff. When the new administrative technologies have been incremental, have not involved major changes in the organization of medical delivery, and have significant benefits clearly demonstrable to the physi- cian, resistance on the part of physician staff has been minimal. Electronic paging systems are an example. By contrast, establishing on-line medical information systems with the capability of process- ing patient care information and performing certain hospital func- tions automatically has encountered more resistance from hospital medical staffs. The introduction of these systems involves signif- icant, if temporary, inconvenience to physicians with patient benefits that are difficult to demonstrate (Appendix E). Limitations on Third-Party Payment Coverage The prevailing system of third-party coverage does not include all types of health care services. Where coverage is lacking or in- adequate, technology is at a particular disadvantage. One obvious example of the effect of coverage limitation is the exclusion from most insurance plans of coverage for preventive medical care, most notably screening services. This creates a bias against the adoption of technology in this category by health care providers. Mammography screening is a telling example. The screening technology has been largely validated for women 50 years of age and older, yet, outside of the federally funded Breast Can- cer Detection and Demonstration Projects (BCDDP's), there is little activity in this area. Fortunately, the technology of mammography has developed in response to a market for its diagnos- tic uses, so the problem of inadequate technological development following an inadequate market has not occurred.
39 THE IMPACT OF DEFENSIVE MEDICINE ON ADOPTION AND USE OF NEW TECHNOLOGY It is often claimed that the medical injury compensation system, which holds hospitals as well as physicians responsible for neg- ligence in cases involving injury to patients, encourages the adoption of equipment-embodied technology. Defensive medicine is virtually always cited as an incentive for physicians to over- use diagnostic services, but there are at present no reliable studies demonstrating the extent of its effect. A recent poll of physicians conducted by the American Medical Association revealed that a majority of physicians believe that unnecessary tests are ordered as a hedge against malpractice. Certainly the evidence is strong regarding the use of particular tests. Rou- tine skull x-rays for all emergency patients with head injury is a frequently cited example. However, the impact of defen- sive medicine on the overuse of existing and established tests, such as skull x-rays, must be distinguished from the impact of defensive medicine on the overadoption of new technology (new tests or more reliable tests). Defensive medicine may well be a potent force influencing use and adoption in the later stages of diffusion, but its importance in earlier stages is questionable. Although much is known about the process by which new medical procedures become standards of medical practice, the point in the diffusion process at which the standard becomes an important in- fluence on adoption and use and the impact of that timing on the pattern of diffusion are largely unknown. It is frequently as- serted that less than 4 years after its introduction, cranial computed tomographic scanning has become a standard of practice for diagnosing certain brain lesions. What this has meant and what it will mean for the use of CT head scanning and the further diffusion of the technology is largely unknown. Study of the operational meaning of "standard of care" for the diffusion of new diagnostic technology is certainly in order. THE IMPACT OF ORGANIZATION OF HEALTH CARE SERVICES ON ADOPTION AND USE The delivery of medical care in the United States is character- ized by two related attributes: disaggregation and specializa- tion. Together these two factors have an immense, though unmeasured, impact on the kinds of technology that are accepted and diffused throughout the health care system. To the extent that present systems of reimbursing for health care services encourage disaggregation and specialization, the health care financing system can be further implicated in creating systematic biases in pressures for and against technology adoption.
40 Disaggregation of Care Disaggregation refers to the delivery of medical care by many small providers who operate essentially independently of and in competition with one another. Disaggregation offers some advan- tages to patients. It increases geographical access to medical care, and it increases patient choice. However, disaggregation has significant implications for the adoption of equipment-embodied technology. In particular, two kinds of technology are likely to be af- fected : (i) Technology offering significant economies of scale in relation to the size of providers. A technology is subject to economies of scale in the production of services if the average cost of producing each unit of service decreases as the volume of service increases. Most equipment-embodied technology, in- volving as it does high initial capital costs, is subject to econ- omies of scale, at least up to the capacity of each machine or system. The recent revolution in the automation of clinical laboratory procedures, beginning with clinical chemistries in the late l960's and continuing into microbiology at present,128 is a dramatic example of the potential for economies of scale in pro- duction. Two separate problems stemming from the phenomenon of econ- omies of scale in the production of medical services must be distinguished. First, there is the problem of a technology whose economies of scale are so great that the entire market for its use is too small to sustain its costs of development, production, and distribution. Second, a technology may be subject to econ- omies of scale that are too large for the markets of individual providers. An example will illustrate the difference. Suppose an automated testing system is developed that is less costly than existing manual methods at volumes above l00,000 per year, based on the assumption that at least l00 machines will be pro- duced and sold. Suppose also that in the entire country only l million tests of this type are required in any year. The market for the machine is therefore limited to at most l0 units. The costs of developing, producing, and marketing the device are thus prohibitive, and the concept may never get beyond a proto- type stage. This is the problem of an insufficient total market. Now suppose that a breakthrough in design enables the machine to be less costly than manual methods at volumes in excess of 5,000 tests per year (with at least l00 machines produced). But suppose that each individual provider operates with a market of only 2,000 tests per year. Individual providers would not real- ize the savings from the economies of scale. This is the problem of a disaggregated market.
4l The first problem, the lack of a total market, does not always argue for policy intervention. The technology just may not be cost-effective, regardless of its economies of scale in produc- tion. Unless the resource costs of developing, producing, and distributing the technology could be significantly reduced through some public policy, the technology is simply not ready for dis- tribution. Some technologies may suffer, however, from a lack of a total market and yet be very much in need of public assistance. For example, some rehabilitative technology appears to suffer from the problem of an insufficient total market.87 The diverse na- ture of the problems faced by the handicapped and the relative]^ small number of individuals who can benefit from any particular device often renders the cost of developing and distributing new devices prohibitive to those who need the assistance. Yet, from society's standpoint, the development of a rehabilitative tech- nology may be quite justifiable. Public subsidies of develop- ment or programs to assist in paying for new devices may be warranted in some cases. The second problem, disaggregated markets, lends itself to corrective policy since it deals with a market structure that is out of balance with the economic environment.* The only way the technology will diffuse under such circumstances is (a) if each provider can manipulate prices or artifically increase volume to pay for the machine or (b) if providers share the services of a regional technology. The first alternative results in over- adoption of the technology relative to its net social benefit. The present system of charge- or cost-based reimbursement, which calculates payment on actual volume, and the freedom of hospitals and physicians to increase the volume of certain procedures, par- ticularly clinical laboratory tests, have encouraged this result. The second alternative, regionalization, may lead to optimal rates of diffusion. No diffusion at all represents a case of under- adoption and a loss of the benefits realizable from the technology. "Regionalization" and "sharing of services" have a long history of proponents who have recognized these problems.4 Indeed, region- alization has been a major goal of federally mandated health plan- ning programs. But regionalization of services faces strong resistance from hospitals and physicians, and the retrospective reimbursement system in operation today does nothing to discourage that resistance. â¢ The regionalization concept also has some natural limitationsâfor example, when services must be provided on an emergency basis or when considerations of patient access outweigh economies of scale in production. *Of course, the level of disaggregation of providers may be in response to important access considerations.
42 (ii) Technology offering significant benefits realizable only through integration of providers or services. Coordinative system-wide technology, such as medical information systems and emergency medical services systems technology, is often subject to nonadditive benefitsâthat is, the benefits accruing from the collaboration of multiple providers outweigh the benefits from individual adoption. For example, the usefulness of the problem- oriented medical information system that records patient data on the basis of medical problems is to a large extent dependent upon the integration of ambulatory care and hospital care data.134 In the words of Lindberg (in Appendix E): To the extent that health care institutions do not work smoothly and sensibly with one another, the medical in- formation system cannot be shared or transplanted. To the extent that health care institutions are balkanized into small administrative parcels, the information systems must of necessity be small as well. It is quite clear why mini-computers are so popular in medicine, and why large data base systems are so rare. The mini-system matches the mini-administrative fiefdom. Emergency medical services (EMS) systems provide a particu- larly cogent example of the problems of coordinative technology. Throughout the country, EMS grew up as a network of independent agencies, including police, volunteer and for-profit ambulance services, and hospital emergency rooms. These participants were often linked by informal agreements and sometimes by ad hoc ar- rangements. In most areas of the country, communication technol- ogy was rudimentary; few ambulance corps were linked by radio to hospitals, and central dispatching was nonexistent. Indeed, there was generally no single agency, public or private, respon- sible for the planning and delivery of emergency medical services on a coordinated basis. The Emergency Medical Services Systems (EMSS) Act of l973 (P.L. 93-l54) was developed to combat these serious deficiencies. Not only did the act provide for grants for the purchase of integrative technology, such as simple com- munications and dispatching systems, but it also provided grants for the establishment of EMSS agencies whose job would be to plan and coordinate the development of system-wide approaches to EMS delivery. However, the EMSS program is based on the assump- tion that at the termination of the grant programs, EMSS agencies will achieve financial self-sufficiency through state, local, or other federal funding sources. Whether the public and quasi- public EMSS agencies will actually be able to achieve financial independence and maintain their role in the health delivery net- work remains to be seen. Third-party reimbursement policies,
43 which at present do not generally recognize system-wide activities and do not provide a level of reimbursement sufficient to cover the costs of operating such a system, could play a major role in this regard. Specialization By specialization we mean the tendency of the health care system to function in specialized settings using increasingly specialized physicians and nonphysicians. Physician specialization in par- ticular has a profound influence on the adoption and use of equipment-embodied technology, although it is not clear whether increasing specialization has caused or resulted from the in- creasing complexity of medical technology. More study is needed of the relationship between technological change and physician specialization. The increasing specialization of physicians in the United States is well documented. The ratio of specialists to general practitioners has increased, and the number of subspecialities has increased. This phenomenon has been observed even within those specialties oriented toward primary care, such as internal and pediatric medicine.1^9 What relationship does this observed trend in specialization have to technological change? It appears that some specialties or subspecialties have developed in response to and around the introduction of new equipment-embodied technology. Many examples of new procedure-oriented subspecialties can be cited: cardio- vascular surgeons; nuclear medicine radiologists; ultrasonog- raphers; and, most recently, gastroendoscopists. Although there is no empirical evidence to support the finding, it is possible that the financial rewards inherent in specialization have en- couraged the adoption of these technologies. Evans25 theorizes that increasing specialization is largely the result of an excess supply of physicians, which encourages physicians "to use tech- nologies which enable them to provide more and different services to patients (particularly if the costs of the services are borne in a hospital budget)." Wessen139 states that: the fee-for-service system encourages the use of more and more specific procedures to which charges can be attached. . . . And the tendency of our reimbursement systems to value specific technological procedures more highly than generalized professional consultations confirms the eco- nomic bias toward the use of specialty services.
44 As the demand for specialists has increased, so has the supply of residencies offered by graduate teaching hospitals. Hadley and Reinhardt48 hypothesize that hospitals provide residency spaces in order to assist in the production of patient care, and that the demand for residents in a specialty service of a hos- pital is responsive and secondary to the demand for the medical services that residents supply. And, because graduate medical education is financed largely through third-party reimburse- ment,52 the ability of hospitals to provide financial support for residents is also a function of patient demand for the spe- cialty services. Sloan111 has tested a similar hypothesis for ophthalmology residency positions and has found that the number of positions offered is negatively related to the stipends hos- pitals must pay residents. Thus, the introduction of new clin- ical or ancillary procedures sets in motion a chain of events leading to continued training and production of specialty physicians and, to the extent that physicians can influence the demand for such procedures, continued increases in their frequency of use. CONCLUSIONS This chapter has attempted to bring together our knowledge, both formal and informal, regarding the impact of different factors on the adoption of equipment-embodied medical technology by health care providers, particularly hospitals. Although there is significant research on the diffusion of health care tech- nology, with notable emphasis on hospital technology, this re- search has not measured the impact of factors that affect all providers and the general environment in which technology dif- fusion takes place. We are left with anecdotal experience, logical analysis, and judgment to ascertain the impact of policy- related factors on adoption and diffusion. The major factors include: (i) the system of reimbursing health care providers for medical services; (ii) the medical injury compensation system; and (iii) the organization of medical services, including dis- aggregation of providers and specialization of manpower. More study is needed to determine the true impact of each of these factors on the development, adoption, and diffusion of various kinds of equipment-embodied technology. In the judgment of this committee, the present system of third-party reimbursement is the primary factor creating sys- tematic biases for and against adoption and use of particular kinds of equipment-embodied technology. The reimbursement sys- tem encourages the use of clinical and ancillary hospital tech- nology and discourages the use of coordinative, system-wide
45 technology. The reimbursement system further encourages spe- cialization of manpower and facilities, which in turn has a distorting effect on the adoption and use of new technology. On balance, the hospital reimbursement system probably creates incentives to overadopt new technology with significant economies of scale relative to the size of individual providers. However, this kind of technology could also be subject to an underadoption problem in the absence of such a lenient cost-based reimbursement system, due to the disaggregation of providers. Were the reim- bursement system to be changed from a cost-based system, region- alization and sharing of services would have to be encouraged in order to induce providers to adopt new high-volume technology whose cost-saving potential lies in its economies of scale.