i 'il 'I 1 ' I I I .1i'l I 1 @ illtltl' Draft: January 26, 1972 'REGIONAL MEDICAL PROORAMS SERVICE AND domPUTER ASSISTED EKG ANALYSIS SYSTEMS Division of Professional and Technical Development @R4igional Medical Programs Service TABLE OF CONTENTS I. Introduction II. Need for,Improved ZltctrbcardiogrAphid Services 111. Description of CAE Systems TV. Current Functional Status of CAE Systems V. Advantages and Potential Advantages of,CAE Systems VI. Limitations @VII. Potential RMPS Roles VIII. Deployment TX. Recommendations Appendix I 1. Introduction The staff report of August 1971 entitled "Computer Assisted EKG Analysis and the Regional Medical Programs" described the difficulties encountered by four Regional Medical Programs in the development and implementation of a computer assisted electrocardiographic system (CAE). The information presented was obtained from review of appropriate materials gontained'in grant applications, correspondence filet, site visit and progress reports, Statements pertaining to the functional capabilities of established CAE systems were derived from a limited review of pertinent'literature, and attempted to define the practical advantages and the insd4quacied of existing systemi. The limitations i@erent.to a report based solely upon these sour.c*g are apparent, Accordingly, the report was submitted to experts in the field of com- puterized electrocardiography both within and without RWS with the purpose of identifying major deficiencies and inaccuracies contained in it,,and@t,o offer appropriate suggestions. Additibnally, an ad hoc panel of experts was convened by RMPS to discuss the report and to .appritt Fd4PS As to the current state of the art. A list of patticipants At this meeting$ and the agenda is included in Appendix I. The@prosent,report@has beenprop:ared@in,direct response to the. .react:Lonsigeherated by, the initial paper and reflects the continued interest on the part of both,@the' Regional Medical Programs Servide@@ and individual,Retional Medical Programs with respect to CAE systems. Three"sotircei,of information have been utilikad: 2. 1. Background and descriptive material derived from the current medical literature included in the annotated bibliography accompanying this report. (Appendix II) 2. Written communications from the project directors of four ongoing RKP supported CAE,progtame. 3, The stenographic transcript of the proceeding of the ad hoc Advisory panel Assembled by RMPS on November 30, 1971. The objectives of this report include the following: 1. To summarize the present functional capabilities of CAE systems with emphasis directed towards practical potentials and applications,, and realistic limitations. 2, To establish suitable roles for RMPS with respect to designi development an(i iL-ipleinentitioii of CAE systems. 3. To provid6 guidelines with respect to system selection, deployment and marketing strategies for use by RMPs interested in CAE systems.,. 3. 4. To provide suitable information for use by the National Advisory Council, RMPS, in establishing policy with respect to sup- port of CAE systems. 5. To recommend appropriate RMPS policy in this area. II. Need for Improved Electrocardiographic Services Disorders of the cardiovascular system are the leading causes of morbidity And mortality in the United States. It has been estimated that over one million Americans die from acute coronary heart disease annually,.and an additional several hundred thousand succumb to cerebro- vancular@diseases. The electrocardiogram is an essential tool in the evaluation.ahd management of these and other disorders. It is ac- knowledged by experts, however, that many areas of the country lack adequate electrocardiographic services. The average practitioner though often familiar with the rudiments of EKG interpretation, relies uIpon the assistance of associates more thoroughly trained in this specialized area. Ideally, an expert cardiologist should be available for this servicei but this is frequently not the case. More comonlyi an internist or general practitioner with an interest in electrocardiog- raphy assumes this function. Clearly, the quality of electtdcardio- graphic serv ces available in.any particular area of the country raa y be quite variable. Many potential solutions to this problem exist, -Train.irLg,more cardiologists or additional training for non- card ologist physicians are two approaches It must be realized,.. 4. however, that the existing deficiencies are as much due to mal distribution of trained personnel As they are to inadequacies in absolute nuImbers. It is all too common to find an abundance of highly trained cardiologists concentrated around a sophisticated medical center while the surrounding community lacks such expertise. The problem is even more obvious when one examines the services available to peripheral comr4un ity hospitals in the more rural areas of the country. It is unrealistic to expect more equitable distri- bution of existing manpower in the near future. The extension of expert services available in modern medical centers into their surrounding regions through application of advanced technology is an attractive alternatives C uter assisted EKG systems represents OMP one example of this@approach. .Aside from the necessity of increasing the availability of EKG there exists a need to improve the quality of EKG inter- pretation. Conv entional electrocardiography suffers from severa shortcomings basic to which are the limitations of the human mind. Even the most skilled cardiologist is subject to fatigue. This affects both his observations and his analysis. Lack of consistency in the application of the multiplicity of diagnostic criteria is another characteristic that impairs the performance of electrocardi- ographers. Finally, human EKG interpretation is biased by factors other than the analytic data provided in the tracings. The electro- cardiographer@has variable access to information regarding the ati nt s age sex,, health status, medication, symptomatology etc., and this 5. influences his estimation of various abnormalities of EKG wavefom. Though such clinical correlation is essential to accurate patient evaluation, it distorts the interpretation of objective data presented in the @. All of these factors contribute to the variability of humn EKG interpretation. This is reflected in the magnitude of disagreement found among experts analyzing the same EKG, and in t e incontittencids detected when the one interprdter reads the same tracing on different occasions. Skilled electrbcardiogriphers exhibit a roprbd ucibility of EKG diagnosis of approximately 807.; the perform- ance of tion-experts is more variable. The digital computer is.well suited to EKG analysis, as will be subsequently discussed, is capable of overcoming many.of the limitations inherent to human interpretation. Any proposed system designed to imprdv6 the availability and quality'of electrocardiographic services must do so within the re straints imposed by existing economic rtsoureds. The cost of a con- Ventioital ZKG,varies co@siddrably throughout the country from a few dollars to thirt dollars or more. These figures re resent the fed p charged by 'the@ptovider institution or physician and do,not necessarily reflect the true coot of the unit cardiogram. Limited data are av ail,- abi6,a'nalyzitig component costs, i.e... technician salaries, software, maintenance, and overhead. One important objective for any EKG service Lneluding,CAE systems must be cost-containment. Additional 'hardwart.:software and personnel,costo should not add significantly 6. to existing expenditures. The service must be provided at comparable costs, or preferably, with cost savings. III. Descri]2ti6n of CAE Systems The digital.computer is capable of analyzing large volumes of specific electrocardiographic variables related to wave contour, magnitude, duration, and frequency. The recognition of deviations from programmed normal values in these parameters permits discrimination of abnormal EKGs from normal ones. Identification of patterns of deviation can then-be translated into specific diagnostic statements, The computer, therefore, may serve only to select out abnormal tracings from a larger population of normal ones. Or, the computer may offer diagnostic interpretations with appropriate alternatives and exclusions. Though,individual systems differ with respect to specific hard- ware, they all share some common features: 1. A device to detect and record the electrical events of the heart and a preamplifier system to amplify the millivolt-range physiologic signals to a level that can be worked with conveniently (data acquisition cart). 2. A system for converting this signal into a mode suitable for computer input (analog-digital conversion). 3. A means of storing and transmitting the coded signal to the digital computer. 7. 4. A digital computer and a suitable computer program Available systems differ with respect to design of acquisition carts (single channel versus three channel), input mode, computer facilities, and printout equ3.pment. The recording technique includes either standard twelve lead tracings, Frank orthogonal leads, or com- binations of these leads. Several contour and rhythm programs are available, most using concept$ of binary logic. The use of nwiti- variatt statistical analysis remains largely developmental. Differ- ences also exist in the form of the computer printout (paper and microfilm), mounting, storage and retrieval capabilities Finally, the types of service provided by any single system vary. Recogtizing the multiplicity of CAE systems currently in operation, and the variety of hardware, software, program And logis- tir-s involved the succeeding section concerned with current functional capabilities of CAE systems will confine itself to general issues and avoid detailed analysis of any single system. IV. :Current Functional.Status of PAE Systems A. Range of Services Stieening: Large volumes of EKGs can be batch processed by existing CAR sy@tems,.rgpidly, reliably,, and at reasonable cost. Patients with abnormal or borderline tracings may be selected out'by the, comIouter for further evaluation. In addition to separating normal 8 from abnormal EKGS. the computer may provide specific diagnostic statements. EKGs determined to be normal by the computer appear to require no further validation, since the incidence of false negative interpretations is less than 1'/.. Depending upon the rigidity of the diagnostic criteria, the incidence of false positive diagnosis may be as high as 20'/.;.therefore tracings with suspected abnormalities require physician validation. Definitive Interpretation: The succeeding sections dealing with the accuracy, volume capacity and cost factors of CAE systems relate:.primarily to those,systems providing definitive inter- pretation. Some general comments are in order at this point. Several EKG wave contour and rhythm programs are currently available and have met with varying success. Most offer standard twelve lead tracings along with vecto.rcardiographic display. Increased utilization of@ the vectotdardiogram (VCG) should be considered advantageous, since it provides the electrocardiographer wIith additional useful information. CAE systems have proven to be quite flexible in that changin' diag- 9 nostic criteria can be accommodated for by appropriate alterations in programming. CAE systems have also shown flexibility with respect to deployment. Computer centers have provided EKG services in a variety of environments including large medical complexes, small and medium sized hospitals, outpatient clinics, small group practices and solo office practices. 9. multiple sources of funding have provided the necessary working capital including NCHSR&D, RMPS, commercial interests and institutional research And development grants. CAE systems are presently available both from commercial enterprises and from several large medical centers* Critical Care Unit Monitoring: On-line EKG monitoring in coronary care (CCU), intensive care (ICU), and respiratory care-units may provide continuous surveillance of cardiac rhythm and contour., Othdr physiologic variables often simultaneously monitored include blood gases, blood pressure,,@respir,atdry rate and temperature. Inade- quacies of,human senses with@respect to continuous observation of multiple variables have-been well documented. The application of CAE systems;in these anviroments is largely developmental.. Exercise Teattni: Several programs have been'designed for continuous analysis of EKG wavelcontour and rhythm during graded exer- eine tettin These programs are essentially research orient Com- 9 ed. put6ritbd analysis@of exercise testing is not widel y available, and must be considered in a developmental stage. Comparative Interpretation: Though several computer pro- grams are undergoing development, relatively few CAE systems provide for even Limited comparative EKG interpretation. Improved technology is available for expansion of storage and retrieval capabilities Which may be experted',to,faci,litate comparative analysis. Digagreementi 10. exists, however, as to which parameters should be compared. Addition- ally, it has been stated by some experts that comparative EKG analysis can be performed mote effectively by human observers. All agree that comparative evaluation of multiple EKGs adds valuable information to patient assessment, and further development of this aspect of CAE systems is to be expected. B. Accuracy Computer analysis of EKGs may be subdivided into t ree a- tinct process66: signal acquisition and measurement, comparison of determined parameters against sets of normal values, and synthesis of patterns of deviation from normal values into diagnostic statements. With respect to the measurement process, the computer appears capable of precision comparable to the most meticulous human effort. Though some vitiation in wave measurement is inherent to the sampling procedure, it is min 1. The computer experiences difficulties in areas similarly troublesome to human observers, i.e., P wave recog- nition detection of-minute r waves, and determination of initiation, duration and termination of Q waves, ST segment and T wave. Unlike their human counterparts, computers are not subject to@ia.tigue, thus assuring consistency-of repetitive determinations. Recognition of deviations from normal values and synthesis of dia nostic statements rely upon both the precision of wave measure- 9 ment and the adequacy of programmed normal values and logic. Computers offer the distinct advantage of being ablelto retain large volumes of information, including the multiple parameters required for EKG analysis. The major limitations are the validity of the progrmmed normal values and diagnostic criteria, and,thdae limitations apply equAlly to both computer and human interpretation. Most CAE systems in current use utilize binary logic and decision tables that attempt to duplicate the analyti c processes of the expert electtocardiographer. The net effect of the computer interface is enhanced reproducibility of interpretation resulting from rigid adherence to progra-med in- structiorts. Errors of omission are minimized though at the expense of 10-207* false positive statements. If accuracy o f CAE systems is defined as degree of agreement with expert human interpretation, then optimal performance should be reflected in 1007. computer-physician agreement. In practice, less, than the ideal is realized. @Approximate figures from systems serving large inpatient po lation's containing a preponderance of abnormal PU tracings indicate that $Si; of@c,omputer analyzed EKGs,require no subse- queht tltoi&tiot@dr modification$ i.e., the Ohysician-computer &gee- meat iii)proac been achieved c his 95% Groatet,agreament has' 4ith ontour analysis (Approx. 90'/.),'vhile rhythm statftents ate less reliable (7-5-80%- agreement. disagre 'nt are @'Two@seneral,reasong'for cotoputer-physician eme identifiable: CAR iiystem.error and physician error. Factors affect- hands @formmee hi-4 iou'ly been alluded to and include per e prev a 12. fatigue, forgetfulness, imprecise measurement, inconsistency of diag- nostic criteria, and bias. CAE system failure ma be separated into, y defects in data acquisition, data transmission, and data analysis. The dominant sources of errors of acquisition are related to inadequate recording 'techniques, improper lea.d.positioning, inad- vertant omission of leads respiratory variations, artifact, and back- ground and transmission noise. Many of these problems are resolvable through improved technician training, and modification of the data acquisition darts. The design of more suitable' electrodes and improved methods for signal filtering are under development, and can be expected to enhance the quality of the recorded signal. Similarly, distortion, noise, and,artifact introduced during telephonic transmission appear to be potentially solvable technological problems. Analytic errors may be further categorized into inadequacies of programming, errors in measurement, and disagreement as to diag- nostic classification. IErrors of measurement relate to the sampling process and, as previously mentioned, moat.often involve failure to detedt P wavesand Q RS$ QT and ST-T segment duration. Artifact, back- ground noise, and unrecognized ectopic beats may further contribute to measurement error. Instances where wave measurement is accurate but diagnostic classification is improper ar.e usually due to remedial errors in pro- ent with respect to d lassification gramming.. Disagreem iagnostic c 13. often reflects minor discrepancies between computer and electrocardi- ographer. Physician inconsistency and ambiguity of diagnostic criteria are most often responsible. In summary, though the degree of cardiologist-computer agree- ment is quite high, A significant incidence of major disagreement still exists, thus necessitating human validation of all computer analyzed EKGS. Except in screening situations in which the computer can be relied upon to separate normal EKGO from equivocal and abnormal ones independently, CAE systems serve to assist the electrocardiographer in his task. The value of the physician-computer interaction will@be discussed in@a subsequent'section (V). C* volume CaEactty A decided advantage of CAR a stems is their,ability to y process large numbers of electrocardiograms rapidly and continuously. The volume capacity of any CAZ system is determined by three major variables:, data acquisition time, computer@time,.and physician vali- dation time.' Acquisition'time is affected by technician skill, number of leads recorded, and type of dart utilized. Using carts durrently available for recording three leads simultaneously, the acquisition time for a 'twelve lead tracing is comparable to recording a standard twelve lead EKG on a one channel cart. Addition of three Frank orthogonal leads increases recording time slightly. The quality of the adquired signal is an important.determinan.t of overall system 14 erformance and this is directly influenced by the skill and motivation p of the recording-technician. Additional training is necessary to assure high quality signal acquisition. Among the factors controlling computer time are the following:. capacity of the computer complexity of computer contour and rhythm programs, input mode (batch processing versus interrupted input), and range of services provided. Comparative interpretation requires Addi- tional time to allow for search and retrieval of stored records. A medium capacity, dedicated computer is capable of processing apprbxi- mately two hundred EKGs'per hour including analysis and data printout without comparative interpretation. From a clinical standpoint, it is advantageous to provide t e physician with an interpreted tracing with minimal delay. The key determinant of turn-around time, i.e., time from inscription of the signal to delivery of_the definitive interpretation, is the additional time required for validation of the computer record. The computer can provide an interpretation within minutes of the inscri tion p of the signal, All computer diagnoses, however, excepting those determined to be entirely normal in screening situation, must be authenticated by an electrocardiographer. At the present time this service may be provided by cardiologists associated with the computer center, local cardiologists, or physicians requesting the electro- ce. physi cardiographic servi It is generally agreed that the cian utilizer should be provided with a validated record, and that review by an electrocaidiog rapher is necessary. It is the degree Af na&4afAnng offered by the CAE system to the electrocardiographer that determines its effective output. Tracings identified as being normal by the computer require no more than casual verification. Similarly, many abnormalities are analyzed so reliably by the CAE system that the reviewer can accept these diagnostic statements with a high level of confidence. Complex contour abnormalities and arrhyth- miss require closer scrutiny. The computer interface accelerates analysis of routine tracings allowing the 6lettrocatdiographer more time to consider the mote complicated ones. The interpretation time of a cardiologist reading predominantly abnormal tracings may be halved, thus potentially doubling his output. Utilizing batch pro- ceasing, turn-around time for a computer assisted EKG interpretation with verification averagestwelve hours or less, D.: Cost FActors., From the standpoint:of the CAE systems provider it is don- venient to examine component costs in three areas: dat a acquisition and transmission, computer processing, and physician validation.. It =at be emphasized that.the.figures mentioned are only approximations, with gignifidant@variati6no existing among individual CAE systems@. .,some of the factors responsible for these differences Will be.identified. Aside from the e ter, the data acquisition cart is the OMPU single most expensive hardware component of CAE systems. 'Cost Of purchase var ies from $6,000 to $10,000, whereas rental fees average $2,000 to $3,000 yearly. Assuming the functional life of this equip- ment to be five to gevenlyears, cost tier unit EKG based upon an annual volume of two,th,oust,nd tracings is $1.00 to $2.00. This is exclusive of maintenance, overhead, supplies, and technician salaries. These factors must,be figured into the total cost of acquisition and apply to both systems utilizing computer analysis, and to conventional EKG services. Where technicians function solely to record EKGS, the unit cost is increased by an amount calculated by dividing technician salary by annual volume of tracings. In institutions generating large volumes of =O, technician salaries add approxima tely $2.00 to unit acquisition costs. Smaller hospitals often use EKG technicians for a variety of functions in addition to EKG recordin thus obscuring 9 the analysis of acquisition costs. Computer costs are influenced by computer size, processing mode$ range of service provided, and data,display form. The most important determinant df@c'omputer costs is efficiency of utilization. Maximal utilization in terms of volume of EKGs processed results in, optimal unit cost. Cost-effectiveness can be achieved on a medium sized (e.g., IBM 1800) dedicated cemputer with.an annual volume of 100,000 tracings. Lesser.volumes are more economically handled using a medium sized computer on,a time-sharing basis. Uninterrupted batch processing is the most efficient input mode. Periodic "stat" or emergency tracings must,be anticipated in any system designed to pro- vide complete electrocardiographic services to inpatient pop ulations, though these interruptions impair the operating efficiency of CAE systems. -The addition of data storage and retrieval capabilities necessary for comparative interpretation increases unit cost, as do 17 modifications in the form and content of data display. Estimates of $1.00 to $3.00 have been made for direct computer costs exclusive of salaries for programmers, clerks, secretaries, maintenance, and over- head. Representative combined cost per unit EKG including adquisi- tion, transmission, prodeasing, printout and return varies from $5.00 to $7.50. However, accurate estimates of total direct and indirect Costs are not available. Finally, there is the cost of validating the preprocessed EKG. As already mentioned the fee charged for a standard eioctro- cardiogram shows marked regional variation and the total cost trans- mitted to the patient@is not an accurate reflection of the component costs of the service. The fees charged by physicians for ihterpre- tation of tracings are highly variable. If the costs of computer preprocessing are-simply,added to the customary fed charged by the Olecttdcardi6grapher, cost containment and reduction are not achieved. The obvious question remains: How can the benefits of computer asoigted EKG analysis be realized wi thout inflating the.coit of the finished product? Most commercial organizations marketing CAIR systems essentially ignore the problem by hot including physician validation as par't,of the service they sell,, From both medical and legal viewpoints, com- puter processed electrocardiograme,lacking individual verif,ication by a Okilled.6lectro6ardibgkaphtr are unacceptable. Whether or not these'comp@ies can@provide CAE:Services with constantly available physician validation at competitive prices is uncertain at this time. Two mechanisms to provide for validation may be offered. The first is employment by the computer center of electrocardiographers for the purpose -of providing validation. A foreseeable problem is the avail- A second appro ability of round-the-clock cover ge. ach is validation a b local cardiologists in the area being served by the computer center. y To avoid increasing EKG costs by addition of computer costs to local ones, same adjustment of physician fees may become necessary. Theor- etically, since computer preprocessing is estimated to reduce the time required for physician interpretation, a reduction in unit cost may be possible. However, this will depend upon the complexity of the tracings, the need for comparison and the number of tracings analyzed CAE systems situated in university medical center com- plexes have been able to provide EKG service with validation on a twenty-four hour a day basis and at a cost of $4.50 to $6.50. There are several reasons for this. Most important are the investi- gative aspects of these programs and the interests of the participating cardiologists. It.is obvious that the bulk of the money generated from the service charge covers.the cost of acquisition and compu.ter processing; a relatively small fraction goes to the physicians over- seeing the operation. Secondly, several additional sources of fun ng are available to these programs including support from private industry, foundations, and internal research and development monies. Thirdly, at least some of the supervisory function is assumed by members of the hous' e staffs, though all tracings are ultimately reviewed by senior staff members. Fourthly, the current service charges are 19. adequate for support of the ongoing programs but fail to include initial developmental and marketing expenses. Lastly, these programs operate on a,large annual volume of tracings. In summary then, several viable CAE systems are available that are capable of delivering,high quality electrocardiographic services at reasonable cost. Though current service charges provide adequate support for established, ongoing systemi they do not cook- ponsate for start-up costs which MAY be considerable. .20 V. Advantages and Potential Advantages of CAE Systems, A. Quality of Electrocardiokraphic Services Computer preprocessing of EKGs minimizes the adverse effects of physician fatigue* forgetfulness, and bias. Rigid adherence to programmed criteria diminishes errors'of omission by alerting the physician to diagnostic possibilities otherwise overlooked, Similarly, sources of physician misdiagnosis become identifiable upon review of the analog.data presented in the computer printout. The computer inter- face allows for greater standarization of diagnostic criteria, thus reducing inter-observer and intra-observer variation, and enhancing analytic re reducibility. Additionally, technicians must be more meticulous during EKG recording to insure optimal signals for computer input; this enhances the quality of the data display and serves as a built-in quality control mechanism. Overall, the quality of the EKG resulting from the cmputer-physician interaction is superior to that furnished by either one separately. B. utilization of Manpower With respect to the EKG technician, utilization of the data acquisition carts is somewhat more complicated than conventional electtocdrdiagraphst-and additional training is required. This is not difficult to achieve and technicians readily adapt to the new carts. Recording time may be slightly increased when fourteen or fifteen lead positions are required; though in general the recording time on 21 multichannel acquisition carts is comparable to standard machines, The computer interface permits automation of cutting and,lmounting of tracings, thus freeing technicians from these time- consuming tasks, Similarly, computer processing provides automated determination of impulse duration, magnitude, direction and frequency, these tasks having previously been the responsibility of the individual interpreting the EKG. The net effect is a,significaut savings in personnel time. Two positive effects upon physician utilization are discernible. Firstly, computer assistance enhances the speed of inter proration of,the@electrocardiographer, diminishing the amount of time necessary for the review of tracings, This it translatable into an increased volume capacity.for the.physician allowing him to examine more records in less-time. The time saved can be invested in other areas demanding his attention* Secondly* CAE systems serve to expand, the,ou timddh of the alettr6c&rdiographic abilities situated within the computet,center, As a result, the general quality of electrocardiography in the community is improved, and areas lacking the essential manpower can have Access to high qualit a nie tly and at manageable y ervicea, conve n Costo@ C. @Regionalization One attractive aspect of CAE systems from the,point of view of RMPS,,, is the,-potential for implementing.planned delivery of 22 essential medical services on a regional basis, This is particularly true for those systems based in a medical.cdnter environment. The linkages established between.provider institutions and peiri heral p medical facilities antipersonnel have the potential for developing beyond service relationships into viable communications networks. It' is difficult to estimate the effectiveness and quality of cooperative relationthips-foiwtored by existing CAE systems. D. Increased Interest It is probable that the availability of CAE systems has and will continue to stimulate increased interest in electrocardiography among practicing physicians This interest is expressed as increased utilization of the electrocardiogram in patient eval uation in both hospital and ambulatory'care enviro=ents, on-a more routine basis. I?,, Coutinuinst Education It has also been suggested that the CAE data display is an offective.educational device. Certainly the information contained in the printout, i.e., analog data, including diagnostic criteria and tracing reproductions can increase the knowledge of the average physician. However, the extent to which it is utilized is de ttrmined by the motivation of the recipient. An evaluation of the educational aspects of CAE systems is not available. 23 Fe, CAE ftstems as a,Reference Point As has been mentioned, little information is availi6 with respect to cost and time,factors associated with the recording and interpretation of conventional electrocardiograms. Similarly, relatively few studies attempt to define the quality of standard EKGS. The develop- mint of CAE programs has resulted in increased understanding of these and other aspects of electrocardiography. This has been evidenced by material presented earlier in this paper dealing with cost factorso volume capacity$ and sources of physician-computer disagreement. 00 Data BAsLe Much of our current knowledge of electrocardiography is based upon analysis of relatively small numbers of tracings. In partt this acc.ounts for the multiplicity of diagnostic criteria resently in uoqi It is in this Aria that computer technology will probably have the greatest impact. The enormous capacity of computers permits analysis and clinicovathologie correlation of hundreds of thousands of electrocardio- grams* enabling us to more accurately define the limits of normality and the determinants of abnormality.' The application of computer techniques hit already pointed but, the limitations of conventional electrocardiography,, and will continue to add now knowledge. Purthik research in these areas, including th lication of,analog computetog multivgriat6 anal sis, a &pp y and high frequency electrocardiography can be expected to alter, the EKG of the future. 24 VI. Limitations The preceding pages have dealt almost entirely with the positive aspects of CAE systems. Unresolved problems do 6xist, however, and need to be identified. A. Costs Provider One persistent problem relates to program start-up. Estab- lished CAE systems with adequate support, in terms of operating volume., are able to sustain themselves on funds generat ed by service charges comparable to the fees received for conventional EKGS. In other words, A potential subscriber can purchase the service from one of several functioning systems.rather than,re@ying,on local resources without increasing the coat to his Pat ient. But the funds accumulated by the Provider in this way cover.only cost of operation. The money expended during program initiation, development and growth may not be recovered. The commercial interest or institution organizing a CAE system must have access to sufficient capit al to cover the costs of start-up. In addition, a period of time must be planned for acquisition of an adequate market and over- Coming technical difficulties. The magnitude of these variables have not yet been clearly defined. Maintenance and operating costs will be determined by local,needs and by willingness of local physicians and institutions to subscribe to the 25 uccessful because of their service. Several CAE systems have been uns inability to establish an adequate market. Utilizer Those individuals and institutions interested in subscribing to a CAE system,must first assess their needs and resources. Partic'l- .Pation in such systems is. expensive, Total direct and indirect costs have not yet been clearly delineated, At the present time, verification by an experienced electrocakdiographer must be.included in any CAE system, The, potential cost benefits detived from the system must be analyzed to separate data acquisition savings, from the savings der ed by reduction in the time required for EKG interpretation. The former is applicable to all EIKG systems, The cost to a physician's office, or a small EKG laboratory must be evaluated against the costs of alternative EKG interpretation systems. Though CAE systems utiliz ation may be the most Advantageous arrangement in certain situations, other approaches existi and should be considered when planning for improvement of electrocardiographic Services. B., p rforma Because of the rapid development of computer technology during the past decade, C!AE systems should be considered to be in,d transitional phase'with respect to'capabilities and range of services. Disagrdement 26 among experts as to what constitutes optimal system components, both hardware and software, continues. Further refinements of data acquisition carts, rhythm and comparativ storage and retrieval ca abilities e programs, p are to be expected. Increased interest in alternative approached, such as multivariate statistical analysis and increased usage of analog computers, will alter the nature of CAE systems of the future. How rapidly these changes will occur is difficult to predict. Whether or not a completely automated EKG system is achievable is also debated. Artifacts, complex arrhythmias, minute inconstant deflections, the need for comparative interpretation, and transmission difficulties are all challenges to this goal. The advantages anticipated with utilizing a CAE system to provide physician assistance must be evaluated against the adequacy of existing services in the region or institution. Emphasis should be directed towards@the delivery of high quality' electrociirdio- graphy by physicians and institutions ladking this service to their patients. Of the various benefits described in section V, the most important, from the viewpoint of RMPS, is the potential for extending the outreach of the expertise within the computer center into the surrounding community. The development of a CAE system to serve a single large medical center complex, without linkage to surrounding institutions, is of lower priority. Though the quality of electro- cardiography as practiced in these centers may be enhanced by the .computer interface, such deployment would have little influence upon the EKG resources of the surrounding region, where the need is more Apparent 27 Related to this, though on a more philosophical level, ate the inherent limitations of the most popular CAE systems; i.ei, those pro- grammed to, reproduce human deciaion-making and logid. Designed to mimic the expert electrocardiographer, such systems assume the limitations of conventional electrocardiography, Some have suggested that greater benefit would be derived by employing computers to do what the human mind cannot do to examine parameters and variables ignored in currect practice. VIII Potential RMPS Roles In direct,response.to the question posed by RMPS, members of the ad hoc panel suggested several possible roles for involvement of individual M4Ps with CAE systems. The following suggestions reflect the opinions of the panel members and Are not necessarily in agreement with RMPS goals and policies. They are presented for information only. A Info=atign4, 1. Assist itL the evaluation of local and@regional EKG resources and needs. 2, Identify appropriate methods for fulfilling such needs. Be Uu-c-ationg Serve'as a@source of information concerning the availability, capabilities, and potential advantages of CAE systems, 2. Inform local physicians and institutions as to the existence of thi of EKO service,, 0 type 34 Encourage participation in such systems where local need is 28 demonstrated. 4. Assist in overcoming local resistance to implementation of CAR systeps, 5. Provide similar information with respect to alternative approaches for the delivery of EKG.service, C, Organization: 1. Encourage communication between the health care providers and health institutions of the region so that c on needs may be recognized. 2, Foster cooperative arrangements among local hospitals with the aim of improving the quality and availability of EKG servi6es, whether utilizing CAE systems or some alternative, D. Consultation: 1. Provide technical and organizational assistance to physicians and institutions interested in subscribing to or initiating a CAE system. 2. Serve as liaison between interested parties and experts in the field of computer assisted electrocardiography. 3. Provide similar consultative assistance to those interested in alternative EKG service system,, E. Support: 1. Make available limited monetary support to assist in ti)d planning and implementation of regional EKG services including CAE systems. 2. Pro vide short term assistance until these systems become 2 self-supporting. F. Research and Development 1. Encourage essential investigative work in areas related to CAE systems -- e.g., exercise testing programs, coronary care unit surveillance, ate, 2* Encourage and support proj6cts concerned with the extension of advanced technology into the special care unit environment. vtil Dep_loirment A. @Comouter Facilities Both:cot=ekeial and institutional computer centers have been utilized. Private industry has played a large role in the development And implementation.@of reliable CAE,Isystems. Sints they have proved successful* And because they can operate on a profit making basis, it is most appro- priate that CAB systems be marketed by commercial organizations, Ongoing researc.h,and development are essential in all,viable industries CAL, svotems being no exIceptioa. Continued development of system hardware,,and software should properly remain the responsibilit of private industry y The use of computer facilities within the medical center environ- ment offers additional advantages including a proximity to expert electro- citaiogiapher, educational potential availability for EKG research, an(. the possibility for establishment of a regional communications network. 30 The availability of both equipment and personnel, and the capacity for sharing computer time and programmer time are also attractive aspects of an institutional computer base. B. Data Acquisition Term na a A variety of sites for deployment of computer terminals have been explored. Most often, mobile data acquisition carts have been utilized. Efficiency is enhanced in situations where the carts are stationary and patients are required to come to some area set aside for the recording of EKGO. This is, however, impractical in many circum- stances. Small and medium sized hospitals generally employ one or several mobile carts that record from patients in the general hospital area, special care units, and in emergency rooms. Multiple terminals from a number of these institutions are linked to a central computer base to furnish the necessary unit volume. Larger institutions may employ many,carts, both stationary and mobile. A computer serving a single medical center complex may receive data from terminals situated in the areas already mentioned, as well as from outpatient facilities, cardiac cattiererization laboratories,and associated extended care facilities. Terminals situated in ambulatory care facilities including clinics, medical foundations, and HeAlth Maintenance Organitations are foreseeable. Provided that the individual facility can generate a suit- able volume of tracings, the linkage of multiple ambulatory care areas 31. to a central computer base may be practical. This has not been demonstrated. rding At this time, small groups and solo practitioners reco alimited number of tracings daily would be better served by some other arrangement. In these a.ituations,,the cost of data acquisition carts becomei prohibitive. IX. Conclusions and Repommandations .Electrocardiographic services are inadequate in many parts Of the counttyo and A need for improving these services exists. The magnitude of this need is quite variable,both among different regions and in different areas of a single region. OrgAnization of, or iub- odription to, a CAE system may b6,ond approach towards fulfilling these need i. $everal existing CAR systems have@demonstrated;their ability to improve both the quality and the availability of electro- cardiographic services, though problems of cost containment continue. RIIP involvement with:CAE@aystems may be appropriate in some instances with the following qualifications: le Regional noode@with respect to electrocardiography must be assessed as part of a more SensIral'ovaluation of regional resources@for the Imanagement of cardiovascular disease, Attention should,be directed need and priorities established. Improvement to areas of'st6atest ,of EKG@servicas :should not be considered alone but should be part of a oordinated plan for-enhancement of regional health services. 32. Isolated attempts to upgrade regional EKG capabilities are not con- sidered to be appropriate roles for RMPS. 2.. Regions concerned with upgrading EKG services should consider several courses of action. Depending upon the availability of exist- ing resources this may be the establishment of cooperative arrange- ments between neighboring hospitals, linkages between local institutions and larger medical center complexes, or use of CAE systems. The thorough consideration of all approa ches, and selection of the' one best suited to fulfill local needs, is of prime importance. 3. Institutions and individuals in regions lacking the necessary personnel, technology, or volume required for support of an entire CAF, system may consider subscription to an established system outside of the re ion if,such 'action@best serves local needs. Individual 9 RMPs may make available;to interested parties, pertinent information related to the existence of and the capabilities of these systems. Such information should,be accompanied by the following recnim-endations: a. The CAE system must provide for validation of individual EKGs by an dxperienced,eledtrocardiographer. b. The service must be available on a twenty-four hoar a day basis. c. the services-offered by the CAE system must be compatible with the needs of the..utilizer. 3-3. d. Careful consideration must be given to'all cost factors including physician validation fees, equipment rental or@purchase costs, transmission charges, and maintenance. 4. Regions possessing the necessary technological expertise, ficilities.and capability for providing a volume of EKGs,sufficient to support a CAE systemmay receive RMP assistance. This may take the form of technical consultation, organizational counseling, inati-@ tutional liaison, and participation in the assessment of regional@ needs, resources, and priorities. RMPs should not assume the respoh- sibility for provision of start-up and maintenance funds. The potetr!- tial profit factor inherent@to CAE systems makes this the concern of private industry. Limited monetary support b RIlPs may be justified y during the organizational phases of CAB system implementation. Such funding is soacifically,intanded to assist in estalblishing suitable working relationships between participants, provide for technical consultation,,,and assist in the evaluation of the proposed system. 5. EKG iyetems,receiving RMP assistance should possess the following characteristics: a. The service offered must be compatible@with the needs of the region and should include both unit validation and twenty.-four hour a day availabilit ,y. b. Priori.ty is to be given to extension of the setvice@ into the regio n so that small and medium sized institutions may ben6fit- computerization of electrocardiographic services within a 34. medical center complex, serving only the components of that complex, is not,of the,highest riority., The major objective of the EKG, p system@should be enhancing the qi4ality-and accessability of EKG services throughout the region. C. Utilization of proven hardware software, and program Ming components, so that immediate benefits can be realized. Emphasis is to be directed at,delivery of service rather than research and development. d. An ade4uate marketing strategy, i.e.,.Provision for the necessary volume,' assurance of local support and utilization of the sygtemi and A ser.vice'charg6 compatible with system viability. e. Total cost to the patient,, including validation, should be comparable,to the customary..EKG costs of the region, and preferably less. 6. RMPS should encourage individual RMP,s to seek Assistance from acknowledged experts in the field during the planning and initi- ation phases of any program. RMPs should act as liaison bet-4een the two so that appropr iate technical consultation is available to project proposers. 7. With respect to extension of computer technology into the areas of exercise testing and oft-line critical care unit monitoring,, both areas must be considered developmental. As such, at the present time, they lie outside of the direct,objectives of RIIPS, i.e., im- proved delivery of medical card. APPENDIX I PARTICIPANTS OF AD HOC COMMITTEE ON COMPUTER ANALYSIS OF ELECTROCARDIOGRAMS AND THE REGIONAL MEDICAL PROGRAMS SERVICE November 30, 1971 Leonard SCHERLISO M.D. (Chairman) Samuel FOX, M.D. .Professor of Medicine President-Elect, Read,-Division of Cardiology American College.of Cardiology School of Medicine George Washington University University of Maryland Washington, D. C. Baltimore, Maryland Hubert V. PIPBERGER, M.D. Cesar A. CACERESO M.D. Chief, President, Clinical Systems V.A. Research Center for CVDP Associates, Inc. V.Ao Hospital Washingtoni D. Co Washington, D. C. Thomas S. CARDILLOO M.D. Leon PORDY, 14.D. oj.edt Director Associate Clinical Professor of art Disease Section Cardiology ochester Regional Medical Program Mt. Sinai School of Medicine University of Rochester Medical Center New York, New York Rochester, New York Mr, T. Allan PRYOR Leonard DREIPUS, M.D. Intermountain R6gionai'M@@edical Clinical Associate Professor Program Rahnemann Medical College and Salt Lake City, Utah Hospital of Philadelphia Philadelphia, Pennsylvania Arthur E. RIKLI, X.D., M.P.HO Missouri Regional Medical Program Robert V. ELLIOTT, M.D. 'Columbia, Missouri c ity Electrocardiographic Interpretative Service Computer Center Ralph SMITH, M.D. 'Denver Colorado Director EKG Laboratory Mayo Clinic Gordon E. ENGEBRETSON, Ph*D. Rochester, Minnesota Associate Director "Cont,nuing Education Mr. Morris WHITE Florida Regional Medical Program Hewlett Packard Company Tampa, Florida Waltham, Massachusetts PARTICIPANTS Continued Harold X4RGULIES M.D. Director Regional Medical Programs Service Edward J. HINM&N, M.D., M.P.H. Director Divisibn of Professional and Technical Development John R. FARRELL, M.D. Chief Clinical.9pecialty Branchi DPTD Kenneth S. GIMEL, M.D. Health Consultant Clinical Specialty Branch, DPM Mr. Earle R. BELUE Health Consultant Clinical Specialty Branch, DPTD HEALTH SERVICES AND MENTAL HEALTH ADMINISTRATION RWIONAL MEDICAL PROGRAMS SERVICE AD HOC ADVISOLY COMMITTEE TO REVIEW SUTEMENT ON COMPUT -ER CTV.OC ANAI,YSIS OF ilTog ARDIO@ Conference Room C B Wing, 3rd Floor Parklawn Building 1000@00 - 3:10 Tuesday, November 10, 1971 Chairman: Leonard Scherlis,, M. D, AGMA I. Introduction and Statement of Purpose - Dr. Harold Margulies 11. Summary of Staff Report on Computer Analysis Of Electrocardiograms @(CAS) Dr. Kenneth Gimbal fail, Q4egtious Conerning'the Current Functional Status of CAE a stems .y A. What are the limitations of CAE systems currently being utilized with respect to accuracy,, precision and reliability? B, Can physician validation of CAE be eliminated? C. What are the actual'time savings in physician man-hours? (Also in Allied Health and Clinical) D, Under what conditions are current systems cost-effective? What are the total unit,coots? E@, @what services should be considered essential to the operation of CAE? (Optimal turn-around time; interval between analysis and validationo@24 hour day availability?) F. What art practical@locatidno for the Acquisitioo units? What it minimum volume to be cost effective G* What are alternatives for delivery of improved alectro- cardiographic services in areas of need. R. What is the role of private entmriso,in the development and Monitoring of CAS? 1. Have CAE- programs met with acceptability from-the physician community in which they are provided? In your practice, do you accept CAE without further verification? Jo, In existing programs, what percentage of operation can be considered -research and development? K, Under what conditions are CAE justified even if unit cost exceeds cost for the same service in the same area? L. Under what conditions would you use CAB as a screening tool? IV. Additional questions or presentations (10 minute limit) from participants, V. AdJourn