Medical Technology and the Health Care System: A Study of the Diffusion of Equipment-Embodied Technology (1979)

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.

APPENDIX THE COST OF CAPITAL-EMBODIED Q MEDICAL TECHNOLOGY* Kenneth E. Warner I. INTRODUCTION Of all the problems which constitute the medical care "crisis," none receives more attention than the consistently and rapidly escalating costs of personal health services, especially those associated with hospital care. Two decades ago, health expen- ditures totaled less than 5 percent of the nation's Gross National Product; today, Americans devote over 8.5 percent of GNP to health. The cost of a day of hospital care grew by more than l,000 percent from l950 to the present, while general consumer prices rose only l25 percent. In the public sector, the health share of the federal budget has risen from half a percent 20 years ago to more than 8 percent. With consump- tion of services increasingly freed from direct financial liability through the vehicle of insurance, and with the supply of services functioning with only limited regulation and controls, there is no clear end in sight to the problem of medical cost inflation.1 *The Committee on Technology and Health Care has elected to use the prase "equipment-embodied technology." The use of the word "capital" rather than "equipment" simply reflects my personal preference. No substantive distinction is intended. Carter administration's proposal to limit hospital cost increases represents a potentially significant initative. As noted below, some analysts believe that inflationary pressures are subsiding of their own accord (McMahon and Drake, l976) . 270

27l The search for a villain in this economic drama has included both people, most commonly physicians, and inanimate objects, especially costly and sophisticated capital equipment. The profusion of such equipment is certainly one of the dominant characteristics of modern medical care. A federal official has concluded that "the long-term cumulative effect of adopting new health care technology is a major cause of the large yearly increases in national health expenditures. ..." (Gaus, l976, p. l2). This conclusion derives from both theoretical reasoning and empirical evidence. The thrust of the theoretical argu- ment, developed in the next section of this paper, is that the environment in which technology adoption and use decisions are made is relatively unconstrained by conventional economic factors and, in fact, is conducive to the adoption and use of technology. The empirical evidence comes from analyses of the causes of hospital cost inflation and from numerous individ- ual case studies, discussed later in this paper. Both of these sources suggest that capital-embodied technology con- tributes significantly to medical cost increases, though no evidence places it as the principal factor, and it should be emphasized at the outset that the empirical evidence on the costs of technology is largely anecdotal and far from defini- tive. The costs of medical technology are numerous and diverse in kind. I will focus principally on direct economic costs— the capital and operating costs associated with capital-embodied technologies. This means that two important categories of costs will be generally ignored. l. Indirect economic costs consist of (a) work productivity losses resulting from application of the technology (productivity gains would be indirect benefits) (Cooper and Rice, l976); and (b) the costs associated with additional medical care services consumed as an indirect result of use of the technology (ies) in question (e.g., the normal medical needs of patients cured or maintained alive who would have died in the absence of use of the technology); and the "negative costs," or benefits, of resources whose use is avoided by use of the technology in question (e.g., CT scanning substituting for pneumoencephalog- raphy).2 2These costs are included implicitly in several studies of the role of technology in health care cost inflation, particularly studies that measure technology as a residual. (These are discussed later.) However, such costs will be neither measured nor even considered explicitly in this paper.

272 2. Noneconomic costs are also ignored. These range from the mundane (e.g., pain) to the subtle and profound, such as: the impact of technology on encouraging physician and nurse specialization, which in turn has complex implications for the delivery of care; the role of "miracle technology" in increasing the demand for'care in general; a reduction in self care (e.g., good personal health practices) resulting from greater reliance on the medical care system, due to a perceived improvement in the efficacy of medical technology; the ethical costs and questions associated with artificial organs, genetic surgery, and so on. While our knowledge of the costs of medical technology is limited, it must be emphasized that even complete knowledge of such costs would not answer the question, Do we have (or use) the "right" amounts and kinds of technology? Cost is only one. side of the economic equation. The marginal cost of a technology must be compared with its marginal benefit to determine optimal usage. While this principle is frequently acknowledged, separation of consideration of benefits and costs pervades the technology literature: Medical journals are replete with analyses of the diagnostic or therapeutic benefits of technologies, with little if any attention to costs, while the health economics literature emphasizes the costs of care and too often ignores the benefits. In the present context, the importance of this is that we need to distinguish cost increases that reflect improvements in care from those that are purely inflationary, i.e., higher prices for the same care. The former may well be tolerable, even desirable, while the latter are clearly undesirable. 1*'5 SMahler (l975) argues that expensive medical technology dic- tates too much of the structure of care delivery systems, and that this technology is one reason that we have "medical care systems," rather than "health care systems." Illich (l975) sees technology as contributing to the "medicalization of society. " view that new technology, though costly, is responsible for saving many lives that were necessarily lost before has been referred to as the "benign theory of hospital inflation" (Davis, l972, p. l359). 5There is little doubt that there is considerable waste in the practice of medicine, e.g. , inappropriate or excessive use of lab tests, whether intended (e.g., for profit or to protect against malpractice) or unintended. It has been estimated that a l0 percent improvement in medical resource utilization efficiency would save more than $5 billion (DeJong and Shaw, l975).

273 The essence and importance of this distinction are captured by Gaus and Cooper (l976, pp. 3-4): [W]e spent 4 billion dollars for new technology [for Medi- care patients in l976] and we do not know if it did any good, much less how much. . . . ... If we had continued providing hospital services to the aged, as they were in l967, then we could have spent that 4 billion dollars last year [to] . . . have + Brought all aged persons above the poverty line [with at least 3.3 million currently living below it]; or + Provided the rent to raise 2 million elderly from substandard to standard housing units; or + Brought all the elderly above the lowest accepted food budget and more; or + Provided eyeglasses and hearing aids to all who needed them [estimated at l8 million needing or wearing glasses and over 3 million needing hearing aids], and more. Which would have helped the most, [medical] technology or food? The profusion of expensive and sophisticated medical tech- nology is likely to continue for many years to come. Given the multiyear lag between the discovery and application of innovation, and the even longer lag from basic research to application, the growth years of NIH through the mid-l960"s and this decade's applied research and development work augur a "technological boom" in the decade to come (DeJong and Shaw, l975). The government's interest in this phenomenon, reflected in the work of the committee, is obvious: Under our current health care system, the government has an economic responsibil- ity in all phases of technology and medicine, from financing the initial research and developmental work to purchasing the technology both directly (e.g., through the VA hospitals) and indirectly (e.g., Medicare). In addition, many believe that the government has a role as "protector of the public" to assure the safety and efficacy of" medical technology (Office of Technology Assessment, l977), a role which has been acknowl- edged legislatively at the federal level recently with the passage of the Medical Devices Amendments of l976 (U.S. Congress, l976). The remaining sections of this paper examine what we know, and how we might learn more, about the pecuniary costs of capital-embodied medical technology, from both a theoretical and an empirical perspective. Section II describes the environment in which medical technology adoption and use

274 decisions are made, suggesting why it is often asserted that the medical system encourages excessive use of technology. The third section defines "capital-embodied medical technology" and the costs of such technology, to serve as a basis for discussion in the ensuing section. Section IV reviews the theory and practice of allocating medical care costs to capital- embodied technology and suggests future directions and data sources that might be exploited to increase our knowledge. The fifth section presents a brief summary and conclusion. II. FACTORS INFLUENCING THE ADOPTION AND USE OF MEDICAL TECHNOLOGY Many observers of American health care believe that there are relatively few constraints on the adoption and use of medical technology by hospitals and physicians. To the contrary, there appear to be positive inducements—both pecuniary and nonpecuniary—favoring adoption and use. As a consequence, the argument runs, we purchase and use too much technology. Furthermore, the economic and professional environment encourages the development and utilization of cost-increasing technology. In a predominantly fee-for-service system charac- terized by widespread insurance coverage, there are few incentives to develop or employ cost-saving technology. Under- standing this environment helps to explain the concern about the excessive use of technology in medicine and the resultant costs, and it places the cost-of-technology issue in perspec- tive as a factor in overall medical cost inflation. This section reviews both noneconomic and economic factors affecting the adoption and use of medical technology. Noneconomic Factors The physician, it has been claimed, is driven by a "technolog- ical imperative" instilled during medical training (Fuchs, l973). Medical schools are showplaces of modernity, and the student's role model, the medical faculty member, is engaged in attempting to advance the scientific frontiers of medicine, owing in large part to the government's support of biomedical research and development. Thus the image of quality medicine with which a medical student leaves training is predicated on a scientific approach to problems, with modern technology constituting the instruments with which that approach is practiced. As is discussed below, the economic system does nothing to discourage the technological imperative in ordinary

275 practice, and indeed seems to reinforce it through relatively complete insurance coverage of the use of much medical technol- ogy. The existence of high-cost, hospital-based technology is considered a factor in the trend toward increasing physician specialization. Increasing specialization, in turn, reinforces the hospital's growing importance as a source of care. In addition, it increases the demand of physicians for still more technology (Rice and Wilson, l975). Possession of modern, sophisticated technology confers prestige on physicians (Feldstein, l97la; Newhouse, l970), and it may well contribute to their economic well-being (Pauly and Redisch, l973). The lay public seems to share the physicians' fascination with technology. We ". . . often look to [technology] to solve problems when less expensive solutions lie elsewhere. This may be particularly true of health care" (Fuchs, l973, p. 58). Growing faith in the power of science in general and curative medicine in particular accelerates the demand for technologi- cally advanced methods of care (Feldstein, l97lb). Physicians' and the public's technology orientations influ- ence the decision making of hospital administrators. It is commonly asserted that administrators acquire sophisticated capital equipment and facilities in order to attract and hold high caliber physicians on their staffs (e.g., Davis, l972; Muller and Worthington, l970). A recent study found evidence to support this assertion (Abt, l975).6 In addition to using technology to compete for staff positions, "adminis- trators themselves derive utility from having the best equipped and most modern facilities" (Davis, l972, p. l358). In the role of the patient's demand agent, the physician is obligated by the "social contract" to provide the "best possible care" (Arrow, l963). In medically desperate situations, °Some observers argue that equipment acquisition decisions, though formally lodged in the administrator's office, reside in fact with medical department chiefs, and that "The opera- tional reality resulting from the physician's policy control has been a physician dominance of the resource allocation function. Consequently hospital investments are primarily evaluated along medical dimensions: rarely are economic con- siderations integrated into the investment evaluation function (Lusk, quoted in Dittman and Ofer, l976, p. 27). Of course, the ability of physicians to make resource allocation decisions with only limited attention to economic considerations is a function of the permissive economic environment discussed below.

276 this implies using the technically best therapy that is avail- able. This may lead to very early adoption and diffusion of a technological innovation (Warner, l975), though diffusion of an unproven medical technology is not restricted to medical crisis situations (Altman .and Eichenholz, l976; Gaus, l976; Gaus and Cooper, l976). Early and extensive diffusion is en- couraged by medicine's protreatment bias and a permissive economic environment. Obviously there is a wide array of innovations from which to choose, due in part to the heavily governmentally subsidized biomedical R&D sector (especially NIH). The direct federal contribution to applied research and developmental work in the medical equipment area is small though growing—most of such R&D is funded by private industry—but for decades basic research has been the ward of the state. This research has contributed both directly and indirectly to the pool of medical technology. Governmental involvement in the medical technology arena can promote the development and diffusion of capital equipment, as does its support of research, but it can also restrict equipment production and use, principally through regulatory policies. Regulation is alternately viewed as a means of compensating for the unorthodox economic environment in which technology adoption and use decisions are made, or as a mech- anism to protect the public against ignorant or unscrupulous users of technology. In some instances, regulation is con- sciously directed toward limiting the spread of "unnecessary" equipment (for example, certificate of need); in others, the objective is assuring high-quality equipment and its appropriate use (e.g., FDA's new medical device regulation [U.S. Congress, l976]). The effects of regulation on technology adoption and use are uncertain, though the available evidence is not encouraging: Where regulation is intended to limit the spread of medical capital, it appears to be reasonably ineffective. For example, where certificate of need has succeeded in limiting growth in hospital bed supply, purchase of other equipment has increased, resulting in no overall savings in capital expenditures (Salkever and Bice, l976). In contrast, the new medical device regulation procedures, which are intended only to assure the safety and efficacy of medical services, have raised the fear that "overregulation" will stifle entrepreneurial initiative and thus reduce the discovery and production of new safe and efficacious devices. ?The capital purchase (CAP) proposed in the administration's cost-containment bill would be expected to have a significant effect on capital expenditures in the future.

277 It has been argued that FDA's regulation of drugs, on which the device procedures are modeled, has had a deleterious effect on drug R&D (Peltzman, l974; Grabowski, l976). With the possible exception of the regulatory mechanisms, the above forces combine to produce a noneconomic environment that is favorably disposed to the adoption and use of modern, sophisticated technology. For a variety of reasons, both economic and noneconomic, such technological change seems biased toward the cost-increasing variety (Feldstein, l97lb). Economic Factors The economic environment of medical care provides some positive incentives and few disincentives to adopt and use technology. Beginning with the subsidization of research and development, the government pumps considerable money into medical schools and elsewhere to encourage the development of new knowledge and technical innovations. But the most salient feature of the medical technology market is the mixture of the seller's profit incentive (e.g., device manufacturers) and the buyer's relatively unconstrained position. For hospitals operating on a cost or cost-plus reimbursement basis, "A quality-enhancing or service-expanding project looks just as good as a cost reducing project in terms of the result- ing cash inflow from . . . reimbursement of depreciation charges" (Silvers, l974, p. 294). In short, there is no economic in- centive to adopt resource-saving technology, at the same time that the noneconomic forces favor adoption of sophisticated and generally costly equipment.8 Some observers even argue 8Given this intuitively appealing logic, it is interesting that studies of the role of cost-based reimbursement have not found it to be a significant contributor to overall hospital inflation. The methodology of some of these studies leaves their findings open to question. For example, Pauly and Drake (l970) simply used a dummy variable to indicate hospitals that were in states in which Blue Cross reimbursed on the basis of costs, thus ignoring the proportion of costs covered by such a reimbursement scheme. Davis (l973) used a measure that captured the proportion. She did not find that cost reimbursement influenced costs within the range of proportions studied. However, she did observe significant increases in average costs in the Medicare period. While Davis proposed alternative explanations, she did not effectively rule out the influence of reimbursement.

278 that the availability of financing governs the rate of adoption of high-cost technology, with the technology's medical efficacy being of secondary importance (Rice and Wilson, l975), as may be demand or costs (Ginsburg, l972). Hospitals acquire capital through a number of mechanisms, including private philanthropy and government grants, borrow- ing, and internal generation of equity capital. None of these forces hospitals "to experience the real discipline of the capital markets" (Silvers, l974, p. 304). Philanthropy and grants provide "free" working capital; however, this source of funds, which once dominated capital budgets, now accounts for only l0 percent of hospitals' capital. Borrowing ability has been enhanced significantly by the tax-exempt status of many bond issues and by the safety factory associated with guaranteed reimbursement through major third parties. Nearly two-thirds of hospital capital funds is now derived from borrowing. Rate controls, inflation, and the limitation of cost-reimbursement to eligible costs have restricted the internal generation of equity funds (Blume, l976). The use of funds to acquire equipment and facilities has been aided by a growing trend toward leasing, which "can serve as a hedge against obsolescence, and . . . affords the hospital greater flexibility in replacing its equipment" (Ofer, l976, p. 5l). 9 But whether through leasing or purchase, hospitals have incentives to overinvest: If interest payments are reimbursed, the effective interest cost to buyers is lower than the true interest. Thus more marginal projects may look acceptable to the purchasers. Also, frequent upgrading of existing services and addition of new ones gives providers greater leeway in the allocation of overhead. Most cost-based reimbursement schemes probably allow considerable latitude in this area (Silvers, l974). For the consumer, increasing insurance coverage and affluence have significantly reduced the real direct (or out-of-pocket) cost of much medical care, especially that provided in hospi- tals.10 Feldstein (l97lb, l977) sees insurance as having 'For many kinds of medical equipment, obsolescence rates appear to run 5 to 8 years (Abt, l975). 0Patients now pay less than one-eighth of the average hospital bill directly, compared with one-half in the early l950's. In addition, increases in real income over the period mean that patients must now work fewer hours to pay the direct cost of a day of hospital care (Feldstein and Taylor, l977).

279 increased the demand for care, particularly for "style" and "high quality." The hospital administrator's response has been "improvements," including the acquisition of fancy technology, which have driven costs up. Completing the circle is the consumer's response to the higher costs—namely, to buy more insurance. "Thus as third party payment has increased over the years, the benefit required to justify a decision in the eyes of doctors and patients has declined. This has led to increased use of resources in all sorts of ways—including the introduction of technologies that otherwise might not have been adopted at all and, more often, the more rapid and extensive diffusion of technologies that had already been adopted to some extent. ..." (Russell, l977, p. 3.) Empiri- cal findings are becoming available that provide evidence that the ability of patients to pay for services, and not simply illness and need, influences the complexity of hospital care (Abt, l975). (See also Russell, forthcoming.) Finally, the efforts of suppliers of medical technology can be credited in part with what some perceive to be rapid and indiscriminate adoption of technologies (Fuchs, l973). While the purchasers of technology operate with limited and certainly unorthodox economic constraints, most of the sellers are conventional profit-seeking firms.11 The situation is not unlike (though not identical to) that of fee-for-service physicians and their patients. It is often claimed that fee- for-service physicians overutilize tests and other services, frequently involving capital-embodied technology, because they will produce profit for the physicians while not (directly) affecting well-insured patients. "In short, when those making the decisions pay none of the cost, resources are used as though they cost nothing" (Russell, l976c, p. 3). The unusual financing relationships that define an economic transaction in medicine have been the subject of many studies, though relatively few have focused on the area of expensive capital equipment, where the potential impact seems especially great. All of the elements come together here to 11The profit potential of the medical market is suggested by the fact that pharmaceutical companies spend over $4,000 per year per physician in advertising and promoting their products (Fuchs, l974). Producers of scientific and medical instruments spend on advertising an amount equal to about 2.5 percent of sales, com- pared with an average rate of l.4 percent for all manufacturing corporations. While the technical or marketing complexity of scientific equipment might require greater advertising, the figures are suggestive of high profit potential (Peterson and MacPhee, l973).

280 produce a situation in which the binding constraint may be the state of the art, i.e., the technology itself, and not, as elsewhere, considerations of all costs and benefits. "[T]he present system allows expenditure, especially for hospital care, to be made for little or no return, when the fact that the resources available to the- economy as a whole are limited forces expenditure on goods and services other than medical care to meet a higher standard. Given the concern about the effects of life style and environment, this does not make sense even if our only goal was to improve the health of the population" (Russell, l977, p. 9). III. DEFINITIONS AND CONCEPTS In order to discuss the allocation of medical costs to capital- embodied technologies, we must define terms more precisely. As this section will suggest, the need for definitions reflects the somewhat arbitrary distinctions created by separating "capital-embodied" technology from all technology and the costs of the former from all operating and capital costs. It is obvious from the Charge to the Committee (memorandum of September 29, l976) that the Committee acknowledges these distinctions. This section should clarify their importance. Capital-Embodied Medical Technology A technology is a defined configuration of inputs used to produce a specified output, either a physical good or a service. Thus a medical technology is a configuration of inputs used to produce a specified medical output. A medical technology may be as simple as a nurse's taking a patient's temperature, or it may be as complex as the combination of surgical specialists, nurse specialists, technicians, supplies, sophisticated equip- ment, and surgical suite used to perform open-heart surgery. In general, medical technologies may be classified by their physical nature12 or by their medical purpose.13 12Congress's Office of Technology Assessment identifies technique (the action of a provider without specialized equipment), drug, equipment (machines and smaller devices and instruments), and procedure (a combination of technique with drugs and/or equip- ment) (Office of Technology Assessment, l976b). 13For example, preventive (or health maintenance), diagnostic, therapeutic, rehabilitative, organizational (management and administration), and supportive (Behney, l976; Office of Tech- nology Assessment, l976b).

28l A capital-embodied medical technology is a medical technol- ogy that utilizes, as one of its principal inputs, the services of one or more pieces of capital equipment. Capital equipment refers to durable items that were not consumed in one (or a few) use(s), but rather that provide services over an extended period of time. This can include anything from a pencil sharp- ener to a computer. In most instances in what follows, I will be using the term to refer to reasonably expensive, major items. (Obviously, to develop an operational definition one would have to define the minimum cost, quantity, durability, and so on.) One might justify the distinction between capital-embodied and other technology, or between expensive major capital equipment and other capital equipment, simply on the grounds of public or medical interest. For example, the CT scanner has created quite a sensation among the lay public, as well as within the medical profession. Perhaps that interest is sufficient to warrant an independent consideration of this innovative technology. However, other than the interest factor, the relevance, or nonarbitrariness, of making defini- tional distinctions between, for example, "major" and "minor" capital equipment must reside in either (a) a technical difference in the circumstances under which major and minor capital equipment are produced, distributed, used, or paid for; or (b) differences as to how such equipment is treated under the law. There is little evidence on the former. How- ever, there are legal distinctions in that, for example, some certificate-of-need laws apply only to equipment or facilities costing more than a specified amount (e.g., $l00,000). Other- wise, the distinction remains arbitrary, and it will introduce difficulties in cost analysis, as described in the next section. The Cost of Capital-Embodied Medical Technology The first and most sensational cost of much medical technology is the capital cost. However, it is generally agreed, and empirical evidence demonstrates, that the most significant costs of major capital equipment derive from the ancillary personnel and supplies needed to use the equipment (e.g., Abt, l975; Ginsburg, l976). Much new medical capital technology is not labor-saving; to the contrary, it creates the need for additional and often specialized technicians and nurses. Thus the true direct costs of a major piece of capital equipment are its fixed (capital) costs (and associated financing costs) plus maintenance and the variable costs associated with its use. This is implicit in referring to "capital-embodied medical

282 technology" rather than just to capital equipment: The latter is simply one component of the technology, one that is useless in the absence of the other variable inputs. The problem in measuring the true costs lies in defining the technology in a systematic and meaningful manner. Is "radiotherapy" limited to the actual administration of radi- ation, or does the technology of radiation therapy include the supportive services that precede and follow the irradiation? Where are the lines drawn? This question is central to identifying the direct costs of medical capital-embodied tech- nology. As is discussed below, narrowly construed, capital- embodied technology probably does not impose a sizable cost burden on the medical care system; broadly construed—that is, to include all services and procedures that have a significant link with the use of capital or with another procedure depen- dent on capital—such technology is probably enormously expensive. A theoretical resolution of the problem is to ask what would have happened to the patient in the absence of the capital-based technology, and then to compare the costs associated with that hypothetical management of the case with that which actually occurs. The difference would be a measure of the direct costs of the capital-embodied technology. The difficulty in making this concept operational is one reason why so little empirical work has been attempted along these lines. It is important to recognize, however, that even if the concept could be translated into effective quantification, it would miss some subtle factors that should be included in a measure of the costs of the technology; for example, the case mix may have changed in response to the existence of the technology (Rice and Wilson, l975); or there may be more cases amenable to treatment by the new technology than would have been identified in the absence of the technology.14 It is also important to distinguish three sources of high cost due to technology. One is simply the existence of a high-cost technology that also produces a significant benefit. Here, the question of the desirability of the technology is one of comparing benefit and cost. A second source is excess- ive use of the technology. A technology with a cost-reducing potential (e.g., it substitutes for a technology that is more costly on a unit or per-use basis) may actually increase costs because it is used so often. The third source has frequently been identified as an important inefficiency in medicine: excess capacity or underutilization of expensive technology. example, the existence of the CT scanner may significantly alter the amount, timing, and possibly nature of brain surgery.

283 In its study of Boston area hospitals, Abt (l975) found 50-60 percent of capacity utilization of autoanalyzers, diagnostic X-ray machines, and patient monitors. A l972 survey found that 62 percent of hospitals equipped with open-heart surgery facilities did fewer than l00 operations annually (Roche and Stengle, l973), while a professional Surgery Study Group had recommended 200 procedures as the minimum number necessary for surgical teams to maintain their skills. In l967, 3l percent of equipped hospitals had not used their open-heart facilities for at least a year (U.S. DHEW, l97l). The cost of underutilization is suggested by the estimate that the short-run cost of maintaining an unused hospital bed is roughly two-thirds the cost of an occupied bed (Altman and Eichenholz, l976). A potentially significant cost of excess capacity may be the added incentive to use the underutilized technology in more marginal or questionable cases. Finally, to repeat, an analysis of the direct costs associ- ated with the technology will miss numerous indirect costs resulting from the availability and use of the technology. The latter could conceivably have a much more significant economic and/or social impact. Technology's Costs and Its Contribution to Cost Inflation A final distinction is between the costs associated with the stock of capital equipment—e.g., the equipment in use during l977—and changes over time in the costs of medical care resulting from additions to and changes in the nature of the stock of equipment. An understanding of this distinction and separate measurement of these different costs are essential to sound policymaking. If it were learned that the existing stock of capital equipment does not impose unreasonable cost burdens, but the flow into the system appeared to be an excess- ive rate, policy should focus on research and devlopment and the diffusion of innovations (e.g., through biomedical R&D funding, FDA certification of devices, etc.). If by contrast 15 1n many instances, "excess" capacity may be justified on the basis of option demand. That is, we are willing to pay a price (i.e., the costs of unutilized capacity) in exchange for the certainty of the ready availability of the technology whenever it might be needed. While option demand is a legitimate basis for unused capacity, the amount of excess capacity often doc- umented considerably exceeds that which option demand would recommend.

284 the stock of capital equipment seemed to be contributing excessively to the current costs of care, while the rate of change in the stock was low or moderate, policy might more profitably focus on the use end of the spectrum (e.g., on reimbursement policies). In the remainder of the paper, I will consider both the costs of technology and technology's contribution to inflation, though the reader should be fore- warned that, as a reviewer of the literature, I will be unable to draw confident conclusions about either. IV. ALLOCATING MEDICAL CARE COSTS TO CAPITAL-EMBODIED TECHNOLOGY The preceding section discussed both conceptual approaches to and problems in allocating medical costs to capital-embodied technology. The first subsection below will extend that dis- cussion by considering methods of measuring costs, with the next subsection presenting the results of attempts at such measurement. Methods of Measuring Costs Case Studies. The first and most obvious approach to measuring the costs of a capital-embodied medical technology is to undertake a case study focusing on a specific technology. A capital-embodied technology is identified and its components defined precisely. The actual costs of inputs are then measured, including reason- able allocations for depreciation, imputed prices where neces- sary, and so on. These are then summed and, ideally, compared with the costs that would have been incurred in the absence of the technology (i.e., using the best alternative technology, if any.16 While this approach may yield excellent estimates of the cost of a given technology, it does not (inherently) permit generalizations or conclusions about the overall cost of capital-embodied technology.17 One must be wary not only 16if benefits from the two technologies differ, they too need to be compared. A benefit-cost or cost-effectiveness analysis might be performed. 17If sufficient uniformity emerged from numerous case studies, one might draw inferences of a larger or more general nature. This is considered in the discussion of the next method and is done, illustratively, later in the paper.

285 of the quality of the studies, but also of the possibility of selection bias. A second case study approach involves adopting treatment of a specified illness as the basis of analysis, rather than beginning with the technology. Here it is assumed that the output (outcome) remains constant, and variations over time in the costs of treatment are measured. Assuming that the treatment in question has a significant capital input, this approach has the appealing logic of defining a technology by a health outcome. The method is expensive, and again it suffers from the usual limitations of case studies (Scitovsky and McCall, l975). A third approach is similar to the first two, though it attempts to introduce broader coverage of hospital (or medical) activities. "Capital technology" is defined (for example, all pieces of equipment with a capital cost greater than or equal to $l00,000) and then analysts specify all related capital-embodied technologies. These are then costed-out in a (rather heoric) manner analogous to that of the first approach. Alternatively, one might focus on a capital-intensive depart- ment or unit (e.g., radiology) and simply modify its budget appropriately, attempting to link inputs to specific services. Estimation Based on Case Studies. A second method combines the first and third case study approaches to produce a rough estimate of the systemwide costs of capital-embodied technology. As in the preceding approach, capital technology is defined and then the budgets of a repre- sentative sample of hospitals are studied to determine direct expenditures on such capital. Rather than attempting to specify the precise production functions, the analyst applies a multiplier to the capital costs, where the multiplier is derived from case studies of the first type. That is, suppose that examination of numerous case studies of individual capital- embodied technologies suggests that the average ratio of total costs to the capital costs alone equals 3. Then the represen- tative hospital's direct expenditure on capital can be multiplied by 3 to estimate the costs of capital-embodied technologies. Obviously this method can provide no more than a ballpark estimate of the desired figure, but in so doing it offers a glimpse of the order of magnitude of the cost of technology. Clearly the credibility of the method depends on the quality and variability of the findings of the case studies.

286 Proxy Methods The great difficulty in finding a systematic approach to answering the question, what are the total direct costs of capital-embodied technology, reflects the lack of a direct measure of "technology" with which one could isolate technol- ogy's costs from other medical costs (Davis, l974). Several researchers have attempted to estimate the total "technology effect" by use of proxies. One approach is to include time in an econometric analysis of inflation in the cost of hospital care, where time is interpreted as representing technological change (Davis, l974).18 While this may provide order-of- magnitude estimates, time can also capture several other factors (Wagner and Zubkoff, forthcoming). In addition, this approach is not intended to distinguish capital-embodied from other technology. A similar problem arises in another technique that has been misinterpreted as providing an estimate of the contribution of technology to inflation. The cost effect of increases in nonlabor input intensity can be established by controlling hospital inflation for unit price increases (wages and nonlabor prices) and the quantity of labor. The residual is the effect of increasing nonlabor inputs (Feldstein, l97lb; Waldman, l972). While this is quite substantial, nonlabor inputs include nondurables and low-cost capital and exclude the labor associ- ated with capital-based technologies. As such, this bears no relationship to the effect of capital-embodied technology on costs. This is discussed further below. Empirical Cost Studies The empirical evidence makes it clear that "if major movable . . . equipment constitutes an important source of hospital cost inflation, it must be because of the complementary inputs required to operate and service it, for total equipment pur- chases amount to less than five percent of hospital costs on the average" (Abt, l975, p. 9). Ginsburg (l976, p. l74) echoes this theme: "[I]nterest and depreciation are a small part of hospital costs and cannot be considered an important cause of cost increases directly. However the indirect effects may be substantial." 18The inclusion of time as a proxy for technological change is a common if not universally accepted practice in economic studies of productivity growth.

287 Case Studies The importance of the noncapital costs of capital-embodied technologies emerges clearly from numerous case studies, despite the fact that few of these carefully examine all of the variable costs associated with use of the capital equip- ment. An instance in which it is relatively easy to measure operating costs occurs when "the technology" is a hospital unit; for example, coronary care units (CCU's). Bloom and Peterson (l973) report that the costs of building and equipping all of the CCU's in Vermont, New Hampshire, Massachusetts, and Rhode Island in l970 was $3.7 million, but the annual operating cost of these units was $6.4 million.19 Russell's (l976a) study of intensive care units (ICU's) found that the capital costs of equipping the units could be relatively small (as little as $2,000 per bed in the late l960's20), but ICU beds' operating costs run 3 times those of ward beds. With ICU and CCU beds totaling 5 percent of all short-stay hospital beds in l975, approximately l5 percent of l975 hospital costs can be attributed to intensive care. Two-thirds of that total represents an excess over what would have been necessary to maintain the same number of ward beds. Given the questionable value of intensive care for many patients (Gellman, l97l; Mather et al., l97l), one must ask how much of that additional expense was necessary. The rapid diffusion of CT scanners provides a classic example of the medical profession's ability to quickly adopt an expensive new technology of unproven value. Undoubtedly a significant technical breakthrough in diagnostic capability, the scanner's potential effects on human health are unestab- n .I lished.*1 The effects on human wealth are becoming quite clear. In l975, $l50 million worth of CT equipment was ordered or installed in the United States. The purchase 19Bloom and Peterson also observe that half of the patients in the CCU's had not had myocardial infarctions and were low-risk patients. O A -"In practice, per-bed costs at times often ran 4 or more times greater than the $2,000. 21While earlier, more accurate diagnosis might lead to decreases in mortality or morbidity, it might produce increases, e.g., by promoting risky surgery.

288 price of a scanner is in the vicinity of half a million dollars. Operating costs are estimated at an average of $300,000 to $400,000 per year. The costs of excess capacity, which already seems to exist in several major metropolitan areas, is illustrated by a calculation that an increase from 40 to 80 examinations per week—i.e., a doubling of the number of patients—increases total costs only l4 percent (Office of Technology Assessment, l976a). Other case studies suggest staggering costs associated with single technologies. The Public Health Service projected that the fifteenth year costs of kidney disease treatment would equal roughly 4 percent of all U.S. health service costs (Gellman, l97l). We currently spend over l2 percent of the nation's health dollars on clinical laboratory services, some $l5 billion in l975 (Banta and McNeil, l977). X-ray services cost almost $5 billion in the same year (Office of Technology Assessment, l976b). In l970 there were some l7 million skull X rays (from a total of 4.2 million examinations, each having multiple films) at a cost in excess of $l20 million. This is a procedure generally of dubious value, which is often per- formed solely to protect against malpractice charges (Office of Technology Assessment, l977). Electronic fetal monitoring adds $35 to $75 to the cost of a delivery. Thus, electroni- cally monitoring all deliveries would cost over $200 million. This estimate ignores the costs, both monetary and otherwise, of the additional caesarian sections that result from electronic 2 Q monitoring (Office of Technology Assessment, l977). If operations are included in the category of capital- embodied technology (due to the capital equipment in operating rooms), then some common surgeries of unclear necessity add immensely to the costs of technology. Every student of medical care is familiar with the current debate on unnecessary surgery (Committee on Interstate and Foreign Commerce, l976) and the link between the supply of surgeons and the amount of surgery that is performed (Bunker, l970). Most experts believe that hundreds of thousands of tonsillectomies and hysterectomies are done needlessly each year, with the attendant risks to the patients. Over $350 million is spent annually on 22The monetary cost implications are obvious. In addition, unnecessary X rays subject patients to unnecessary radiation dangers. 23The benefits of electronic fetal monitoring are unestablished. The most certain consequence has been a sizable increase in the number of caesarian sections.

289 appendectomies; yet recent evidence is causing clinical researchers to question the necessity of this standard operation (Office of Technology Assessment, l977). The effects of coronary artery bypass surgery are uncertain; but Americans spent over $650 million on this surgery last year, at more than $l0,000 per operation. A propo- nent of the procedure argues that the United States should prepare to do 80,000 coronary arteriograms a day to screen for coronary disease. The national bill for such a screening program would come to more than $l0 billion a year (Office of Technology Assessment, l977). If, as has been predicted, coronary artery bypass surgery becomes the most common surgery in the United States, with approximately 700,000 operations per year, the annual cost of this procedure would total over $7 billion (Neuhauser and Jonsson, l974). In one of the few studies to systematically examine the costs associated with several major pieces of hospital capital equipment,25 Abt Associates concluded that it is the capital's complementary inputs, such as labor and supplies, which con- tribute most significantly to hospital cost inflation. Total equipment purchases per se accounted for less than one-twentieth of hospital costs. However Abt did observe that total equip- ment expenditures had risen 23 percent per year from the mid- l960 's through the mid-l970's. Thus, "Weighting this percentage by the percent of equipment in total costs, [they] estimate the direct contribution of equipment expenditures to hospital per diem cost inflation at l.l5 percentage points. The per diem cost inflation in [their] sample of l5 Boston hospitals was l3.4 percent per annum over the same time period, implying that the direct contribution of the growing equipment-intensity of service provision to inflation was on the order of nine percent (l.l5/l3.4)" (Abt, l975, pp. 9-l0). This is probably the most systematic direct estimate of the contribution of capital cost to hospital inflation, but it ignores those "complementary inputs" that convert a piece of equipment into a usuable component of a technology. In 24The surgery seems to reduce angina pectoris, but this might be a placebo effect. Through an unorthodox controlled clinical trial, a placebo effect was established in the l950's for mammary artery ligation (Neuhauser and Jonsson, l974). 25These were: cardiac catheter labs, automated blood analyzers, patient monitors, diagnostic X-ray machines, computers, and dishwashers.

290 addition, as the authors observe, their sample of Boston hospitals had a disproportionately large number of teaching hospitals, which tend to purchase more capital equipment than nonteaching hospitals. Hence the above estimate likely over- states the contribution of equipment purchases to inflation nationally. Proxy Measures in Studies of Hospital Cost Inflation Employing a variety of proxies for technology, scholars who have studied hospital cost inflation attribute anywhere from 30 to 50 percent of that inflation to technology. It should be noted at the outset of this discussion that there is much confusion in the interpretation of the findings by proxy, and, while 50 percent would seem to provide a high upper bound, the true contribution of technology to inflation, particularly high-cost technology, may be less than the "low" estimate of 30 percent; clearly the contribution of expensive capital-embodied technology must be less than the total effect of technology. The upper-bound, or 50 percent, estimates are based on the studies that break up hospital cost inflation into labor and nonlabor components and price and quantity components within each of those (e.g., Feldstein, l97lb; Waldman, l972). Over the last two and half decades hospital wage rates have risen rapidly, hospital employment less so. Assuming that the Consumer or Wholesale Price Index adequately reflects the prices of nonlabor inputs in hospitals, the quantities of nonlabor inputs have been increasing rapidly. Overall, changes in hospital input prices account for slightly more than half the rise in hospital costs, with the remainder representing additional quantities of old and new equipment, supplies, and labor. Gaus and Cooper (l976, p. l) have labeled this "essen- tially the 'technology factor)." Others have picked up on such evidence, and interpretations of it, to echo the con- clusion that nearly half of recent hospital cost increases is due directly or indirectly to medical technology (Office of Technology Assessment, l976b). However, the only conclusion warranted by these data is a tautological restatement of what was calculated: Half of hospital cost increases has been due to unit price increases; the other half represents increases in the quantities of inputs. Much of the latter might simply

29l reflect changes in tastes or attitudes or behavior—changes in demand. Other estimates of the "technology factor" place technology's role in cost inflation between 30 and 40 percent. For the early Medicare period of l968 through l97l, Redisch (l974) claims to explain 39 percent of the rise in operating cost per adjusted patient day, and 42 percent of the rise in oper- ating cost per adjusted admission, as the result of "explosive growth" in a handful of physician-controlled medical services: pathology tests, nuclear medicine procedures, anesthesiology, pharmacy items, lab tests (inpatient and outpatient), radiology procedures (inpatient and outpatient), therapeutic radiology procedures, and blood bank units drawn.27 Davis (l974) uses time as a proxy for technological change in a study of cost increases from l962 through l968 in approximately 200 private, nonprofit hospitals. She suggests that her finding of a 2 percent increase in cost per year, controlling for demand and supply variables, is a reasonable upper limit on the effect of technological change. This translates into 38 percent of the predicted increase in expenses per admission. While plausible, Davis1 upper limit argument is not entirely con- vincing (Wagner and Zubkoff, forthcoming). In another study, Baron (l974) estimates that technology and case mix changes, together, account for 35 percent of the cost increase in adjusted patient days and 30 percent in adjusted admissions. Baron's contribution is to acknowledge the change in case mix that, while potentially occurring all the time, appears to have shifted significantly with the advent of Medicare and Medicaid. Unfortunately, he did not separate the technology and case mix factors. 26strictly speaking, this can legitimately be considered an increase in technology, if technology is defined simply to be inputs producing medical services. However, given the context of the word "technology" in this paper and in most current discussions on the topic, the 50 percent residual in cost inflation after controlling for unit prices should not be labeled the "technology factor." 27While Redisch's regressions do indicate that growth in these services was a significant factor, his attribution of all the explained variance to the service variables is unwarranted. His equations also include highly significant time dummy variables.

292 In general, hospital cost inflation studies single out in- creasing demand, promoted by growing insurance coverage, as the most important factor in inflation. Wage catch-up in the hos- pital industry through the decade of the l960's certainly in- fluenced inflation, but in no way can it be considered "the" explanation (Feldstein and Taylor, l977); and as noted above, the cost-plus reimbursement theory of inflation has received little support. While all of the major studies give credence to the demand-pull theory, it should be recognized that the na- ture of the demand increases is more complex than that which is normally construed as demand-pull. For example, Redisch's (l974) conclusion, that a significant amount of the inflation rate can be explained by increases in the utilization of physician- controlled medical services, suggests that physicians' demands may have been pulling costs up, and not simply patients' demands. The importance of recognizing the significance of demand is to distinguish those increases in input intensity that simply reflect increases in demand from those that represent technolog- ical change. Much of the concern with the costs of technology seems to focus conceptually on the latter, while alarm appears to derive in part from erroneously attributing both sources of actual cost increases to technological change. The obvious pol- icy significance is that demand-induced inflation may require a different policy approach than inflation produced primarily by technological change. Toward an Estimate of the Costs of Capital-Embodied Technology None of the inflation-based studies distinguishes capital- embodied from other technology; and none definitively captures technology alone. With the exception of the Abt study, no re- search has attempted to examine the costs of capital-embodied technology beyond the case study level, and the Abt findings re- late only to the costs of expensive capital equipment in a non- representative sample of hospitals. However, the literature reviewed above permits some tentative conclusions, or guessti- mates, as to the costs of capital-embodied medical technology. Based on the case studies (and recognizing the potential hazards of their nonrandom method of selection), we can assume that on average the operating (variable) costs associated with capital- embodied technologies at least equal and probably exceed the capital cost by a factor of 2, 3, or more. Using as a base Abt's estimate of the direct contribution of high-cost equipment ex- penditures to hospital cost inflation (8.58 percent), this sug- gests that, in the Boston area, capital-embodied technology may have accounted for between l7 percent and 34 percent of hospital

293 cost inflation in recent years, at least in the hospitals sampled by Abt. However, as these hospitals are disproportionately in- volved in medical education,31 and teaching hospitals consume more capital equipment than nonteaching hospitals, these estimates should be reduced if they are to represent the situation nation- wide. While it is no more than an educated guess, I would there- fore place the contribution of expensive capital-embodied technology to hospital cost inflation at less than a quarter, quite possibly considerably less. In any event, the procedure for estimating the contribution seems reasonable. All that is needed are data comparable to those Abt derived, but based on a representative sample of hospitals. A more systematic review of the case study literature might refine one's sense of the vari- able cost: fixed cost ratio. The cost of capital-embodied technology in a given year could be established in a similar manner. First one would need to de- termine capital expenses per year for major equipment and then this figure would be multiplied by l plus the variable-to-fixed cost ratio. Needless to say, determination of the latter is cen- tral to the overall estimation, and the ratio is clearly sensi- tive to how broadly technology is defined. Based on Abt's findings that the rate of growth in capital inputs exceeds the rate of growth of other inputs, the cost of capital-embodied technology in a given year, as a percent of total cost, should be less than the percentage contribution of such technology to inflation. The conceptual work and data needed for further analysis are suggested above. Once concepts are agreed upon, data acquisition becomes simply a cost-accounting exercise, with analysts turning to actual hospital budgets to procure representative figures. A few states have good uniform accounting systems that might yield satisfactory numbers (e.g., California, Massachusetts, New Jersey, New York, Washington). In order to determine whether or not they will reimburse for particular capital items, rate-setting com- missions in at least three states (Washington, Connecticut, and New York) consider the noncapital costs related to equipment pur- chases (Gaus and Cooper, l976). Their cost-impact studies might provide useful data and suggest methodological approaches to performing analysis. Examination of the data in other central- ized systems might be fruitful, though few of these will be representative of national patterns of medical care (e.g., the 31Over half of the sample are teaching hospitals, while nationally only about l0 percent of hospitals are affiliated with medical schools.

294 Veterans Administration hospitals or Kaiser). A few public sources could be explored (e.g., the "Guide Issue" of Hospitals), though data in these are generally reasonably crude. Other po- tentially interesting sources of information include the manu- facturers of medical capital equipment and federal funders of technology RSD. And, of course, analysts would have to scrutin- ize the case study literature to determine a reasonable multi- plier. The perplexing problem is the starting point: conceptually and operationally defining "capital-embodied technology." This requires defining capital equipment in a meaningful way (e.g., choosing a capital cost minimum that corresponds to a legal requirement, such as certificate of need) and specifying bound- aries as to what constitutes a technology that is capital- embodied. Loosely defined, capital-embodied technology could encompass almost all of medicine; narrowly defined, it might be restricted to 6 or 7 percent of medical costs. The difficulty in arriving at an acceptable definition relates to the arbitrari- ness of distinguishing capital-embodied from other technology. While there may be a structural reason to make such a distinc- tion—for example, concern with equipment and facilities subject to certificate of need—there is no obvious logical reason. V. SUMMARY AND CONCLUSION To date, social policy with respect to medical technology has been fragmented and piecemeal, a reflection of the evolution of our medical care system. The perception of the need for a uni- fied policy reflects either a belief that there is something fundamental that distinguishes capital-embodied technology from other facets of care, or else a frustrated acceptance of the in- evitability of an incrementalist approach to contolling the sys- tem, combined with a belief that capital equipment is responsible for considerable medical cost inflation. As Wagner and Zubkoff observe (forthcoming, p. 6), "To the extent that the tendency to overinvest in and overuse sophisticated services is just a part of a larger tendency to overuse health services or to invest too many labor or nonlabor resources in the production of hospital services, the problem is not related to technology itself and should not be singled out as a technology problem." This is a perception which warrants serious consideration by policymakers. The literature sheds little light on the quantitative cost importance of capital-embodied technology; neither does it answer the question of whether the tendency to overinvest in and overuse sophisticated services is qualitatively distinct from a similar

295 general tendency in the provision and consumption of medical care. However, an assimilation of the numerous case studies (tempered by an awareness of selection bias) can be combined with the analysis of more general studies to draw some tenta- tive conclusions: • The high costs associated with capital-embodied technol- ogy result principally from the ancillary personnel and supplies required to use the capital equipment. These variable costs, which may include specialized labor (nurses or technicians), run from l to 3 or more times capital costs. • Adoption and use of capital-embodied technology is prob- ably a significant though not primary cause of hospital cost in- flation. A methodology was suggested as to how one might estimate both the current cost of such technology and its con- tribution to inflation. For high-cost equipment, as a guessti- mate I place the latter at less than 25 percent of the annual hospital inflation rate. The former would appear to be smaller still. • The failure of the literature to deal directly and thor- oughly with capital-embodied technology reflects the difficulty of defining medical production functions in a uniquely meaning- ful way. The same problem arises in considering medical tech- nology more generically. In part, this difficulty suggests an arbitrary element in defining and distinguishing these terms. • The true social cost of a capital-embodied technology, or of any medical care, cannot be evaluated in an output vacuum. That is, lacking knowledge of the benefits derived from a tech- nology, one cannot determine whether the technology's costs are purely inflationary or whether they represent a "good buy," pro- ducing a benefit of at least comparable magnitude. Still, theory suggests the existence of strong incentives in the existing medi- cal care system for overutilization of technology; case studies seem to confirm significant and widespread instances of this. • There are reasons to anticipate continuation of a "tech- nology boom." Given the multiyear lag between research and technology application, the generous finding of R&D a decade ago and the recent emphasis on technology development augur numerous capital-intensive innovations during the next several years. Part of the "technology boom" of recent years undoubtedly owed to the growth in private and public insurance coverage. For example, Russell (l977) documented sharp increases in ICU's and diagnostic radioisotope facilities in small hospitals with the advent of Medicare and Medicaid; electroencephalographs rose rapidly in larger hospitals. It can be argued that "the histor- ical pressures for hospital inflation have been greatly reduced

296 in the mid-l970)s. . . . [F]uture increases in demand for hos- pital services similar to those experienced before are unlikely to occur" (McMahon and Drake, l976, p. l36). With 90 percent of all direct hospital costs covered by insurance, it is probably correct that the economic pressure for increasing the quantity of services will be small. . However, this will not necessarily alter the growing demand for higher quality of services, with its attendant implications for the use of sophisticated tech- nology. The general concern is expressed by Gaus and Cooper (l976, p. ll): "Given our tremendous technology-induced costs, it is apparent that in the U.S. today we have neither the capacity to determine how much of what technology is appropriate nor the mechanism to control adoption." In the extreme, the fear is that the state of the art in technology is the binding constraint on medical costs. Unless we learn how to assess the cost- effectiveness of technologies and then how to implement effec- tive controls on adoption and use, we can anticipate that technology will continue to contribute to rapid inflation in the cost of medical care. REFERENCES AND BIBLIOGRAPHY Abt Associates, Inc. l975. Incentives and Decisions Underlying Hospitals' Adoption and Utilization of Major Capital Equipment. HRA Contract No. HSM ll0-73-5l3 (September). Altman, Stuart H., and Eichenholz, J. l976. "Inflation in the Health Industry—Causes and Cures." In Health: A Victim or Cause of Inflation?, pp. 7-30. M. Zubkoff, ed. New York: Prodist. American Public Health Association. l974. "Resolutions and Position Papers." American Journal of Public Health 64(2) (February):l73-96. Arnstein, S. l976. "Technology Assessment: Opportunities and Obstacles for Health Managers." Paper presented at the Second International Congress on Technology Assessment, Ann Arbor, Mich. (October 26) Arrow, K. l963. "Uncertainty and the Welfare Economics of Medi- cal Care." American Economic Review 53 (5) (December):94l-73. Arthur D. Little, Inc. l976. Automated Electrocardiography in the United States. DHEW Contract No. (HRA)230-75-02l2 (August). Banta, H., and McNeil, B. l977. "The Costs of Medical Diagnosis: The Case of the CT Scanner." Processed. Baram, M. l976. "Medical Device Legislation and the Develop- ment and Diffusion of Health Technology." Background paper

297 for the Conference on Health Care Technology and Quality of Care, Boston University (November l9-20). Baron, D. l974. "A Study of Hospital Cost Inflation." Journal of Human Resources 9(l)(Winter):33-49. Behney, C. l976. Studies of the Implications of New Medical Technologies. Staff paper for the National Council of Health Planning and Development, HRA, DHEW (August). Berry, R., Jr. l970. "Product Heterogeneity and Hospital Cost Analysis." Inquiry 7(l)(March):6-75. Bloom, B., and Peterson, O. l973. "End Results, Cost, and Pro- ductivity of Coronary Care Units." The New England Journal of Medicine 288(2)(January ll):72-78. Blue Sheet. l977. Drug Research Reports 20(l8)(May 4). Blume, F. l976. "Hospital Debt Management and Cost Reimburse- ments." Chapter 7 in "Financial Management Under Third Party Reimbursement." K. Bays, W. Breen, and A. Ofer, eds. Topics in Health Care Financing 3(l)(Fall). Bunker, J. l970. "Surgical Manpower. A Comparison of Opera- tions in the United States and in England and Wales." New England Journal of Medicine 282(3) (January):l35-44. Collen, M., ed. l973. Technology and Health Care Systems in the l980's. National Center for Health Services Research. DHEW Publ. No. (HSM) 73-30l6. Committee on Interstate and Foreign Commerce, House of Repre- sentatives. l976. Cost and Quality of Health Care: Un- necessary Surgery. Report by the Subcommittee on Oversight and Investigations. Washington, D.C.: U.S. Government Printing Office (January). Community Service Council, Inc., Alabama Regional Medical Pro- gram. l974. "Description of Initiatives and Alternatives for Containment of High Health Costs." Birmingham (August). Cooper, B., and Rice, D. l976. "The Economic Cost of Illness Revisited." Social Security Bulletin 39 (2) (February):2l-36. Davis, K. l972. "Rising Hospital Costs: Possible Causes and Cures." Bulletin of the New York Academy of Medicine 48(ll) (December):l354-7l. Davis, K. l973. "Theories of Hospital Inflation: Some Empir- ical Evidence." Journal of Human Resources 8(2)(Spring): l8l-20l. Davis, K. l974. "The Role of Technology, Demand and Labor Mar- kets in the Determination of Hospital Costs." Chapter l5 in The Economics of Health and Medical Care, pp. 283-30l. M. Perlman, ed. New York: John Wiley & Sons. Dejong, M., and Shaw, K. l975. "Issues in Medical Technology." Chapter 2 in Health Planning: A Systematic Approach, pp.25-59. By H. Hyman. Germantown, Md.: Aspen Systems Corp.

298 Dittman, D., and Ofer, A. l976. "The Impact of Reimbursement on Hospital Cash Flow." Chapter 4 in "Financial Management Under Third Party Reimbursement." K. Bays, W. Breen, and A. Ofer, eds. Topics in Health Care Financing 3(l)(Fall). Feldstein, M. l97la. "Hospital Cost Inflation: A Study of Nonprofit Price Dynamics." American Economic Review 6l(5) (December):853-72. Feldstein, M. l97lb. The Rising Cost of Hospital Care. Wash- ington, D.C.: Information Resources Press. Feldstein, M. l973. "Econometric Studies of Health Economics." Harvard Institute for Economic Research, Discussion Paper No. 29l (April). Feldstein, M. l977. "Quality Change and the Demand for Hos- pital Care." Written for the Council on Wage and Price Sta- bility, Executive Office of the President. Feldstein, M., and Taylor, A. l976. "Differences Among Hos- pitals in the Rates of Increase of Cost and Related Charac- teristics." A report to the Council on Wage and Price Stability (December). Feldstein, M., and Taylor, A. l977. "The Rapid Rise of Hos- pital Costs." Staff report of the Council on Wage and Price Stability, Executive Office of the President. Washington, D.C.: U.S. Government Printing Office (January). Fuchs, V. l973. "Health Care and the U.S. Economic System." In Technology and Health Care Systems in the l980's. M. Collen, ed. National Center for Health Services Research. DHEW Publ. No. (HSM) 73-30l6. Fuchs, V. l974. Who Shall Live? Health, Economics, and Social Choice. New York: Basic Books. Gaus, C. l976. "What Goes Into Technology Must Come Out in Costs." National Leadership Conference on America's Health Policy (April 29-30), pp. l2-l3. Gaus, C., and Cooper, B. l976. "Technology and Medicare: Al- ternatives for Change." Presented at Conference on Health Care Technology and Quality of Care, Boston University (November l9-20) . Gellman, D. l97l. "The Price of Progress: Technology and the Cost of Medical Care." Canadian Medical Association Journal l04(5)(March):40l-6. Ginsburg, P. l972. "Capital Investment by Non-Profit Firms: The Voluntary Hospital." Michigan State University, Econo- metrics Workshop Paper No. 7205 (November). Ginsburg, P. l976. "Inflation and Hospital Cost Investments." In Health: A Victim or Cause of Inflation?, pp. l64-78. M. Zubkoff, ed. New York: Prodist. Ginsburg, P. l977. "The Diffusion of Hospital Technology." Processed.

299 Gordon, G., and Fisher, G., eds. l975. The Diffusion of Medi- cal Technology: Policy and Research Planning Perspectives. Cambridge, Mass.: Ballinger. Grabowski, H. l976. Drug Regulation and Innovation—Empirical Evidence and Policy Options. Washington, D.C.: American Enterprise Institute. Hiatt, H. l975. "Protecting the Medical Commons: Who is Responsible?" New England Journal of Medicine 293(5)(July 3l). Iglehart, J. l977. "And Now It's Carter's Turn to Try to Con- trol Health Costs." National Journal 9(l5)(April 9):55l-56. Illich, I. l975. Medical Nemesis. London: Calder and Boyars. Kaluzny, A. l974. "Innovation in Health Services: Theoret- ical Framework and Review of Research." Health Services Research 9(2)(Summer):l0l-20. Klarman, H. l968. "Increase in the Cost of Physician and Hos- pital Services." Inquiry 7(l)(March):22-36. Klarman, H. l974. "Application of Cost-Benefit Analysis to the Health Services and the Special Case of Technology In- novation." International Journal of Health Services 4(2) (Spring):325-52. Laffey, W., and Lappen, S. l976. "Tax-Exempt Hospital Financ- ing: Revenue Bonds." Health Care Management Review l(4) (Fall):l9-30. Lave, J., and Lave, L. l970. "Hospital Cost Functions." Amer- ican Economic Review 50(3)(June):379-95. Lave, L. l976. "The Effect of Inflation on Providers." In Health: A Victim or Cause of Inflation?, pp. ll7-29. M. Zubkoff, ed. New York: Prodist. Lee, Maw Lin. l97l. "A Conspicuous Production Theory of Hos- pital Behavior." Southern Economic Journal 38(l)(July):48-58. Mahler, H. l975. "Health—A Demystification of Medical Tech- nology." Lancet 2(7940)(November l):829-33. Mather, H., et al. l97l. "Acute Myocardial Infarction: Home and Hospital Treatment." British Medical Journal 3(5770) (August 7) : 334-38. McMahon, J., and Drake, D. l976. "Inflation and the Hospital." In Health: A Victim or Cause of Inflation?, pp. l30-48. M. Zubkoff, ed. New York: Prodist. Mueller, M., and Gibson, R. l976. "National Health Expenditures, Fiscal Year l975." Social Security Bulletin 39(2)(February): 3-20. Muller, C., and Worthington, P. l970. "Factors Entering into Capital Decisions of Hospitals." In Empirical Studies in Health Economics, pp. 399-4l5. H. Klarman, ed. Baltimore: The Johns Hopkins Press. Neuhauser, D., and Jonsson, E. l974. "Managerial Response to New Health Care Technology: Coronary Artery Bypass Surgery."

300 Chapter l4 in The Management of Health Care. By W. Abernathy, A. Sheldon, and C. Prahalad. Cambridge, Mass.: Ballinger. Newhouse, J. l970. "Toward a Theory of Nonprofit Institutions: An Economic Model of a Hospital." American Economic Review 60(l)(March):64-74. Ofer, A. l976. "Third Party Reimbursement and the Evaluation of Leasing Alternatives'." Chapter 5 in "Financial Management Under Third Party Reimbursement." K. Bays, W. Breen, and A. Ofer, eds. Topics in Health Care Financing 3(l)(Fall). Office of Technology Assessment, U.S. Congress. l976a. The Computed Tomography (CT or CAT) Scanner and Its Implications for Health Policy. Draft report (September). Office of Technology Assessment, U.S. Congress. l976b. Develop- ment of Medical Technology—Opportunities for Assessment. Washington, D.C.: U.S. Government Printing Office. Office of Technology Assessment, U.S. Congress. l977. Efficacy and Safety of Medical Technology. Draft report. Pauly, M. l974. "Hospital Capital Investment: The Roles of Demand, Profits, and Physicians." Journal of Human Resources 9(l)(Winter):7-20. Pauly, M., and Drake, D. l970. "Effect of Third-Party Methods of Reimbursement on Hospital Performance." In Empirical Studies in Health Economics, pp. 297-3l4. H. Klarman, ed. Baltimore: The Johns Hopkins Press. Pauly, M., and Redisch, M. l973. "The Not-For-Profit Hospital as a Physicians' Cooperative." American Economic Review 63(l)(March):87-99. Peltzman, S. l974. Regulation of Pharmaceutical Innovation— The l962 Amendments. Washington, D.C.: American Enterprise Institute. Peterson, R. , and MacPhee, C. l973. Economic Organization in Medical Equipment and Supply. Lexington, Mass.: Lexington Books. Rapoport, J. l976. "Diffusion of Technological Innovation in Hospitals: A Case Study of Nuclear Medicine." Processed. Redisch, M. l974. "Hospital Inflationary Mechanisms." Paper presented at the Western Economics Association Meetings, Las Vegas, Nevada (June l0-l2); revised October l974. Rice, D., and Wilson, D. l975. "The American Medical Economy— Problems and Perspectives." Paper prepared for the Inter- national Conference on Health Care Costs and Expenditures, Fogarty International Center, Bethesda, Md. (June 2-4). Roche, J., and Stengle, J. l973. "Facilities for Open Heart Surgery in the United States: Distribution, Utilization, and Cost." American Journal of Cardiology 32 (2) (August): 224-28.

30l Rosenthal, G. l976. "Inflation: Directions for Research." In Health: A Victim or Cause of Inflation?, pp. 3l7-29. M. Zubkoff, ed. New York: Prodist. Russell, L. l976a. "The Diffusion of New Hospital Technolo- gies in the United States." International Journal of Health Services 6(4):557-80. Russell, L. l976b. "Inflation and the Federal Role in Health." In Health: A Victim or Cause of Inflation?, pp. 225-44. M. Zubkoff, ed. New York: Prodist. Russell, L. l976c. "Making Rational Decisions About Medical Technology." Presented at November 23 meeting of AMA's National Commission on the Cost of Medical Care, Chicago. Russell, L. l977. "How Much Does Medical Technology Cost?" Paper presented at the Annual Health Conference of the New York Academy of Medicine, New York City, April 29. Russell, L. Forthcoming. "The Diffusion of Hospital Technol- ogies: Some Economectric Evidence." Journal of Human Resources. Russell, L., and Burke, C. l975. Technological Diffusion in the Hospital Sector. Prepared for the National Science Foundation, Office of National R&D Assessment, under grant number RDA75-l4274 (October). Salkever, D. l972. "A Microeconometric Study of Hospital Cost Inflation." Journal of Political Economy 80(6) (Nov- ember/December) :ll44-66. Salkever, D., and Bice, T. l976. "The Impact of Certificate- of-Need Controls on Hospital Investment." Health and Society 54(2)(Spring):l85-2l4. Salkever, D., and Bice, T. Forthcoming. "Certificate-of-Need Legislation and Hospital Costs." Draft of chapter to be included in Hospital Cost Containment: Selected Notes for Future Policy. M. Zubkoff and I. Raskin, eds. New York: Prodist. Scitovsky, A., and McCall, N. l975. "Changes in the Costs of Treatment of Selected Illnesses l95l-l964-l97l." Health Policy Program Discussion Paper, School of Medicine, Uni- versity of California, San Francisco (September). Silvers, J. l974. "The Impact of Financial Policy and Struc- ture on Investments in Health Care." Chapter 2l in The Management of Health Care. By W. Abernathy, A. Sheldon, and C. Prahalad. Cambridge, Mass.: Ballinger. Strum, H. l967. "Technological Developments and Their Effects upon Health Manpower." Monthly Labor Review 90(l)(January): l-8. Thomas, L. l974. "The Technology of Medicine." In Lives of a Cell, pp. 3l-36. New York: Viking Press. Also in New England Journal of Medicine 285(24)(December 9, l97l):l366-68.

302 U.S. Congress. l976. Congressional Record—House. "Medical Device Amendments of l976" (March 9), pp. Hl7l9-Hl759. U.S. DHEW. l97l. Towards a Comprehensive Health Policy for the l970's: A White Paper. Washington, D.C.: U.S. Govern- ment Printing Office. U.S. Public Health Service. l968. Benefit/Cost Analysis of Kidney Disease Program: Report to Surgeon General. Washington, D.C. Wagner, J., and Zubkoff, M. Forthcoming. "Medical Technology and Hospital Costs: Notes for Policy." Draft of chapter to be included in Hospital Cost Containment: Selected Notes for Future Policy. M. Zubkoff and I. Raskin, eds. New York: Prodist. Waldman, S. l972. "The Effect of Changing Technology on Hospital Costs." Research and Statistics Note, Social Security Administration, Office of Research and Statistics. DHEW Publ. No. (SSA)72-ll70l (February 28). Warner, K. l975. "A 'Desperation-Reaction' Model of Medical Diffusion." Health Services Research l0(4)(Winter):369-83. Warner, K. l977. "Treatment Decision Making in Catastrophic Illness." Medical Care l5 (l) (January):l9-33. Worthington, N. l975a. "Expenditures for Hospital Care and Physicians' Services: Factors Affecting Annual Changes." Social Security Bulletin 38(3). Worthington, N. l975b. "Factors Contributing to Annual Changes in Expenditures for Hospital Care and Physicians' Services." Office of Research and Statistics, Social Security Administration. Zubkoff, M., ed. l976. Health: A Victim or Cause of Infla- tion? New York: Prodist.