Citation
Background Paper 2: Case Studies of Medical Technologies: Case Study 2

Material Information

Title:
Background Paper 2: Case Studies of Medical Technologies: Case Study 2 The Feasibility of Economic Evaluation of Diagnostic Procedures: The Case of Computed Tomographic Scanning
Series Title:
The Implications of Cost-Effectiveness Analysis of Medical Technology
Creator:
United States. Congress. Office of Technology Assessment.
Wagner, Judith L.
Publisher:
U.S. Congress. Office of Technology Assessment
Publication Date:
Language:
English
Physical Description:
viii, 24 p. ; 28 cm.

Subjects

Subjects / Keywords:
Tomography ( LCSH )
computed tomographic scanning ( KWD )
Diagnostic imaging ( LCSH )
Cost effectiveness ( LCSH )
resource allocation ( KWD )
Genre:
federal government publication ( marcgt )

Notes

General Note:
A case study by the Office of Technology Assessment (OTA) that explores "the feasibility of economic evaluation of diagnostic procedures" (p. 4).
General Note:
Title on report: The Implications of Cost-Effectiveness Analysis of Medical Technology: Background Paper 2: Case Studies of Medical Technologies: Case Study 2: The Feasibility of Economic Evaluation of Diagnostic Procedures: The Case of CT Scanning

Record Information

Source Institution:
University of North Texas
Holding Location:
University of North Texas
Rights Management:
This item is a work of the U.S. federal government and not subject to copyright pursuant to 17 U.S.C. §105.

Aggregation Information

IUF:
University of Florida
OTA:
Office of Technology Assessment

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

The Feasibility of Economic Evaluation of Diagnostic Procedures: The Case of CT Scanning April 1981 NTIS order #PB81-209777

PAGE 2

CASE STUDY # 2 THE IMPLICATIONS OF COST-EFFECTIVENESS ANALYSIS OF MEDICAL TECHNOlogy APRIL 1981 BACKGROUND PAPER #2: CASE STUDIES OF MEDICAL TECHNOLOGIES CASE STUDY #2: THE FEASIBILITY OF ECONOMIC EVALUATIO N OF DIAGNOSTIC PROCEDURES: THE CASE OF CT SCANNIN G Judith L, Wagner, Ph. D. Senior Research Asociate, The Urban Institute Wshington, D.C. I 1 OTA Background Papers are documents that contain information believed to be useful to various parties. The informtion under-girds formal OTA assessments or is a n outcome of interna exploratory planning and evaluation. The material is usually not of immediate policy intereest such as is contained in an OTA Report or Technical Memorandum, nor does it present options for Congress to consider. 1 J

PAGE 3

Library of Congress Catalog Card Number 80-600161 For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D.C. 20402

PAGE 4

Foreword This case study is one of 17 studies comprising Background Paper #2 for OTAs assessment, The Implications of Cost-Effectiveness Analysis of Medical Technology. That assessment analyzes the feasibility, implications, and value of using cost-effectiveness and cost-benefit analysis (CEA/CBA) in health care decisionmaking. The major, policy-oriented report of the assessment was published in August 1980. In addition to Background Paper #2, there are four other background papers being published in conjunction with the assessment: 1) a document which addresses methodological issues and reviews the CEA/CBA literature, published in September 1980; 2) a case study of the efficacy and cost-effectiveness of psychotherapy, published in October 1980; 3) a case study of four common diagnostic X-ray procedures, to be published in summer 1981; and 4) a review of international experience in managing medical technology, published in October 1980. Another related report was published in September of 1979: A Review of Selected Federal Vaccine and Immunization Policies. The case studies in Background Paper #2; Case Studies of Medical Technologies are being published individually. They were commissioned by OTA both to provide information on the specific technologies and to gain lessons t-hat the broader policy aspects of the use of CEA/CBA. Several of the cally requested by the Senate Committee on Finance, Drafts of each case study were reviewed by OTA staff; by could be applied to studies were specifimembers of the advisory panel to the overall assessment, chaired by Dr. John Hogness; by members of the Health Program Advisory Committee, chaired by Dr. Frederick Robbins; and by numerous other experts in clinical medicine, health policy, Government, and economics. We are grateful for their assistance. However, responsibility for the case studies remains with the authors. Director

PAGE 5

Advisory Panel on The Implications of Cost= Effectiveness Analysis of Medical Technology John R. Hogness, Panel Chairman President, Association of Academic Health Centers Stuart H. Altman Sheldon Leonard Dean Manager Florence Heller School Regulatory Affairs Brandeis University General Electric Co. James L. Bennington Chairman Department of Anatomic Pathology and Clinical Laboratories Childrens Hospital of San Francisco John D. Chase Associate Dean for Clinical Affairs University of Washington School of Medicine Joseph Fletcher Visiting Scholar Medical Ethics School of Medicine University of Virginia Clark C. Havighurst Professor of Law School of Law Duke University Barbara J. McNeil Department of Radiology Peter Bent Brigham Hospital Robert H. Moser Executive Vice President Americans College of Physicians Frederick Mosteller Chairman Department of Biostatistics Harvard University Robert M. Sigmond Advisor on Hospital Affairs Blue Cross and Blue Shield Associations Jane Sisk Willems VA Scholar Veterans Administration

PAGE 6

OTA Staff for Background Paper #2 Joyce C. Lashof, Assistant Director, OTA Health and Life Sciences Division H. David Banta, Health Program Manager Clyde J. Behney, Project Director Kerry Britten Kemp, Editor Virginia Cwalina, Research Assistant Shirley Ann Gayheart, Secretary Nancy L. Kenney, Secretary Martha Finney, Assistant Editor Other Contributing Staff Bryan R. Luce Lawrence Miike Michael A. Riddiough Leonard Saxe Chester Strobel OTA Publishing Staff John C. Holmes, Publishing Officer John Bergling* Kathie S. Boss Debra M. Datcher Patricia A. Dyson* Mary Harvey* Joe Henson

PAGE 7

Preface This case study is one of 17 topics being issued that comprise Background Paper #2 to the OTA project on the Implications of Cost-Effectiveness Analysis of Medical Technolology. The overall project was requested by the Senate Committee on Labor and Human Resources. In all, 19 case studies of technological applications were commissioned as part of that project. Three of the 19 were specifically requested by the Senate Committee on Finance: psychotherapy, which was issued separately as Background Paper #3; diagnostic X-ray, which will be issued as Background Paper #5; and respiratory therapies, which will be included as part of this series. The other 16 case studies were selected by OTA staff. In order to select those 16 case studies, OTA, in consultation with the advisory panel to the overall project, developed a set of selection criteria. Those criteria were designed to ensure that as a group the case studies would provide: l l l l l l l examples of types of technologies by function (preventive, diagnostic, therapeutic, and rehabilitative); examples of types of technologies by physical nature (drugs, devices, and procedures); examples of technologies in different stages of development and diffusion (new, emerging, and established); examples from different areas of medicine (such as general medical practice, pediatrics, radiology, and surgery); examples addressing medical problems that are important because of their high frequency or significant impacts (such as cost ); examples of technologies with associated high costs either because of high volume (for low-cost technologies) or high individual costs; examples that could provide informative material relating to the broader policy and methodological issues of cost-effectiveness or cost-benefit analysis (CEA/CBA); and examples with sufficient evaluable literature. On the basis of these criteria and recommendations by panel members and other experts, OTA staff selected the other case studies. These 16 plus the respiratory therapy case study requested by the Finance Committee make up the 17 studies in this background paper. All case studies were commissioned by OTA and performed under contract by experts in academia. They are authored studies. OTA subjected each case study to an extensive review process. Initial drafts of cases were reviewed by OTA staff and by members of the advisory panel to the project. Comments were provided to authors, along with OTAs suggestions for revisions. Subsequent drafts were sent by OTA to numerous experts for review and comment. Each case was seen by at least 20, and some by 40 or more, outside reviewers. These reviewers were from relevant Government agencies, professional societies, consumer and public interest groups, medicaI practice, and academic medicine. Academicians such as economists and decision analysts also reviewed the cases. In all, over 400 separate individuals or organizations reviewed one or more case studies. Although all these reviewers cannot be acknowledged individually, OTA is very grateful for their comments and advice. In addition, the authors of the case studies themselves often sent drafts to reviewers and incorporated their comments. These case studies are authored works commissioned by OTA. The authors are responsible for the conclusions of their specific case study. These cases are not statements of official OTA position. OTA does not make recommendations or endorse particular technologies. During the various stages of the review and revision process, therefore, OTA encouraged the authors to present balanced information and to recognize divergent points of view. In two cases, OTA decided that in order to more fully present divergent views on particular technologies a commentary should be added to the case study. Thus, following the case

PAGE 8

studies on gastrointestinal endoscopy and on the Keyes technique for periodontal disease, commentaries from experts in the appropriate health care specialty have been included, followed by responses from the authors. The case studies were selected and designed to fulfill two functions. The first, and primary, purpose was to provide OTA with specific information that could be used in formulating general conclusions regarding the feasibility and implications of applying CEA/CBA in health care. By examining the 19 cases as a group and looking for common problems or strengths in the techniques of CEA/CBA, OTA was able to better analyze the potential contribution that these techniques might make to the management of medical technologies and health care costs and quality. The second function of the cases was to provide useful information on the specific technologies covered. However, this was not the major intent of the cases, and they should not be regarded as complete and definitive studies of the individual technologies, In many instances the case studies do represent excellent reviews of the literature pertaining to the specific technologies and as such can stand on their own as a useful contribution to the field. In general, though, the design and the funding levels of these case studies was such that they should be read primarily in the context of the overall OTA project on CEA/CBA in health care. Some of the case studies are formal CEAS or CBAS; most are not. Some are primarily concerned with analysis of costs; others are more concerned with analysis of efficacy or effectiveness. Some, such as the study on end-stage renal disease, examine the role that formal analysis of costs and benefits can play in policy formulation. Others, such as the one on breast cancer surgery, illustrate how influences other than costs can determine the patterns of use of a technology. In other words, each looks at evaluation of the costs and the benefits of medical technologies from a slightly different perspective. The reader is encouraged to read this study in the context of the overall assessments objectives in order to gain a feeling for the potential role that CEA/CBA can or cannot play in health care and to better understand the difficulties and complexities involved in applying CEA/CBA to specific medical technologies. The 17 case studies comprising Background Paper #2 short titles and their authors are: Artificial Heart: Deborah P. Lubeck and John P. Bunker Automated Multichannel Chemistry Analyzers: Milton C. Weinstein and Laurie A. Pearlman Bone Marrow Transplants: Stuart O. Schweitzer and C. C. Scalzi Breast Cancer Surgery: Karen Schachter and Duncan Neuhauser Cardiac Radionuclide Imaging: William B. Stason and Eric Fortess Cervical Cancer Screening: Bryan R. Luce Cimetidine and Peptic Ulcer Disease: Harvey V. Fineberg and Laurie A. Pearlman Colon Cancer Screening: David M. Eddy CT Scanning: Judith L. Wagner Elective Hysterectomy: Carol Korenbrot, Ann B, Flood, Michael Higgins, Noralou Roos, and John P. Bunker End-Stage Renal Disease: Richard A. Rettig Gastrointestinal Endoscopy: Jonathan A. Showstack and Steven A. Schroeder Neonatal Intensive Care: Peter Budetti, Peggy McManus, Nancy Barrand, and Lu Ann Heinen Nurse Practitioners: Lauren LeRoy and Sharon Solkowitz Orthopedic Joint Prosthetic Implants: Judith D. Bentkover and Philip G. Drew Periodontal Disease Interventions: Richard M. Scheffler and Sheldon Rovin Selected Respiratory Therapies: Richard M. Scheffler and Morgan Delaney These studies will be available for sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, D. C. 20402. Call OTAs Publishing Office (224-8996) for availability and ordering information.

PAGE 9

Case Study #2 The Feasibility of Economic Evaluation of Diagnostic Procedures: The Case of CT Scanning Judith L. Wagner, Ph. D. Senior Research Associate The Urban Institute Washington, D.C.

PAGE 10

Contents Conclusions . . . . . . . . . . . . . 21 LIST OF FIGURES Figure Ale. Page l. A Model for Economic Evaluation of Diagnostic Procedures . . . 6 2. Disease Detection Rates for Two Tests . . . . . . . . 19 3. Case-Finding Costs for 100 Patients . . . . . . . . 19

PAGE 11

Case Study #2: The Feasibility of Economic Evaluation of Diagnostic Procedures: The Case of CT Scanning Judith L. Wagner, Ph. D. Senior Research Associate The Urban Institute Washington, D.C. INTRODUCTION In the past, the evaluation of medical procedureswhen it has occurred at allhas focused on the assessment of the procedures effectiveness in altering the status of patients health or improving the accuracy of diagnosis. Economic criteria, such as the immediate or induced costs of a procedure, have seldom been included in the evaluation. Recently, however, as the Nations ability to pay for new advances in medical procedures has been increasingly taxed, interest in economic evaluation has been rising, and the application of economic criteria has gained acceptance. One manifestation of the increasing legitimacy of economic evaluation of medical technologies is the passage of the Health Services Research, Health Statistics, and Health Care Technology Act of 1978 (Public Law 95-623), establishing a National Center for Health Care Technology which is mandated to consider, among other things, the cost effectiveness of medical technologies. In spite of this new interest, the actual application of economic evaluation to medical procedures remains limited. Except for preventive services such as immunization and asymptomatic screening (11,31,32,33,34,40), little effort has been devoted to performing this type of evaluation, Although some recent literature represents a start in this direction (42), the territory of economic evaluation of medical procedures remains largely uncharted. Nowhere do questions of the feasibility and usefulness of economic evaluation arise more than in the area of evaluating diagnostic procedures. Diagnostic testing has been widely characterized as subject to uncontrolled and explosive growth (16), and the existence of tradeoffs between the information obtained and the cost of such testing is often noted (22). But just how such tradeoffs should be measured is an issue that has not been pursued at length. Computed tomographic (CT) scanning was introduced in 1973. Initially limited to the head, CT scanning can now be used to detect diseases in other parts of the body. The use of this diagnostic technology has initiated a controversy of unprecedented proportions. This controversy has been cast in terms of the tradeoff between the benefits and costs of CT scanning. Consequently, a great deal of literature extending beyond the boundaries of traditional clinical evaluation has emerged. Numerous articles and reports have been published that describe the historical pattern of diffusion of CT scanners in hospitals and private practices throughout the United States (5,10,35), the operating costs of CT scanning units (13,14), patterns of utilization of CT scanners in particular institutions (2,25,29), and most important here, the cost effectiveness of CT scanning (15,17,23,24,41,46). Many of these studies have been reported in clinical journals, demonstrating the intense interest of the medical community not only in the .3

PAGE 12

clinical uses of CT scanning, but also in the impact of this new technology on the practice of medicine. Interestingly, almost all of the studies of the economics of CT scanning have been authored by physicians or scientists without major collaboration with economists. Judging by the quantity of published research, one would surmise that the state of the art of economic evaluation of diagnostic proceduresCT scanning, in particularis well advanced. In fact, the opposite is true. The studies of the cost effectiveness or economic benefit of CT scanning reveal major conceptual and methodological weaknesses that not only mitigate these particular studies value, but also cast doubt on the overall feasibility of conducting economic evaluations of diagnostic procedures. The purpose of this case study is to explore the feasibility of economic evaluation of diagnostic procedures. Several questions are addressed. To what extent can cost-benefit analysis (CBA) and its derivative, cost-effectiveness analysis (CEA)the cornerstones of economic evaluation methodologybe applied to diagnostic procedures? What, if any, methodological and conceptual obstacles to analyses are there? How useful such analyses? CBA and CEA are methods eating scarce resources among performing such are the results of to assist in alloalternative uses, These methods were developed primarily to evaluate large public-sector investments such as highways, dams, and airports. When applied to diagnostic procedures, they are intended to provide information on two related questions: 1) Under what circumstances should the procedure be performed? and 2) How much investment in capacity to perform the procedure is justified? The answer to the second question rests on thorough study of the first, for only by knowing when a procedure should be performed can one assess how much investment in capacity is justified. The fundamental argument of this case study is that the conceptual, methodological, and data problems inherent in answering the first question (i.e., when should a given diagnostic procedure be used?) often create such inaccurate and unreliable results that CEA/CBA is generally not worthwhile. In only a few situations (to be enumerated later) is such economic evaluation likely to be particularly illuminating and therefore useful to medical decisionmakers. Such a pessimistic prognosis for the application of CBA and CEA needs to be explained and supported with evidence. In this case study, therefore, we begin with an idealized model of economic evaluation of diagnostic procedures and then use this model to explore the implications of problems that occur in attempts to apply it. Some typical second best approaches, designed to circumvent these problems, are then described, with particular emphasis on the use of these approaches in the evaluation of CT scanning. The economic evaluations of CT scanning include three general analytic approaches: 1) good faith but limited attempts to apply the principles of economic evaluation, 2) estimates of the impact of CT scanning technology on health care expenditures, and 3) analyses of the amount and location of CT scanning capacity required to meet a predetermined level of demand. Each of these approaches is assessed with respect to its limitations as valid economic evaluation and the particular questions usefulness methodological weaknesses that raise as to the ultimate feasibility and of economic evaluation. CBA AND CEA: THE THEORETICAL IDEAL The purpose of CBA and CEA is to help decisionmakers determine the best allocation of resources among possible alternative uses. CBA is used to measure the difference between the value of all benefits resulting from an activity, both at present and in the future, and the costs. If this difference, or net social benefit, is positive, the activity would be said to be worth its costs, and

PAGE 13

the decision should be made to allocate scarce resources (labor, capital, etc. ) to producing the benefits. Straightforward as this decisionmaking tool is, the benefits accruing from an activity, particularly those undertaken in the public sector, are often difficult even to identify, much less to value. Because CBA requires enumeration and valuation of all significant benefits, analysts have derived the seemingly simpler technique of CEA. Here, a single measure of effectiveness is designated, and the ratio of cost to effectiveness guides resource allocation decisions among competing alternatives. Applying these principles to medical procedures, CBA would enumerate and place a value on all benefits (both positive and negative) derived from performing a procedure on patients with a specified set of conditions and would compare those benefits to the cost of performing the procedure. The resulting net social benefit would indicate whether the procedure should be performed under the specified conditions. In CEA, a measure of procedure effectiveness would be designated, and the ratio of that single measure to cost would be the critical item for resource allocation. Lives saved, life-years saved, quality-adjusted life-years saved, disability-days saved, and age-adjusted disability days saved are measures of effectiveness often chosen in studies of health care programs (43). Although CEA eliminates the problem of placing a monetary value on benefits, its scope is limited. While the scope of CBA is the whole society, CEAS is limited to the comparison of strategies with similar impacts. Since diagnostic procedures may have widely varying effects some procedures may affect mortality while others may affect only the quality of lifeCEA can only be used to compare a narrow set of procedures with similar purposes. The application of CBA and CEA to diagnostic procedures can be described with reference to figure 1. A set of patients with specific presenting signs and symptoms faces two or more alternative diagnostic pathways, each representing a specific combination of diagnostic tests. These alternative pathways may represent different diagnostic tests or even different testing sequences. For example, pathway A might represent the administration of a radionuclide (RN) brain scan followed by a CT head scan if the RN scan is positive or equivocal, while pathway B might represent a CT scan only. Each of these two diagnostic pathways carries with it certain resource costs and has specific diagnostic results. Four possible diagnostic results are shown for each pathway: true positive, true negative, false positive, and false negative. Each of these outcomes implies a course of therapy (with negative findings implying no therapy), each involving its own resource costs. Finally, the therapies result in health outcomes measured by relevant indicators. In CEA, a single indicator or an index of indicators is chosen, and the impact of the pathway on this indicator is compared to the total resource costs (both therapeutic and diagnostic) associated with the pathway. The diagnostic pathway with the lowest ratio of cost to effectiveness is preferred. In CBA, all relevant outcomes, both good and bad, would be measured and their value assessed. Notice that the purpose of economic evaluation is to decide under what conditions, if any, a diagnostic procedure should be used. Typically, a procedure will be of undisputed and obvious value for some groups of patients. The important question for resource allocation is not whether the diagnostic procedure or the equipment producing it is justified or whether the procedure raises or lowers health care costs, but how the procedures should be used in the practice of medicine. Only by knowing the costs and effects of treating each group of patients with the procedure is it possible to know whether there is too much or too little capacity to perform the procedure in any community. As seemingly obvious as this point is, it has not been recognized by the authors of many studies purporting to be analyses of the cost effectiveness of CT scanning. Too many of these studies, whose titles include the term cost effectiveness, are simply historical descriptions of the aggregate impact of CT scanning on health

PAGE 14

Figure 1 .A Model for Economic Evaluation of Diagnostic Procedures I I Dx(A) + I q Dx(A) Patients with specific signs, symptoms, or other demographic characteristics Dx(B) I I I False A A care costs in a hospital or community, These proved by increasing, decreasing, or rearrangstudies are plagued with measurement probing CT scanning capacity in the United States. 1 lems, but even without those problems, the I Similarly, estimating the impact of a new procedure on medistudies would offer no guidance on whether the care or public program expenditures, while useful in Federal budgexisting allocation of resources could be imeting, is not CEA. PROBLEMS IN APPLYING THE IDEAL MODEL The ideal model of economic evaluation is however, arise in or are exacerbated by the atstraightforward enough, but in attempts to aptempt to apply the model to diagnostic proceply it to diagnostic procedures, serious probdures. Problems in the following areas merit a lems arise. Some of the problems are common closer look: to all economic evaluations. For example, the l question of how one should compare costs and effects accruing at different times has not yet been fully resolved. z Several of the problems, l identifying homogeneous patient groups for analysis, specifying appropriate diagnostic pathways, measuring diagnostic accuracy,

PAGE 15

Q measuring diagnostic and therapeutic costs, and l specifying outcomes of the diagnostic/ therapeutic process. Each of these areas is discussed below. Identifying Homogeneous Patient Groups Application of the ideal model of economic evaluation requires dividing the universe of patients into reasonably homogeneous groups based on their presenting conditions and personal characteristics. The patients in each group, with a common set of signs and symptoms and perhaps other important characteristics, such as age or sex, are treated uniformly with respect to the diagnostic pathways applied. The outcome of a CEA can be extremely sensitive to the criteria (signs and symptoms) used to identify a patient group. These criteria can be so general that they lump together patients with very different problems, or they can be so detailed that they essentially create a unique group for each individual patient. If, for exampIe, an analysis of CT head scanning is based on a universe of patients with all types of neurological complaints, the CEA would average together the costs and outcomes associated with very different kinds of patients. Patients with recent strokes would be analyzed together with patients complaining of persistent headaches. At this extreme, the usefulness of economic analysis in discriminating between the appropriate and inappropriate uses of CT scanning is questionable. Conversely, if the criteria are so detailed that few patients fall into each group, the analysis would need to be repeated for each of the many groups of patients and would be prohibitively expensive. Moreover, such detailed criteria are impractical for use in diagnostic decisionmaking because they are likely to include subjective signs that are difficult to incorporate into a diagnostic protocol. The problem of patient grouping is frequently noted in connection with clinicaI trials (38). As the number of groups of patients increases, the sample size required to show statistically significant effects also increases, possibly exhausting the supply of patients available for entry into a clinicaI trial. At the same time, aggregating may obscure significant benefits available to particular subgroups. But the problem is even more acute in economic evaluation, which must estimate costs associated with diagnosing and treating patients. Cost data are rarely available on a disaggregated basis, and allocation of aggregate costs to individual patient groups involves a loss of accuracy in cost measurement that can be quite serious. Specifying Appropriate Diagnostic Pathways The outcome of economic evaluation varies with the number of diagnostic pathways studied. Omission of significant alternatives can seriously compromise the results. If a new diagnostic test is developed for a previously undiagnosable disease, then the pathways in contention are easy to specify: the new test versus no test. But when a new diagnostic procedure adds information to that available from an existing arsenal of procedures, then each diagnostic pathway comprises a procedural sequence in which the performance of later procedures is conditioned on the achievement of certain results from earlier procedures. Moreover, two sequences that differ onIy in their rules for continuing the testing process must be considered separate diagnostic pathways, since each sequence will have yields and costs that are different from those of the other sequence. Hence, the number of pathways to be considered may be quite large, and the best pathways may inadvertently be excluded from analysis. The importance of comparing alternative testing sequences places economic evaluation squarely at odds with principles for evaluating the accuracy of diagnostic procedures. Some of the best clinical studies of diagnostic procedures have measured the diagnostic accuracy of a given procedure independent of the outcome of the tests that precede or follow it. Indeed, a cardinal principle in measuring diagnostic accuracy of interpretive tests is that the reader or analyst should not be aware of the outcomes of previous tests (1). Yet, if knowledge of such outcomes would improve the accuracy or timeliness of di-

PAGE 16

agnosis in a real clinical setting, and hence the potential outcomes of therapy, then that effect should be considered in the economic evaluation. Conversely, if knowledge of previous results tends to bias the findings in a particular direction, and hence leads to more misdiagnosis and inappropriate and costly treatment, that should also be considered in the economic evaluation. Unfortunately, the best efficacy studies specifically eschew the measurement of diagnostic accuracy in such biased but realistic settings. It is not surprising, then, that estimates of the relative frequency of outcomes of different diagnostic pathways are often lacking. Any aspect of diagnosis or treatment that changes the parameters of the diagnostic pathwayeither diagnostic accuracy, therapeutic effectiveness, or resource costsrequires definition of a new diagostic pathway. Moreover, a particular diagnostic pathway assumes a particular mix of inpatient and outpatient treatment. If two treatment protocols are identical except for the setting of care, then the protocols must be considered as separate pathways, since their costs may differ widely. Technological change, which may improve the accuracy of a diagnostic test or the effectiveness with which the diagnosed disease can be treated, also requires the specification of new pathways. One accepted way of dealing with uncertainty about or changes in such parameters without constructing additional pathways is use of sensitivity analysis. Sensitivity analysis systematically assesses the effect of parameter changes on the solution to the economic analysis. The question is posed as follows: What percentage change in the diagnostic accuracy of test X would be needed to induce a change in the analytic findings? The approach is powerful when only one or two parameters are subject to change or uncertainty, but it breaks down under conditions of general uncertainty about the appropriate value of many parameters. Unfortunately, most economic evaluations of diagnostic procedures face such conditions, It should now be clear that the need to specify and compare a large number of alternative pathways to one another taxes both available data and analytical resources. Specifying Diagnostic Accuracy The model pictured in figure 1 is constructed as if each alternative diagnostic pathway yields onl y two diagnostic findings: positive and negative. Often, a positive finding is associated with several diseases or different stages of a disease. Hence, estimates of the total sensitivity (i.e., false-positive rates) and specificity (i.e., true negative rates) 3 of a diagnostic pathway are generally inadequate for economic evaluation, Each type of positive finding must be linked to the particular therapeutic pathway it evokes, so that the cost of that pathway can be estimated. Thus, for any patient group with specific presenting signs and symptoms, accurate economic evaluation requires data not only on the total sensitivity of a diagnostic pathway, but also on the rates of detection of the full spectrum of disease states that the diagnostic pathwa y migh t identify. Such detailed estimates are seldom available. Measuring Diagnostic and Therapeutic Costs Economic evaluation requires measurement of the diagnostic and therapeutic costs incurred when patients embark on a particular pathway. A CBA should measure the cost of all present and future health care stemmin g from the health problem that precipitated entry into the pathway in the first place. The cost of future health care made necessar y by a false-negative finding, for example, should be included. In CEA, the criterion for inclusion of health care costs depends on the chosen indicator of effectiveness. The use of disability days as the effectiveness indicator would obviate the need for measurement of health care costs associated with treament of disability itself. Similarly, a cost-effectiveness study using the 5-year survival rate as the outcome measure of interest should count all costs up to the day of outcome

PAGE 17

measurement. The cost of a false negative would be relevant only if it induced the delivery of additional medical care prior to the point at which effectiveness is measured. Thus, with CEA, there is a natural stopping point to cost measurement that is dictated by the time horizon implicit in the effectiveness measure. Most of the cost-effectiveness studies of CT scanning have treated the measurement of diagnostic and therapeutic costs cavalierly, ignoring the difficult conceptual and measurement problems and taking ad hoc approaches without acknowledging their limitations. The authors of these studies either do not appreciate the difficulties inherent in cost measurement, or they recognize and are overwhelmed by the problems, and therefore ignore them. Nevertheless, the problems are great and may defy solution. Cost is a measure of valuein this case, the value of productive resources (labor, equipment, supplies, etc. ) that must be forgone to other uses when a particular diagnostic/ therapeutic pathway is chosen for the patient group under study. The appropriate economic concept is that of incremental or avoidable costs, those incurred as a direct result of committing a given increment of patients to a diagnostic therapeutic pathway or that can be avoided if the same patients do not take the pathway. Avoidable costs are treated as variable, while unavoidable costs are assumed to be fixed The definition of what is avoidable or variable depends on the time perspective of the decision and the size of the increment. Generally, the longer the time perspective and the larger the increment, the more costs are avoidable. Suppose, for example, that the decision is whether to perform CT scanning on patients with suspected stroke in a hospital with a scanner currently used at .50-percent capacity for other types of patients. In the very short run, the only truly avoidable costs are the materials, supplies, and extra wear and tear on equipment caused by serving the stroke patients. Most labor costs may be invariant with serving the new patients, because technicians must be paid for full workshifts whether they are busy or not. Consequently, the appropriate measure of cost would exclude these fixed operating costs. Also, because the equipment exists to serve other

PAGE 18

assume that the service will be provided using the most efficient technology currently available to produce the expected diagnostic and therapeutic results. Therefore, the appropriate measure of capacity cost is not the original cost of the equipment and plant, but the current cost of replacing it. Depending on the rates of technological change and general inflation, replacement cost may be quite different from the original capital cost of the equipment in any institution. But whereas the original cost of existing equipment is readily available from accounting records, estimation of current costs of replacing capacity is fraught with difficulty. Product improvements in CT scanning, for example, have increased equipment costs and improved the flexibility of CT scanning and the quality of resulting images. If these quality improvements are unnecessary to produce the level of diagnostic accuracy specified for the pathway, the measured costs of CT scanning should not be based on these new models. Whole-body scanners, for example, are increasingly used in studies of the head. Since the purchase price and operating costs of head scanners are less than those of body scanners, it would be incorrect to allocate to head scanning patients a portion of the extra costs associated with a body scanner. But how will one estimate the cost of replacing head scanners when they are no longer manufactured? The measurement of original capital cost, while inappropriate, is at least precise. Whichever concept of capacity cost is used original capital cost or current replacement costit must be allocated between the patient group under study and all other patient groups. A common rule of thumb is to allocate these costs in proportion to the share of total volume accounted for by the patient group under study. Underlying this rule is the implicit assumption that the costs of any excess capacity should be borne equally by each patient group using the service. In fact, this assumption may bias the analysis against certain patient groups. Headache patients, for example, may be able to wait weeks for a CT scan without negative effects on outcome, but may not have to do so because excess CT scanning capacity has been made available to serve trauma patients on an emergency basis. CT scanning costs allocated to the headache patients should rightly be calculated as if the equipment were used to full capacity, while the trauma patient group should bear the entire cost of the excess capacity. Such differential cost allocation has not been attempted in the CT scanning papers reviewed by this author. Several economic studies of CT scanning have used hospital procedure charges to estimate the cost associated with the use of CT scanning. This approach is unacceptable, because procedure charges bear no relation to the incremental costs of treating particular patient groups. Even if each charge were set to equate the revenues from a procedure with its associated costs, which is seldom the case, the unit charge so constructed would be unlikely to reflect optimal utilization of capacity or appropriate allocation of indirect costs and overhead. It should be noted, however, that in most American hospitals charges for specific procedures are often completely unrelated to their costs. Profits from one procedure typically subsidize losses from another. Indeed, the hallmark of a good hospital administrator is that charges have been manipulated to maximize revenues. Consequently, the use of procedure charges as a proxy for incremental cost seriously compromises an analysis. The extensive reliance on procedure charges in economic studies of CT scanning is curious in light of the relatively detailed information available on the operating costs of CT scanning units (13,14). Although the obstacles are formidable, improved estimates of the cost of some procedures are within reach. Another common practice in economic studies is to use physicians fees as estimates of the value of physicians inputs into the diagnostic and therapeutic processes. Although such fees represent actual payments, they are not accurate reflections of the value of physician inputs. Third-party payer policies have created anomalies in fee structures, so that some diagnostic procedures offer higher returns on the physicians investment of time and skill than others. In particular, fees for new procedures are often set higher than those for established procedures. A better approach to valuing physicians serv-

PAGE 19

ices would be to use carefully constructed relative value scales based on the time and skill requirements of individual procedures. Construction of good relative value scales is expensive, however, and is unlikely to be undertaken as part of the evaluation of medical procedures. Some recent tentative evidence suggests that the California Relative Value Scale may be reasonably well correlated with the level of difficulty of surgical procedures i(19), but more investigation of this question is needed before it would be possible to recommend reliance on such a system. To summarize, the obstacles to accurate measurement of the resource costs associated with a diagnostic/therapeutic pathway are great, but substantial improvement in cost estimates would be possible if researchers would attempt to approximate long-run incremental cost. At present, most cost analyses are grossly inaccurate, so much so, in fact, that wrong inferences have been drawn about the usefulness of CT scanning. Specifying Outcomes of the Diagnostic/Therapeutic Process way. CEA assumes that a single measure or index can capture the essential nature of the outcomes. The former type of analysis challenges our present capacity to objectively quantify all types of effects, both good and bad, while the latter runs the risk of oversimplifying the decisionmaking problem. The problem of specifying and measuring outcomes is common to all evaluation and is therefore beyond the scope of this case study. One type of outcome, however, is unique to economic evaluation of diagnostic procedures: the benefits derived from negative findings. Negative findings are certainly valued by patients, for such findings can have dramatic impacts on their peace of mind. The reassurance value of negative findings is generally ignored in CEA, because satisfactory methods for estimating changes both in patient satisfaction and in the value that patients place on those changes are lacking at present (42). Consequently, CEA is likely to be biased against diagnostic/therapeutic pathways offering high specificity (i. e., high true-negative rates) relative to those pathways with high sensitivity (i. e., high true-positive rates). CBA requires the specification of all significant kinds of outcomes of each diagnostic pathECONOMIC EVALUATIONS OF CT SCANNING Only two studies of CT scanning with which the author is familiar are good-faith efforts to implement the ideal model of economic evaluation in a comprehensive way. One such study, by Knaus and Wagner (24), is successful; a second study, by Baker and Way (4), represents a failure to understand and carry out the principles of economic evaluation. Knaus and Wagner (24) analyzed the effectiveness and costs of two alternative diagnostic pathways for patients presenting with sudden severe headaches but no focal signs; such patients are generally viewed with skepticism as candidates for CT scanning. The analysts in this study compared the outcomes of two diagnostic pathways: 1) performing a CT scan on all such patients referred; and 2) performing a CT scan only on patients presenting additional neurological signs (i. e., depriving the patient of a C T scan unless and until his or her condition deteriorates). Only one possible positive finding was considered: a ruptured intracranial aneurysm resulting in subarachnoid hemorrhage. Because early diagnosis of this disease significantly improves the prognosis of surgical intervention, the authors concluded that the cost of each life saved by early detection is about $511,000, or $23,000 per life-year saved. Knaus and Wagner observed that the analysis depended on the attainment of a diagnostic yield at least as high as that obtained at a large teaching hospital: about one aneurysm per 125

PAGE 20

toms, and other factors that might improve the cost-effectiveness ratio for headache patients. It is instructive to ask why Knaus study met so many criteria for good economic evaluation. A critical determinant appears to be the availability of empirical estimates of the outcome of surgery, medical therapy, and no intervention in patients in various symptomatic stages of cerebral aneurysms. Without such probability estimates, the study would have foundered. The availability of data from clinical trials made possible the decision analytic approach for an admittedly narrow set of diagnostic findings for patients with headaches. The study by Baker and Way (4) of CT scanning in the abdomen is a manifest demonstration of what can go wrong when clinicians attempt to apply the principles of economic evaluation. The authors randomly selected the case histories of approximately 200 hospitalized patients who had received body scans in 1976 or 1977. Each patient was classified according to the organ system studied (e. g., biliary, liver, etc. ). Baker and Way then constructed an arbitrary scale of effectiveness, where a value of 1 was assigned to a patient for whom the CT scan had a lifesaving effect, a value of 18 was assigned to a patient if the CT scan actually precipitated a death, and a value of 10 indicated no net value. The costs of CT scans and other related diagnostic tests were presumabl y estimated from the procedure charges, although the authors did not specify their methods. A ratio of cost to effectiveness was constructed for each organ system, with the effectiveness scale value as the denominator. The authors concluded that the differences between the categories [organ systems] are insignificant, and the average result suggests that, in general, the value of the scans to medical care in relation to the cost of scans was low, Apart from the obvious question of the reliability of subjective assessments of effectiveness and the cost measurement techniques, two criticisms of Baker and Ways study are relevant here. First, it is utterly meaningless to compare the cost-effectiveness ratio of CT scanning across patient groups. The appropriate comparison is between CT scanning and other diagnos-

PAGE 21

of CT scanning on the cost of diagnosis is a logical starting point but is generally an insufficient approach to economic evaluation. The fundamental weakness of the studies remaining to be discussed, therefore, is not their failure to consider therapies and outcomes. Rather, it is their choice of diagnostic pathways. Each one of these studies compares actual patterns of diagnosis with and without CT scanning. Most are pre-CT/post-CT comparisons of diagnostic costs in an institution or country. Given the system of medical practice in force today, however, there is no uniformity from physician to physician in the use of diagnostic procedures on similar patients. Since one physician may order a battery of tests simultaneously, while another may move cautiously from one to another, each diagnostic pathway comprises many approaches to diagnosis. Thus, to show that the introduction of CT scanning has historically raised or reduced diagnostic costs is to show that and nothing more. That finding holds no implications for resource allocation, because the pathways are too amorphous to determine how CT scanning should fit into the diagnostic process. Homogeneous patient groups, however, are likely to undergo diagnostic strategies that are reasonably uniform. Therefore, empirical studies of the impact of CT scanning on diagnostic cost are more likely to be useful as resource allocation tools when these studies are concentrated on a narrow set of presenting signs and symptoms. Three studies of specific patient groups with relatively unambiguous and homogeneous signs and symptoms were reported by Larson and colleagues at the University of Washington. They analyzed the impact of CT scanning on the utilization and cost of diagnostic services for patients with presenting conditions suggestive of brain tumors (26), cerebrovascular disease (27), and hydrocephalus (30). Admission forms were abstracted for all inpatients admitted to the university-affiliated hospital with signs and symptoms suggesting one of the three diseases during the year prior to and 2 years after the introduction of a CT head scanner. Patterns of service

PAGE 22

delivery and medical care in the pre-CT and post-CT patient groups were compared, including the utilization of and charges for neurodiagnostic procedures; the length of hospital stay; the time required to reach final diagnosis; the level of detail of the diagnosis; and the type of therapy employed, CT scanning had a major impact on the configuration of diagnostic studies used for each of the three types of patients, but the relative benefits resulting from these changes varied among the three. In the study of patients with suspected brain tumors (26), CT scanning significantly reduced the use of RN brain scanning (from 85 to 38 percent of patients), angiography (from 65 to 50 percent of patients), and pneumoencephalography (PEG) (from 15 to 4 percent of patients). The length of hospitalization, speed of diagnostic workup, and timing and type of therapy for these patients, though, did not change significantly. The overall detectability of brain tumors increased markedly after CT scanning became available. Total per-patient hospital charges for neurodiagnostic procedures did not change significantly. The authors therefore concluded that CT scanning as it was integrated into the neurodiagnostic process at the study hospital was cost effective, because it permitted reductions in the use of invasive and somewhat risky procedures but did not raise hospital charges or otherwise affect prognosis, In the case of patients with suspected cerebrovascular disease (27), the introduction of CT scanning significantly reduced the number of lumbar punctures and RN brain scans, but the total per-capita hospital charge for neurodiagnostic procedures increased about 33 percent. The length of hospitalization and speed of diagnostic workup did not change, and although the level of detail of the discharge diagnosis was greatly increased, such detail was unnecessary to plan therapy. In the judgment of the authors, CT scan results were not essential to establishing a diagnosis in these patients but played a confirming role. In patients with suspected cerebrovascular disease, therefore, CT scanning was found to be of questionable value. The introduction of CT scanning in the diagnostic workup of children with suspected hydrocephalus (30) led to major reductions in RN brain scans and PEGs. CT scanning also reduced the total charges for neurodiagnostic procedures. Speed of diagnosis, length of hospitalization, and kind of therapy did not change. Thus, CT scanning was found to be particularly appropriate for such patients. All three Larson studies depended on the use of procedure charges as a proxy for the economic costs involved in performing the neurodiagnostic procedures. The resulting distortion of estimated cost impacts is instructive. In the study of suspected hydrocephalus (30), total estimated diagnostic charges dropped precipitously with the advent of CT scanning, largely because of the high charge for a PEG examination ($492 at the study hospital). The high charge at least partly reflects that procedures relative rarity and the existence of excess capacity. The long-run incremental cost of a PEG examination is probably much lower. Were such examinations not painful and risky, an d therefore to be avoided wherever possible, a substantial change in the estimated cost of this procedure could have reversed the findings regarding the benefit of CT scanning in patients with suspected hydrocephalus. In fact, a PEG procedure charge of $2 OO would have resulted in no difference between pre-CT and post-CT total per-capita charge for neurodiagnostic procedures, and the analysis would have rested on the implications of substituting CT for PEG examinations for patient morbidity and risk. The investigation did not question the appropriateness of inpatient versus outpatient workups. Some neurodiagnostic studies can probably be performed on an outpatient basis. By accepting the existing settings of care as the basis for comparison, the studies may have overlooked particularly efficient alternatives. The three Larson studies do show, however, that with good hospital data systems and effort and persistence on the part of the investigators, it is possible to identify patients with reasonably homogeneous presenting signs and symptoms. The ability to obtain medical records categorized by admission problem rather than discharge diagnosis is critical for estimating the outcomes of different diagnostic pathways, and

PAGE 23

although the Larson studies do not specify pathways with sufficient clarity, their patient identification methodology would support such analyses. A more recent study by the same authors investigated the impact of outpatient CT scans on headache patients (28). Three cohorts of pa tients were selected for study: one cohort containing patients seen before installation of a CT scanner, a second cohort containing patients seen shortly after installation, and a third cohort containing patients seen 1 year after installation of the CT scanner, The patients in each of these cohorts were selected from a review of electroencephalogram (EEG) records, where headache was the reason for the EEG. Thus, the patient sample excluded those who may have had CT scans without EEGs. The cost of diagnostic tests was computed using procedure charges; the diagnostic testing charges per patient increased 16 percent after the introduction of CT scanning. The neuroradiologic diagnostic yield of patients with normal neurological workups was low, and, in the case of CT, was negligible. This finding led the authors to suggest that abnormal findings on neurological examination should be a screening criterion for performing CT scans and other neuroradiological procedures for patients with headaches. The fact that diagnostic charges increased after the introduction of CT scanning says very little about the appropriate allocation of resources; the more powerful evidence presented by the authors is the apparent absence of real clinical benefit, even with respect to diagnostic capability, deriving from CT scanningand, indeed, from most neurodiagnostic procedures. If radiological and other technology-bound procedures are to be performed on patients with no focal findings, then this study has not adequately demonstrated that CT scanning cannot lower the cost of diagnosis if it is appropriately substituted for other tests. A clear cost-reducing option, however, appears to ensue from stopping the diagnostic workup of the patient when the clinical examination reveals no abnormalities. In light of the importance of such a conclusion, it is unfortunate that the sample of patients in this study was chosen by excluding those who may have had CT scans without EEGs. More needs to be done to refine the study of diagnostic procedures for patients with headaches. Ambrose and his colleagues (2) in Britain studied the impact of CT scanning on the diagnosis of head injuries, but they did not attempt to attach cost estimates to the observed changes in utilization of diagnostic procedures and exploratory surgery. The introduction of CT scanning at the study hospital dramatically reduced the use of arteriography, PEG, and exploratory craniotomy without changing the mortality rate; however, once CT examinations were available, they were ordered so frequently that net costs may have increased. Nevertheless, the realized effect is not the relevant question for resource allocation. CT scanning can undoubtedly reduce the cost of diagnosis for some patients by substantially reducing the high cost, morbidity, and mortality associated with the diagnostic procedures it replaces. The real question though, is how to achieve the best results. Like previous studies, the Ambrose study failed to explore how CT scanning might optimally be integrated into diagnostic and management protocols. In a recent study of CT scanning in acute head trauma at a U.S. teaching hospital, Zimmerman, et al. (47) noted a progressive decrease in utilization of skull X-rays following the introduction of CT scanning, but they also observed that a certain proportion of the present skull radiographic examinations are done because of the referring physicians adherence to traditional ways of evaluating trauma patients. A critical question that economic analysis might address is exactly when, if at all, skull X-rays should be employed in the evaluation of patients with head trauma. Bahr and Hodges (3) investigated the impact of CT scanning on the total cost of hospital care (measured by billed charges) for neurological and neurosurgical inpatients at a university hospital. The total average hospital bill for these patients decreased about 10 percent, after adjusting for inflation, but the decrease was not statistically significant. The study population was disaggregated into three diagnostic groups:

PAGE 24

1) patients with extracerebral collections, 2) patients with proved intracranial tumors, and 3) patients with vascular disease. The total inflation-adjusted hospital bill for patients in the first two categories decreased by 23 and 35 percent respectively, while that of the third group increased by 30 percent. It is difficult to interpret such findings because of the possibility that the patient mix changed over the period, especially if the availability of CT scanning enabled testing of some as outpatients. In addition, the study population is ambiguous, consisting of patients with all types of neurological complaints. The disaggregated cost comparisons are based on final diagnosis, not on the patients presenting signs and symptoms. Consequently, the cost or savings by diagnosis are irrelevant to the use of CT scanning under specified presenting conditions. The difficulty of interpreting cost impacts based on procedure charges has already been noted. In one of the earliest economic impact studies, Wortzman, Holgate, and Morgan (45) reviewed the records of over 200 inpatients who were given CT head scans at a Canadian hospital. In each case, the authors judged whether an angiogram or PEG had been prevented or prompted by the CT scan. The hospital costs (measured by charges) saved or induced by use of the CT scan were calculated on the assumption that the hospital stay for an angiogram is 1 da y and that for a PEG examination, 3 days. Net savings resulting from the availability of CT scanning were estimated at $300 per patient. In a second part of the study, the authors reviewed all outpatient CT scans at the hospital and, with the help of a survey of referring physicians, estimated that 58 percent of the outpatients would have been hospitalized had the CT scan been unavailable. Assuming that these patients would have stayed the average length of stay in neurology (14 days), the authors estimated that CT scanning saved $1,490 per outpatient scan. The vulnerability of the savings estimate to changes in assumptions is obvious and needs no further comment. Nor does the use of charges to reflect cost. The method for determining the degree of substitution of CT for other procedures is judgmental and therefore subject to investigator and respondent bias. Most important, however, the aggregated nature of the patient group under study mitigates the value of the findings. It is certain that savings, however measured, could be further increased by more detailed analysis of presenting signs and symptoms. In a recent reappraisal of their early study, Wortzman and Holgate (45) revealed discrepancies between their estimates of potential savings and realized effects in subsequent years. Hospital stays and admissions did not decline as predicted in the study. Although the number of angiograms and PEGs did decline, the cost of these procedures, as measured by charges, rose, thus wiping out much of the estimated savings. The procedure charge increases were necessitated by decreases in utilization without concomitant capacity reduction. But the authors wrongly assumed that estimated cost savings should be altered to reflect these changes. The principle of long-run incremental cost requires the assumption that capacity will ultimately adjust to reduced demand. A study by Thomson (41) at Frenchay Hospital in Great Britain traces changes in the use of particular procedures and services in the 2 years following the installation of a CT scanner. There, as in most other institutions, the number of arteriograms, RN brain scans, and PEGs declined after the introduction of CT scanning; the length of stay of neurosurgical inpatients declined as well, and admissions were avoided. The cost of operating a CT scanner at the study hospital was meticulously calculated using a modified concept of capacity replacement cost and assuming operation at full capacity. The economic cost of other neurodiagnostic procedures made unnecessary by CT scanning were not measured so carefully, but were assigned relatively low values in order to bias the study against CT scanning. Hospital stay savings were estimated at the average daily cost of a National Health Service hospital bed. CT generated net savings for the hospital, and, by inference, for the National Health Service, although the savings were minimal, Since Thompson made no attempt to follow a specific patient group but merely reported on in-

PAGE 25

stitutionwide changes in utilization of services, it is impossible to know the kinds of patients for whom CT scanning is cost effective. Other authors have documented such institutionwide changes (26). This kind of information, even when linked to good cost estimates, ignores the central problem for medical decisionmaking and health facilities planning: To what uses should CT scanning capability be applied? Two studies have attempted to estimate the impact of CT scanning on the national cost of diagnosis in the United States. Willems, et al. (44) estimated net changes in total U.S. expenditures for diagnosis due to the introduction of CT scanning. The performance of a CT scan generated a technical charge, professional fee, and associated hospital room and board charges, all of which totaled an amount between $295 million and $428 million in 1976. Savings resulting from reduction in PEGs, RN brain scans, and arteriograms were deducted from this total, leaving a net estimate of $181 million to $39 0 million in increased expenditures due to CT scanning in 1976. Like analysts in most of the other studies previously described, WilIems used procedure charges to estimate the cost of most procedures. Although charges are an inadequate surrogate for resource costs, they do reflect real transfers from payers to providers of health care; in that sense, the total expenditure burden imposed by CT scanning is of interest. It is unknown, however, whether the expenditures were used to subsidize the provision of other services that hospitals provide at a loss or, perhaps, whether the expenditures were used to inflate physician or other professional incomes. A study by researchers at Arthur D. Little, Inc. (17) represents the only effort to date to estimate the national impact of both head and body scanning on the cost of diagnosis. The national impact on the utilization of competing diagnostic procedures and exploratory surgery in 1977 and 1980 was estimated by a variety of sources, including consultations with experts in the field. The technical cost of providing each type of procedure was estimated assuming equipment at full utilization and current costs of replacement of equipment. (Except for exploratory surgery, charges were not used to estimate procedure costs. ) To arrive at a unit cost for each of the various diagnostic procedures, the national average hospital charge was applied to the hospitalization required by cedure. It was predicted in the A. room and boar d additional days of each type of proD. Little study that the availability of CT head scanning would raise diagnostic costs by about $29 million in 1977 and $31 million in 1980. These low estimates of net impact contrast with the results of the Willems study but can be explained by different assumptions about the effect of CT scanning on exploratory surgery and skull X-rays. The predicted impact of CT body scanning is subject to greater speculation since, at the tim e of the study, body scanning had only begun to find its way into the diagnostic process. The authors predicted that in 1980, as a result of the use of CT scanning, the estimated cost of diagnosing abdominal and mediastinal disorders will have increased by $152 million. This estimate is based on the assumption that the availability of CT body scanning will reduce the number of exploratory surgeries by 50 percent and that no other diagnostic procedures will be affected by CT scanning. Although the ultimate significance of these estimates for decisionmaking is most certainly low, the A. D. Little study represents the best effort to date at estimating the resources costs associated with performing diagnostic procedures. Capacity costs were based on current equipment replacement prices and full capacity. However, indirect costs (administrative expenses in departments outside of the CT scanning unit) were arbitrarily assigned a value of 50 percent of direct costs, a level that probably bears no relation to the actual incremental costs associated with the existence of the CT scanning capacity. The technical cost of surgery an d routine costs of hospital care were based on charges rather than costs. With the high cost of hospital construction today, hospital room and board charges reflecting original capital costs probably understate replacement costs, In spite of these shortcomings, the method for estimat-

PAGE 26

ing procedure costs is a model for economic evaluation of CT scanning. Studies of the Cost of Case Finding A commonly advocated approach to economic evaluation is the calculation of the cost of case finding, which is variously defined as the cost per positive finding, the cost per correct finding, or the cost per true positive (6). The cost of case finding is a concept borrowed from the economic evaluation of screening procedures. It is particularly useful in determining which of two or more mutually exclusive screening procedures is more cost effective in detecting a particular disease in an asymptomatic population. A procedure (or sequence of procedures) that costs the least per true positive finding is preferred to all others. Unfortunately, the concept does not travel well. When applied to analysis of CT scanning, which is used on symptomatic patients in conjunction with other tests, the cost of case finding is meaningless. The results of studies using the concept illustrate its limitations. Carrera, et al. (9) calculated the cost per abnormal CT scan for three types of cases: 1) headaches without neurological findings, 2) headaches with neurological findings, and 3) suspected temporal lobe epilepsy. Because of differences in the diagnostic yield of CT scanning among these three types of cases, the cost per abnormal CT scan examination was $4,000; $1,300; and $2,000 respectively. 7 In a separate study, Knaus and Davis (23) calculated the casefinding cost of CT scanning for 25 leading indications for CT head scanning. The case-finding cost at the study hospital ranged from $411 for patients in coma to $3,500 for patients with headache. Larson, et al. (28) estimated that the cost of case finding for headache patients without abnormal neurological findings was $11,901. The authors of all three papers questioned the cost effectiveness of CT scanning in patients with headaches without neurological 7 A CT procedure charge of $240 was used as a surrogate for cost findings. Any conclusion about cost effectiveness is premature, however, for the cost of case finding can only be compared to the benefits of finding abnormalities in the patient group. Evens and Jest (12) have used the concept of case-finding cost to compare CT head scanning and RN brain scanning on patients with neurological complaints. Using estimates of sensivitity and specificity of CT and RN scanning, the authors calculated the cost per correct finding: CT scanning cost $141 per correct finding; RN scanning cost $73 per correct finding. Although, on the surface, this finding would imply that RN scanning is the procedure of choice, Evens and Jest observed that CT is a more sensitive test. Hence, in a large number of cases, the combination of the two tests would be ordered, with a joint cost per correct finding of $214. If an initial negative RN scan generates a confirmatory CT examination in at least 60 percent of neurological cases, then an initial CT scan costs less per correct finding than an initial RN examination. Notice that Evens and Jest used a variant of case-finding cost. Whereas Carrera (9) and Knaus and Davis (23) calculated the cost per abnormal or positive finding, Evens and Jest calculated the cost per correct finding. A third approach, the cost per true positive finding, has also been used. g The relative merit of these alternative measures of case-finding cost depends on the errors in diagnosis and implications for therapy of each alternative. In fact, the most appropriate measure can be selected properly only if the costs and benefits of pursuing diagnostic pathways to their ultimate outcomes are known. This point can be illustrated algebraically. Let TP, FP, TN, and FN denote the proportion of true positives, false positives, true negatives, and false negatives expected from a given diagnostic pathway in a specified population of patients. Let C be the total cost of performing See, e.g., B. J. McNeil and S J. Adelstein, Measures of Clinical Efficacy. The Value of Case Finding in Hypertensive Renovascular Disease, 1975 (32).

PAGE 27

diagnostic procedures on these patients. The three case-finding cost options are: Cost per true positive = c TP (100) Cost per abnormal finding = c (Tp + Fp) (1OO) Cost per correct finding = c (TP + TN) (100) Suppose that two diagnostic procedures, A and B, have the disease detection rates shown in figure 2 when performed on a patient group with a 0.50 disease prevalence rate. Assume also Figure 2. Disease Detection Rates for Two Tests + Tes t result s Diagnosti c test A B that both procedures cost the same to administer, $10 per patient. Figure 3 gives the casefinding costs of each procedure for each 100 patients as measured by the three case-finding options. In figure 3, diagnostic procedure B would be preferred if the option cost per correct finding were used, and diagnostic procedure A would be preferred under the options cost per true positive and cost per abnormal finding. Suppose, however, that the implications of a false negative are major (e.g., a fatal but completely treatable disease goes undetected), while those of a false positive are minor (e. g., inexpensive additional tests are required). Clearly, diagnostic procedure A would be preferred, and use of the cost per correct finding would be unacceptable. Similarly, if the implications of a false positive are major (e.g., application of inappropriate, expensive, and risky therapy), while those of a false negative are also great, then the use of cost per correct finding is superior. Thus, to accurately determine which measure of case-finding cost is preferred, it is necessary to know something about the subsequent benefits and costs resulting from a particular diagnostic procedure. The more that is known, the more accurate can be the weights put on the different kinds of errors. Thus, casefinding costs do not bypass the need for information concerning the full implications of embarking on a diagnostic pathway. Figure 3.Case-Finding Costs for 100 Patients Cost per Cost per Cost per true positiv e correct findin g abnormal finding (C/TP) (C/TP + TN ) (C/TP + F? )

PAGE 28

least 500 angiograms and 200 PEGs could be eliminated each year, the CT scanner would save money. The cost analysis rightly assumed that all radiographic equipment is operated at capacity both before and after the introduction of CT; this avoids ascribing to CT scanning savings that would otherwise be obtainable from consolidating radiographic services in larger hospitals, In fact, the whole question of CT scanner location appears to be inextricably linked to the optimal distribution of neuroradiologic capacity as a whole, Costs could be reduced by regionalizing such capacity, but these savings would have to be compared to the added cost of transportation to regional facilities. For emergency patients, the cost savings would also have to be weighed against reduced access to care. Thus, the Swedish study, though excellent in its costing methodology, is still limited as a planning approach, since it analyzes the effect of introducing CT scanning to an already existing system in which the present allocation of radiological resources may very well be inappropriate. Bartlett, et al. (7) reported on a comparison of alternative options for the location of CT scanning capacity in a particular region of England. The study assumed that the demand for CT scans would be 3.5 per thousand residents per year. Alternative configurations for meeting (and exceeding) that demand were then compared. The critical question was whether CT scanning capacity should be installed in all hospitals with neurosurgery, neurology, and accident centers. The not so startling conclusion was that more scanners would cost more money and the added cost of installing a CT scanner in every accident center (far exceeding the required scanner-to-population ratio) must be compared to the lifesaving potential of the scanner in head trauma. We are thus led full circle back to the basic question of the conditions for which CT scanning is economically justified. An American study by Greenwald, et al. (18) is an application of a mathematical optimization technique to the question of scanner capacity and location. The authors formulated the problem thus: How does one choose the number and location of CT scanners such that the sum of CT

PAGE 29

operating and patient travel costs is minimized? The solution must meet two constraints: 1) that the regions demand for CT scans is met, and 2) that each scanner is assigned a patient load not exceeding its capacity. Other considerations, such as the desirability of placing CT scanning capability in centers of radiological excellence, were not included in the model, but easily could be if the potential locations of CT scanners were limited to those sites meeting certain criteria. The authors showed that when patient travel costs are taken into account, more scanners can actually reduce the total cost of meeting a given level of demand. As demand for CT services increases or decreases, however, the optimal number of facilities also changes. Hence, the model begs the question of the optimal number of scanners by assuming that the answer to the prior questionunder what conditions is CT scanning worth its costs has a 1 read y been answered. CONCLUSIONS Taken together, the literature on the economics of CT scanning reveals a plethora of studies that either ask the wrong questions, or, in asking the right ones, must make heroic and unsupportable assumptions. Few studies have addressed the fundamental question of economic evalution: For what kinds of patients is CT scanning worth its costs? The energy devoted to documenting historical cost savings resulting from CT scanning is particularly misplaced. This literature offers no real insight into the question facing health care decisionmakers today. Studies of the cost of case finding, which appear on the surface to be more useful in guiding resource allocation decisions, offer such a partial view of costs and effects of any diagnostic pathway that their results are more obfuscating than enlightening. Finally, those methods designed to assist directly in the big resource allocation decisionthe placement of CT scanning capacitymust ignore the unanswered question of conditions of cost-effective use. The mathematical programing approach to facility location has a rich history, especially with health care facilities. It is worth noting that if CT scanning is considered a necessary diagnostic tool for head trauma and other emergency situations, then the formulation suggested by Greenwald, et al. (18) may not be appropriate. Instead, a preferred formulation might be to minimize the number of scanners required to insure that all points in the region lie within some maximum time or distance criterion. 11 The resulting configuration of CT scanning facilities would very likely be quite different from that obtained in the Greenwald, et al, study. Although the methodological and conceptual obstacles to conducting useful economic evaluations are great, the literature shows that some can be overcome. Unfortunately, the obstacles have been dealt with in pieceseach study has offered a solution to a particular problem, but no study is completely satisfactory in all areas. In particular, the proper estimation of resource costs is an area that appears to be a good candidate for major improvements with relatively littIe effort, although some conceptual problems will continue to exist. Certainly the use of charges as a surrogate for cost can be eliminated. The identification of patients by specific signs and symptoms also appears to be feasible, although the cost of research based on specified patient groups will depend on the existence of good hospital data collection systems. The problem of identifying patients by signs and symptoms will be intensified with the use of CT body scanning because of the ambiguity of signs and symptoms related to diseases of the abdomen.

PAGE 30

When the data needed to perform valid economic evaluations are considered, it is not surprising that so few appropriately constructed studies exist. The cost of analyzing the signs and symptoms of the many different kinds of patients who are reasonable candidates for CT scanning would be great in itself; to imagine conducting such analysis for all types of patients and diagnostic procedures is unrealistic in the extreme. At best, a few controversial indications for expensive diagnostic procedures might be the subjects of economic evaluation. Research to date on CT scanning suggests that patients with headaches and suspected cerebrovascular disease would be excellent candidates for additional analysis with more appropriate methods of economic evaluation. If one cannot expect a comprehensive analysis of the appropriate conditions of use of a diagnostic procedure, how can those responsible for decisions about the level and location of diagnostic capacity make rational choices? The task is impossible, as most health planning agency personnel have suspected all along. Instead of attempting to find analytic approaches to support what is essentially a political decision, health planning agencies might do better to set fairly arbitrary regional capacity ceilings and use economic evaluation to explore the cost and access implications of alternative locational choices. Even better, the incentives might be improved for those who are most knowledgeable about the diagnostic alternatives: the providers themselves. If we had greater confidence that their incentives were consistent with the goals of public policy, the limitations of formal evaluation would be far less troublesome. 1. 2. 3. 4. 5. 6. 7. 8. 9. Abrams, H. L., and McNeil, B, J., Medical Implications of Computed Tomography (CAT Scanning ), N. Eng. J, Med. 298:255, 1978. Ambrose, J., et al., EMI Scan in the Management of Head In juries, Lancet 1:847, 1976. Bahr, A. L., and Hodges, F. J., Efficacy of Computed Tomography of the Head in Changing Patient Care and Health Costs: A Retrospective Study, Am, ]. Roent. 131:45, 1978. Baker, C., and Way, L. W., Clinical Utility of CAT Body Scans, Am. ). Surg. 136:37, July 1978. Banta, D., Number and Distribution of CT Scanners, 1978, report to the Bureau of Health Planning and to State and local health planning agencies (Hyattsville, Md.: National Center for Health Services Research, August 1978). Banta, D., and McNeil, B. J., The Costs of Medical Diagnosis: The Case of the CT Scanner, Health Care Mgt. Rev, 3:7, 1978. Bartlett, J. R., et al,, Evaluating Cost-Effectiveness of Diagnostic Equipment: The Brain Scanner Case, Br. Med. ]. 2:815, 1978. Bunker, J. P., et al, (eds. ), Costs, Risks, an d Benefits of Surgery (New York: Oxford University Press, 1977). Carrera, G. F., et al., Computerized Tomography of the Brain in Patients With Headache and 10. 11. 12. 13. 14. 15. 16. 17. With Temporal Lobe Cost Effectiveness, /, 1978. Epilepsy: Findings and Comp. Tomog, 2:108, Creditor, M. C., and Garrett, J. B., The Information Base for Diffusion of Technology: Computed Tomography Scanning, N. Eng. J, Med. 297:49, 1977. Dales, L. G., et al., Evaluation of a Periodic Multiphasic Health Checkup, Meth. Inform. Med. 13:140, 1974. Evens, R, G., and Jest, R. G., The Clinical Efficacy and Cost Analysis of Cranial Computed Tomography and the Radionuclide Brain Scan, Sem. Nuc. Med. 7:129, 1977, Economic Analysis of Body Computed Tomography UnitsIncluding Data on Utilization, Radio, 127:151, 1978. Economic Analysis of Computed Tomography Units, Am. ], Roent. 127:191, 1976. Evens, R. G,, et al., The Utilization, Reliability, and Cost Effectiveness of Computer Cranial Tomography in Evaluating Pseudotumor Cerebri, Am, J. Radio. 129:263, 1977. Freeborn, D. K,, et al., Determinants of Medical Care Utilization: Physicians Use of Laboratory Services, Am. J. Public Health 62:846, 1972. Gempel, P. A,, and Harris, G. H. Comparative

PAGE 31

18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. Cost Analysis: Computed Tomography vs. Alternative Diagnostic Procedures, 1977-1980 (Cambridge, Mass.: Arthur D. Little, Inc., 1977). Greenwald, H. P., et al., Transportation or CT Scanners: A Theory and Method of Health Resources Allocation, Health Serv. Res. 14:207, 1979. Hsiao, W. C., and Stason, W, B,, Toward Developing a Relative Value Scale for Medical and Surgical Services, Health Care Fin. Rev. 1;23, 1979. Institute of Medicine, Computed Tomographic Scanning: A Policy Statement (Washington, D. C.: National Academy of Sciences, 1977). Jonsson, E., and Marke, L. A., CAT Scanners: The Swedish Experience, Health Care Mgt. Rev, 2;37, 1977. Kassirir, J. P., and Pauker, S. G., Should Diagnostic Testing Be Regulated? N. Eng. ]. Med. 299:947, 1978. Knaus, W. A., and Davis, D. O., Utilization and Cost-Effectiveness of Cranial Computed Tomography at a University Hospital, ]. Comp. Assist. Tomog. 2:209, 1978. Knaus, W. A., and Wagner, D., CT Scanning Patients With Headache: A Cost-Benefit Calculation, Am. ]. Neuro. (in press). Knaus, W. A., et al., Impact of New Technology: The CT Scanner, Med. Care 15:533, July 1977. Larson, E. B., and Omenn, G., The Impact of Computed Tomography on the Care of Patients With Suspected Brain Tumor, Med. Care 15:543, 1977. Computed Tomography in Patients With Cerebrovascular Disease: Impact of a New Technology on Patient Care, Anl. ]. Roent. 131:35, 1978. Diagnostic Evaluation of Headache: Impact of Computerized Tomography and CostEffectiveness, ]. A.M.A. 243:359, 1980. Effect of Computed Tomography on Utilization of Cerebral Angiography, Am. ]. Roent, 129:1, 1977. Impact of Computed Tomography on the Care of Patients With Suspected Hydrocephalus, Am. ]. Roent. 131 :41, 1978. McNeil, B. J., and Adelstein, S. J., Determining the Value of Diagnostic and Screening Tests, ]. Nucl. Med. 17:439, 1976. Measures of Clinical Efficacy: The Value of Case Finding in Hypertensive Renovascular Disease, N. Eng. ]. Med. 293:221, 1975. McNeil, B. J., et al., Measures of Clinical Ef 34. 35. 36. 37. 38. 39, 40. 41. 42. 43. 44. ficacy: Cost-Effectiveness Calculations in the Diagnosis and Treatment of Hypertensive Renovascular Disease, N. Eng. ]. Med. 293:216, 1975. Neuhauser, D., and Lewicki, A. J., What Do We Gain From the Sixth Stool Guaiac? N. Eng. j. Med. 293:266, 1975. Office of Technology Assessment, U.S. Congress, Policy Implications of the Computed ToMography (CT) Scanner, GPO stock No 052-003 -00565-4 (Washington, D. C.: U.S. Government Printing Office, 1978). The Implications of Cost-Effectiveness) less Analysis of Medical Technolology, GPO stock No. 052-003 -00765-7 (Washington, D. C.: U.S. Government Printing Office, August 1980). Orkand Corp., Planning Approach, Criteria, and Standards for Specialized Trauma Services: Optimal, Intermediate, and Minimal Levels of Care (Silver Spring, Md.: Orkand Corp., Sept. 30, 1976). (Also available as publication No. HRP-0015437 from Department of Commerce, National Technical Information Service, Springfield, Va. ) Pete, R., et al., Design and Analysis of Randomized Clinical Trials Requiring Prolonged Observation of Each Patient, Br. ]. Cancer 34:585, 1976. ReVelle, C., et al., Facility Location: A Review of Context-Free and EMS Model s, Health Serv. Res. 12:129, 1977. Schoenbaum, S. C., et al., Benefit Cost Analysis of Rubella Vaccination Policy, N. Eng. ]. Med. 24:306, 1976. Thomson, J. L. G., Cost-Effectiveness of an EMI Brain Scanner: A Review of a Two-Year Experience, Health Trends 9:16, 1977. Weinstein, M. C., Economic Evaluation of Medical Procedures and Technologies, in Medical Technology, J. Wagner (ed. ), Urban Institute Conference, NCHSR Research Proceedings Series, DHEW publication No. (PHS) 79-3254 (Washington, D. C.: National Center for Health Services Research, Office of Health Research, Statistics, and Technology, PHS, DHEW, 1979). Weinstein, M. C., and Stason, W. B., Foundations of Cost-Effectiveness Analysis for Health and Medical Practices, N. Eng. ]. Med. 296:716, 1977. Willems, J., et al., The Computed Tomography (CT) Scanner, in Medical Technology: The Culprit Behind Health Care Costs? proceedings of the 1977 Sun Valley Forum on National Health, S. H. Altman and R. Blendon (eds. ),

PAGE 32

DHEW publication No. (PHS) 79-3216 (Washmography: An Evaluation of Cost-Effectiveington, D. C.: DHEW, 1979). ness, Radio. 117:75, 1975. 45. Wortzman, G., and Holgate, R. C., Reapprais47. Zimmerman, R. A., et al., Cranial Computed al of the Cost Effectiveness of Computed ToTomography in Diagnosis and Management of mography in a Government-Sponsored Health Acute Head Trauma, Am. ], Roernt. 131:27, Care System, Radio. 130:257, 1979. 1978. 46. Wortzman, G,, et al., Cranial Computed To-


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ENWFC30VK_I5S1W7 INGEST_TIME 2017-05-18T16:56:45Z PACKAGE AA00055059_00017
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES