BRUCE I. BLUM

INFORMATION SYSTEMS AT THE JOHNS HOPKINS HOSPITAL

Computers are playing an increasingly important role in patient care. Ten years ago, systems were limited primarily to processing data; today they are becoming an integral part of the medical deci­ sion-making process. This article reviews the Johns Hopkins Medical Institutions' experience with clinical information systems and describes two applications in detail.

INTRODUCTION cluded in this category are the hospital information The potential benefit of computers in health care system, the clinical laboratory system, and surrogate delivery was recognized in the early 1960's. It had medical record systems. Excluded from considera­ already been established that about one-fourth of all tion are financial information systems, embedded activity in a hospital was dedicated to information computers (e.g., the CAT scanner), and systems with limited data base facilities (e.g., digital physiological processing. I Clearly, the computer could be used to manage this function, provide surveillance and re­ monitoring systems and EKG stations). minders, and assume some of the labor-intensive ac­ tivities. Unfortunately, the technology of the 1960's THE JOHNS HOPKINS HOSPITAL could not economically and reliably support the con­ cepts of the early visionaries. The result was over­ The Johns Hopkins Hospital (JHH) is an 1100-bed promise, under-delivery, and a decade of skepticism. hospital located in the inner city of Baltimore, Md. By the early 1970's, the diligence of researchers The hospital annually provides about 320,000 days of and developers had been rewarded. Systems were in inpatient care; there are also over 350,000 outpatient operational use; in many cases the cost of operation visits per year, of which 77,000 are emergency cases. was fully supported by patient care revenues. Prog­ As part of the Johns Hopkins Medical Institutions, ress continued, and 1975 represents something of a the hospital is also involved with teaching, research, watershed year in the scientific literature of medical and specialized medical care. information science. Prior to that year, most papers were either speculative or descriptive. Since the oper­ Hospital Information Systems ational half-life of the early systems was shorter than In 1974, JHH organized itself into a group of the normal publication cycle, many papers presented largely independent units.4 Each unit (e.g., the Sec­ defunct applications. By 1975, however, key contrib­ tion of Surgical Science or the Department of Medi­ utors reported on fi ve years' experience with a sys­ cine) was given broad financial responsibility and au­ tem. Data were evaluated and system benefits were thority in the formulation of its individual budget. identified. Although most applications still were sup­ The reorganization led to dramatic reductions in cost ported by research funds, progress was made in the growth and provided a model for other hospitals. development of transportable systems that could be With respect to information systems development, justified on the basis of cost or benefit. this reorganization implied that JHH would provide Building on newer, less expensive equipment, the central facilities, such as financial management sys­ use of computers proliferated. The 1982 market for tems and an admission, discharge, and transfer sys­ hospital' information systems is projected to be close tem. Individual functional units, on the other hand, to $1 billion. 2 Over 95070 of all hospitals of 100 beds would be free to purchase or develop local systems if or more use or plan to use computers. 3 Of course, they could be justified. The result was a major com­ much of this use will be for financial and administra­ mitment to a central system that supported all JHH tive management. Nevertheless, computers are play­ units plus the independent development of smaller ing an increasing role in the process of patient care. systems that would meet specialized needs. In what follows, we trace the use of information Figure 1 is a diagram of the clinical computer com­ systems at the Johns Hopkins Hospital. This history plex at the East Baltimore campus. The JHH central records both success and failure. Because the primary computers are located in the basement of the Rutland focus is the impact on patient care, the discussion is Avenue garage. They consist of one IBM 3031 and limited to clinical information systems. These are de­ two IBM 4341 computers with 3.5 gigabytes of mass fined as applications that manage clinical data and storage. This system supports 350 video terminals that also retain an extended data base. Systems in- throughout the hospital; it also provides batch sup-

104 Johns Hopkins A PL Technical Diges( Outpatient Research Department of Department Center Oncology Laboratory Medicine Pediatric Anesthesiology computer computer computer computer computer computer Oncology Professional Laboratory Operating fee CLiNFO Clinical Comprehensive Information Medicine room system system System child care System scheduling JHH central computers i I I I I I L ______Interfaces to ---- _J : L ______J.. ______-- other JHH systems ______..J r--=-=----:---:--:-----:-:--~-~---Patient identification Outpatient identification Administrative systems Outpatient registration Admission, discharge, transfer Core Record System Outpatient Inpatient systems systems Hopkins Patient System Radiology reporting • Laboratory results reporting Laboratory results reporting Ancillary service systems • Radiology reporting

--- On- line terminal access Hopkins Patient System ancillary support ---Computer-to-computer interface • Radiology ----- Tape data exchange • Pharmacy Figure 1 - Johns Hopkins Hospital clinical information systems. port for all of the hospital's administrative systems reports, and communicating with the JHH cen­ and some of the School of Medicine's research needs. tral computer. 5,6 The clinical functions supported by the JHH cen­ 2. Oncology Center. A pair of computers supports tral computers are divided into four general cat­ a 56-bed tertiary care facility for cancer pa­ egories: tients that also handles 500 outpatient visits a week. The computers are also used to operate 1. Administrative systems. These include the iden­ the prototype Core Record System. tification of patient history numbers from the 3. Department of Pediatric Medicine. This system name and demographic information; the re­ is dedicated to the Comprehensive Child Care cording of such general information as address, program. It is used to manage administrative next of kin, and insurance coverage; and pa­ data, present clinical information, and support 7 tient location within the hospital. quality assurance and research analysis. ,8 2. Inpatient systems. All hospital-wide inpatient 4. Department of Anesthesiology and Critical functions are now being merged into a single Care Medicine. A small computer is being used integrated unit called the Hopkins Patient Sys­ to manage scheduling for the operating rooms tem. and critical resources. There are plans to in­ 3. Outpatient systems. As in most large hospitals, tegrate this system into a much larger resource the large number of outpatient visits limits the scheduling system. scope of the automated support. Access to some current information is available through In addition to these patient care systems, other the Hopkins Patient System. systems throughout the Medical Institutions are used 4. Ancillary services. A hospital is normally di­ for research and administration. Of particular note vided into patient care services (e.g., surgery, are the following: pediatrics) and ancillary services (e.g., phar­ macy, radiology, clinical laboratory). Each sys­ 1. CLINFO. This system has been specially de­ tem that provides a patient care service also in­ signed to allow individual investigators to man­ 9 cludes functions that are uniquely directed to age and analyze data from clinical trials. ,IO the management of the ancillary service units 2. Professional Fee Billing. A commercial system (e.g., work lists for the clinical laboratory). is used for professional fee billing. This system can support new clinical applications for the outpatient units. In addition to the JHH central computer complex, there are several smaller computer facilities that are dedicated to the local needs of specific functional Development of the Systems units: Under the leadership of Richard Johns and Donald Simborg, the Department of Biomedical Engineering 1. Department of Laboratory Medicine. A net­ was instrumental in developing several major clinical work of computers manages the processing of systems in the late 1960's and early 1970's. information within the clinical laboratory . This A prototype ward management system 11 was im­ involves interfacing with automatic analyzers, plemented in the early 1970's. It was designed to or­ processing requisitions, preparing laboratory ganize and assist in carrying out physicians' written

Volume 4, Number2, 1983 105 B. I. Blum - JHMI Information Systems nonpharmaceutical orders in a 30-bed acute inpatient and any special instructions. These envelopes ward. The system processed all physician orders and have a clear window on the reverse side so that produced action sheets for each nursing team and pa­ a technician can check to see that the contents tient. Special action sheets dealt with diets, utility match the instructions printed on the front rooms, weight, specimens, intake and output, and vi­ side. tal signs. Standing order sheets were also printed. 4. After the envelopes are filled and checked, they During its trial period, the system was well received are placed in a tray for delivery to the nursing by the users. A detailed analysis that compared auto­ unit. Each tray typically contains all drugs re­ mated and nonautomated 30-bed units demonstrated quired for a two-hour period. One hour prior that the system was able to reduce errors. 12 A sum­ to the time of administration, a list of drugs is mary of the findings is given in Table 1. Unfortunate­ printed, and the pharmacy technicians adjust ly, the cost of equipment was too great to justify full­ the trays to reflect any changes made since the scale implementation, and the system was abandoned envelopes were first printed. The tray and list after the concept had been demonstrated. are then delivered to the floor for administra­ tion.

Table 1 - Summary of ward errors with and without a pro­ By having the computer manage the current orders totype ward management system. and by using multiple checks for accuracy, the num­ ber of medication errors was reduced by about 750/0 Automated Unautomated (see Table 2). Because the computer is used to process Units Units all orders, automatic billing and preparation of daily and discharge drug profiles are possible. The avail­ Orders examined ability of the profiles reduces clerical activity and (total no.) 856 857 makes timely data available in an easy-to-read Transcription errors format. (0J0) 1.7 7.3 Communication er- rors (total no.) 9 70 Table 2 - Summary of pharmacy errors with and without a Uninterrupted com- computerized unit dose system. munication errors Automated Unautomated (no.) 6 38 Units Units Failure to carry out order exactly (070) 5.8 14.7 Observations (total) 1234 1428 Wrong route 2 Wrong dose 7 76 Extra dose 3 8 A different application of a computerized clinical Unordered drug 5 18 system was developed for the inpatient pharmacy in Wrong form 4 1970.13,14 Its goals were to reduce errors in the admin­ istration of medication, to lower the number of nurs­ ing hours required for medication-related activities, The automated unit-dose system was in operation­ and to eliminate waste of medication. The system al use from 1970 to 1983. Its scope was limited to the uses unit-dose drugs (that is, drugs that are pre­ departments of Medicine, Pediatrics, and Oncology. packaged in the unit of use) and works as follows: The decision to discontinue its use was based upon its operating cost and the need to implement a new phar­ 1. Physician drug orders are entered into the data macy system for the Hopkins Patient System. An base at a video terminal. This may be done by a equivalent pharmacy system with an identical flow medical professional or a clerk at the nursing was written for the Oncology Center and now station, or in the pharmacy. At JHH, all orders operates on the Oncology Center's computer. A are entered in the pharmacy. The order entry manual, unit-dose cart system is used in all other program will not accept inappropriate doses, units; an automated pharmacy system is scheduled as methods of administration, etc. part of the implementation plan for the Hopkins Pa­ 2. After the order is entered, a pharmacist reviews tient System. the order against the patient's current drug pro­ A third system, the Johns Hopkins radiology re­ file (i.e., a list of current orders and drugs ad­ porting system, was developed in the late 1960' s to ministered). If there are questions, he will call speed and facilitate the generation of radiology re­ the physician. ports. 15 It is used throughout the hospital and fol­ 3. Periodically, the system processes all active or­ lows this flow: ders and prints dose envelopes that contain the name and location of the patient, the time of 1. On completion of filming, each study is given scheduled administration, the drug description, an accession number, and the patient identifier

106 Johns Hopkins APL Technical Digest B. I. Blum - JHMI In/ormation Systems information is entered by a clerk into the sys­ method of dictation and stenography. The system is tem's data base. easy to use. The radiologists generate reports almost 2. In the reading room, the radiologist signs on to as quickly as they can dictate, and the final report is the reporting terminal. All reports generated immediately available for verification and retrieval. after this sign-on will contain his name at the Finally, the availability of computer-stored reports is bottom. useful for research, medical audit, teaching, or re­ 3. To begin reporting, the radiologist enters the generation of lost medical records. (A commercial accession number of the case, and the terminal version of the system was marketed by Siemens; be­ displays a frame with logically arranged lists of cause of limited sales, it is no longer available.) words and anatomic diagrams specific to each In parallel with the development of these stand­ type of examination (see Fig. 2). The report is alone systems, there has been a continuous effort to generated by probing (touching) selected words produce an integrated hospital system. One key com­ or phrases; colored probe points indicate the ponent of such a system is the patient locator called availability of more detailed frames. the admission/ discharge/transfer system. In the mid- 4. Once the report has been generated, it is re­ 1970's, such a system was developed at JHH, one viewed in text form at a video terminal. If it is that both maintained the current inpatient census and accepted, END EXAM is probed and the pro­ linked the various ancillary systems. The success of cess continues. this approach led the hospital to evaluate its needs in 5. Periodically, the newly generated reports are the area of hospital information management. After printed, and the printed reports are distributed performing an intensive analysis, JHH decided to in­ to the ordering physicians. On-line access to the stall a comprehensive hospital information system. report, however, is available at all designated The system chosen was the IBM Patient Care System, video terminals in the hospital as soon as END which was originally developed by the Duke Medical EXAM is probed. Center and is marketed by IBM. 16 The needs of each hospital are different. A con­ This system has been in routine use since 1972. The siderable period of time is necessary before the re­ cost of operation is comparable to the traditional quirements are established and the system can be in-

E ROSION(S• ~ FOREIGN MAnER

FRACTURE•

- CHIP• TRAUMA• • · HEALED• •

Figure 2 - Sample radiology display from the commercially available system.

Vo lume 4, N umber 2, 1983 107 B. I. Blum - JHMI In/ormation Systems stalled. At the present time, the Hopkins Patient mary patient records, current census, and ap­ System provides access to clinical reports in the nurs­ pointments. ing units and selected outpatient areas. New and 3. Schedule patient care activities and therapies as replacement functions are being installed; several determined by clinical algorithms and research years will be required before the complete system will studies. Since most cancer therapies involve se­ be implemented. quences of actions over long periods of time, Along with the creation of a strong centralized sys­ the ability of the computer to provide remind­ tem, several functional units developed systems for ers was considered an important asset in deal­ their local needs. Two of these systems provide an in­ ing with the problem of information overload. teresting contrast in requirements and scope. The 4. Maintain clinical data for retrospective first is the Oncology Clinical Information System, analysis. designed to manage data and to support decision­ making for cancer therapy. The second is the Core In July 1975, one year before the Center was to Record System, designed to produce a low-cost sum­ open, work on a prototype began. I ? The first task in mary medical record for use in ambulatory care. developing a prototype was to convert the existing Both systems were cost justified on the basis of their automated tumor registry file into a format that contribution to patient care; neither development could be expanded to meet the needs of the Center. was supported by any government grants. The data base was transferred and programs were de­ veloped to produce patient abstracts such as that THE ONCOLOGY CLINICAL shown in Fig. 3. At the same time, work began on the INFORMATION SYSTEM development of formats that could display clinical data effectively. Flow sheets and printer plots were The Johns Hopkins Oncology Center is one of 22 designed, and, after many iterations, the formats Comprehensive Cancer Centers established as part of presented in the following sections were established. 18 the National Cancer Plan initiated by the Cancer Act The prototype system operated on a computer at of 1971. The Center has major programs in lab­ the Applied Physics Laboratory. The tumor registry oratory and clinical research, education, and col­ was a functional system with a current data base. It laborative activities with community physicians. ran in the batch mode and was used for searches, re­ Care of adult patients is centered in a 56-bed facility ports, abstract preparation, and quality assurance. that can also serve 500 outpatients per week. At any The remainder of the prototype system was used on a one time, 2000 patients are being treated under one or more of several hundred formally established treatment plans called protocols. lOtiNS I-tOPKINS u Oc:, P I TA L ~1~ T 0 RY 'lJQ : QilCrJL 'GY CENTF'R "lAUE : Development of the System :),Q,TE: '1/2/f<2 PHnN£ CL A.5 5 'IN: The development of the system was begun in the SP(lUSE ORG 0 211210~1 A I RT I.j OATE O'\/OQ/30 p n p, (n o ~n "' TTT£n 0 2 /02/ iR OTSCl-itlRC,EO 0 3 /1 S / 1A mid-1970's, when plans were made to consolidate SEX ~ QACf WI-I I TE MAP . STATUS SING L r;" various oncology activities into a single facility. fl"'JL.LOW li P s ro. T US • 09/0 5/18 DEC " A SE D Ot A(:NnSTS PRTIoI AQY .. 1 Although there was no funding for a computer or for -_...... --_ .... -----_ .. -_ .... -- -_...... -_ .... ------_ ...... -- -_ .... -_.. -_ .. -_ ...... -_...... -_ .... ~ software, the facility's architecture and staffing pro­ OU TS I DE OI AGN(1SIS 1211' : ANNE A RU NDEL GENI •• HnSD . jections were predicated on the assumption that a J~H DI AGNOS IS 02118 computer system would be available. Over 250 wall :1 62.3 U PP~P LO BF . LUN" : 80 42 / 3) OAT ,ELL CAReINO. A ( T-1 62) jacks were set into place, with cables terminating in a : GR An E : POORLY DIFFER EN TlATf. D : EXTENT : DISTA NT computer room. : LATF RL: LT OR.. AN INVOLVEM ENT ONLY ;BAS I S OF DIAGNOSI S : P AT ~OLOGY. CYTOL OG Y. X- RAY. OT HER: OJ HH REPORT S During the period of planning, it was decided to .: C ARCINOHA or LUL LUNG WITH MET. . ; ' 8-1 - BONE .AP.OW BX .: HETlSTl T I C SMA LL CELL CA P C I~ O M' CONSI ST ENT: develop a prototype system. At that time, outpatients :WITH LUNG PRIM> RY. OSL 18-1759 - MED IASTI NAL Ly.PH NODE ( OSL): : were being treated in a Johns Hopkins clinic, and in­ : . ETA STAT IC SOA LL C E' LL CAR CTN OO. CO NSISTENT WITH LU - G PRIUPY. patients were being treated in facilities at Baltimore ~ ~ ;;~;;;~~; .. -~;; ;;;;-:-;------_ ...... -- .. ---_ .. ------_ .. ~ City Hospital. At each site, procedures had been es­ ~ - - ;; ~ ;--~- -- ; ~ --- ~-; ~ ;~ -~ -; ~;; -~ - ~ ~ ~-~-~~;~;; ;;; ~ ; ------: : 1 2 171 : BIOP : : OTH : BPONCHO SCO P Y BX. : 12111 : : BTOP : : OTH : " EDT ASTINOS COPY BX. tablished for patient management based on the care­ : 0 2102118: : CH E" : : JHH : NO TE 1 : 0 2102118: : BTOP : : J ~ H : NOTE 2 ful evaluation and monitoring of clinical data. : 061 14 /18: : RIOP : : JHH : MNE .A PR !)" ( POS . 18-9 524 ) : t~O T E S - ... - - - .. ---- Based on the Oncology Center's clinical practices, : 1 J '1'19 ... CY T O '(&,>.J. ,l ORT A,"' YC I N, AMO VP 1 6 it was determined that an information system was re­ quired that would H T5TOPY : .. ----- ...... -----.. .. ------.. ----_ ...... --.. _-- -- -_...... -_ ...... _...... :

:~ E n I C A L HI STORY 1050 P EP TIC ULCER. 1965 TB 1. Organize and display the clinical data as an in­ : SOC ll L HI S TORY Str,(nI(E: 2 p OD X 2 1} YRS. 3- 4 PDD X 8 YRS. : ON SET nf' sy li!. PT n ~ s 1 2 / ": PT . n , ANK P olI N tegrated report that combined inpatient and : '1 CC UPAT I QN TEL EPH ON < n p rR AT n R : ALLERGY PENICI LL IN outpatient data in ways that could indicate :COVJIIF:NT S DE'I . tr.ln~: BAPTTST trends and facilitate decision-making. 2. Manage the basic patient record functions, in­ Figure 3 - Sample patient abstract, Oncology Clinical In­ cluding a hospital-wide tumor registry, sum- formation System.

108 Johns Hopkins APL Technical Digest B. I. Blum - JHMI Information Systems limited basis. Because of the difficulty in establishing there was no automated backup. It was therefore de­ a current data base, the display system was most ef­ cided to add a second computer to permit adding fective for retrospective studies; its primary contribu­ functions and to provide backup. tion was the establishment of format requirements The second computer was installed in 1980. Stan­ for outputs. dard MUMPS was now available; the previous ver­ In 1976, in part as the result of a gift from the Edu­ sion was no longer used. Because the two versions cational Foundation of America, sufficient funds were sufficiently different, it was decided to repro­ were available to purchase a computer and support gram the system rather than convert it. This became the development of a system for the Oncology Cen­ known as the Phase II system. It also was decided to ter. Called the Phase I system, it would be a conver­ use a new programming tool, TEDIUM™ , which sion of the prototype to an interactive system on a would provide data base management system func­ dedicated computer. A PDP-II computer was-8elect­ tions and improve system maintainability. 23 -25 U nfor­ ed with the MUMPS operating system. Work on the tunately, TEDIUM was only in its initial stage of de­ prototype stopped, and development of the new sys­ velopment, and the staff was faced with the follow­ tem began. 19 ing problems: (a) conversion of what was, by now, a Figure 4 presents a summary of the development complex system, (b) maintenance of the old MUMPS activity. As can be seen, the prototype system was system, and (c) use of a new and poorly documented converted to an on-line system with the prototype's tool. The result was a major slip in the schedule and data base in nine months. The reasons that the sys­ some user frustration. However, the conversion was tem could be rapidly implemented were (a) the exis­ successful. Parts of the new system were put into op­ tence of a prototype so that the design requirements erational use in 1981, and the old MUMPS system were well known, (b) the use of MUMPS with its was retired in 1982. powerful data access tools and interactive debugging The two computers are now linked with distributed capability, and (c) the fact that all activity was de­ data base software; continuous clinical and adminis­ voted to new applications with no effort spent on sys­ trative support is provided. There is a direct link to tem maintenance. 20 the Department of Laboratory Medicine's computer Development of the system continued and new system, and all test results are automatically trans­ functions were added: bacteriology reports sorted by ferred to the Oncology Clinical Information System. date, organism, or specimen; appointment system, The computer also supports an Oncology Center including scheduling of tests; pheresis system to man­ pharmacy with a common data base. age blood product donation and transfusion; admin­ Because the system manages a very large data base, istrative functions such as charge capture and report­ attention is now being directed to the implementation ing of level of care; and protocol-directed daily care of tools to manipulate the data base for retrospective plans. 2 1,22 analysis. Called the Phase III system, parts are in op­ The success of the Phase I Oncology Clinical In­ erational use; all basic tools will be in place by the formation System began to limit its effectiveness. end of 1983. The system became saturated at 20 concurrent jobs, production processing required 20 hours a day, and it Data Display and Management was difficult to add new functions to the existing sys­ The data base contains information on about tem. Further, during periods of computer downtime 37,000 patients, including 9000 who have been treated in the Oncology Center. For all patients in the data base, it is essential to have the basic identifica­ 1975 19761977 1978 1979 1980 1981 1982 1983 tion (name and history number), demographic data I I (e.g., age, race, sex, place of birth), and diagnosis Prototype (e.g., location of disease, type of disease, date diag­ • • • nosed). For the patients treated at the Oncology Cen­ ..t. Center opens ter, additional information is required for therapy, t Computer delivered administration, and evaluation. L..-.,. On-line access The abstract, shown in Fig. 3, provides on-line ac­ I L-A Daily care plans cess to much of the summary data required by the health care team. The abstract contains identification Second computer delivered and administrative data. For each primary tumor '--___~, ~ew programs t site, there is a block that gives the diagnosis and a I In use I converslon. summary of treatment. These blocks combine both complete codes and text. For example, for the patient whose L:Analytic abstract is shown in Fig. 3, the disease is coded as tools 162.3, Upper Lobe, Lung. While this is effective for t------IITEDIUMTM development searches, there is a less precise text description that is more meaningful to the physician, i.e., "Carcinoma Figure 4 - History of the development of the Oncology of LUL lung with met." A summary of the pathol­ Clinical Information System. ogy report provides more detailed information.

Volume 4, Number 2, 1983 109 B. I. Blum - JHMI Information Systems The presentation and searching of abstract data otics given, the blood products transfused, and the are generally straightforward. The management of temperature in degrees above normal. clinical data, however, is more complex. Some pa­ To illustrate how such plots can be used in patient tients are treated for many years, and all of their in­ care, note how this single plot combines information patient and outpatient test results, medications, and to provide an overview of the patient's progress. vital signs are retained. The data base now contains Therapy for the leukemia with cytosine arabinoside over 4.5 million data points for those patients treated and daunomycin was begun on February 16. With at the Oncology Center. It is not uncommon for pa­ the convention used by this protocol, this date is re­ tients to have over 100 data points recorded in a numbered treatment day (TX DAY) 1. Based on a single day. Clearly, special tools are required to pre­ laboratory research model of cell kinetics, a second sent these data in formats that facilitate medical cycle of cytosine arabinoside chemotherapy is given decision-making. 26 beginning on day 8. The effect of this treatment is The most common format for data display is a readily seen. A large number of malignant leukemic chronological tabulation called a flow sheet. The sys­ cells (W) are eliminated, falling from a pretreatment tem provides a capability for ordering flow sheets level of 80,000 to a level of 100 by day 5. (see Fig. 5) containing specific data for a given time At the bottom of the plot, the patient's maximum period. This presentation is clearly superior to the daily temperature is indicated in degrees above nor­ collection of laboratory slips that usually confronts a mal (centigrade). The temperature is elevated at the physician; in fact, in most settings a physician gen­ outset of treatment. As a toxic reaction to the chemo­ erally transcribes the information on the laboratory therapy, bone marrow aplasia (signified by a white slips onto his own flow sheet. blood cell count of less than 1000) is added as an ad­ A major deficiency of the flow sheet is its inability ditional medical problem. Two antibiotics are begun; to indicate the time axis. This deficiency is effectively the temperature falls to normal. Protocols for the overcome in a plot such as that shown in Fig. 6, management of fever in the absence of normal white which shows the first 36 days of therapy for a leuke­ blood cells are a routine part of the organized ap­ mia patient treated with a combination of two anti­ proach to clinical management that must be taken in­ tumor drugs. The platelet and white blood cell counts to consideration in the design of this computer are plotted on a logarithmic scale. Below the plot are display shown the chemotherapy administered, the antibi- To continue with this illustration, the impact of a standard protocol for administration of blood plate­ lets is also seen. Platelet levels, plotted as "P" on the graph, are monitored daily. When the number of h lSlllWi NO: 00183014& t-.lAfio;l: •••••••••• platelets in the blood falls below 20,0001 cubic milli­ v A'IE. : v41.!O/82 STAN [) A~ [) "l.no meter, there is a danger of hemorrhaging, and plate­ : ... -.. ------.. --_.. .. -_ .. --- : .. ------: .. ----- .. : .. ----- .. : ------: ------.-.-.: lqrlAH81: I Orl AR87 : 31 MApa2 : 0 I APkH2 : 01 APH82 : 0 1 ~ P ~ S 2 0 4A PR82 : let transfusions are given. This plot uses a horizontal J7Rl lI P .y 3~:llAY lo: nA Y J7 : DAY 38 : DAY 39 : DAY 4 0 DAY 41: : ,J7 '!l4 I.~y 11 : [) AY 12 : I)AY 23:IJ Ai 2 4 :UAY l5 : DAY 2 6 D ~Y 2 '1 : : PAT Sl AT tiS------: ------_: • ___ • __ : ______.. : ______.. : ______._ .. : line drawn at 20,000 to provide a visual guide to the : TE:'.P Df;G C 37 . ,1: 37 . 5 : 37 . 4: 3"/ .5: 30 .9: 37.4 37.3: medical management team. Transfusions of platelets : Pll[.S~ . 120 : 122 : I L~ : 1 2 4: 1 00 : 1 0 4 148 : : Hf,SP I ~H' 74: 70 : 24 : lO : 22: 12 24: : army ... f A" KG 61 . 1: 6 1. 0 : bl . ~ : 0 1. 2 : 60 .7: 6 1. 0 bl. C : and the responding rise to higher levels on the follow­ :F ~C 3 . ~0: 3. 7 0 : 3 .6: 3.~: : (, ARE Lf nl, 4 : 4 : 4 : 4: 4 : 4 4 : ing day are indicated along the bottom of the plot :f' LU 10 flV) .[. 1465 : 135 0 : 940 : 2 115: 1 6~0 : 4110 2435: :,LUln fP;J) % Q4 0 : 1 5 4 ~ : 675: 4 80 : 600 : 0 3 0 : and in upward deflections of the plotted line. In this : TO fl. [ " lAK ML 24 0 ~ : 2895 : 1~05 : 25~5 : 2290 : 1235 246 5 : : llI'! '''; l llll ~L 14~0 : 9 7 5 : l 0 75 : 13 HO : 1000 : 1875 8 7 5 : patient, it is clear that platelet responsiveness exists : URIHSI'lIOL CC 0 : 0 : 0 : 0 : 0 : 0 0 : : 1. I i! STnLlL ~1. 5 0 . : 250 : 0 : 8~0 : 275: 0 8 00: and continues throughout the treatment. : F"rS nlfr ~I 0 : 0: J~ O : 550 : 0 : 0 5 0 : : 1 UTL Oll l ,"I, l500 : 1 22~ : l425 : nso : 1275: 6 5 0 1725: By day 19, normal white blood cells are seen to be : I'UlIIiT I' [f "I. - 95 : [67 0 : - HoO : - 185: 1015 : 5 8 5 740 : : KCALfl . ~ . ) ~C4L 1144 : 936 : : PH II 'lfl . V. ! G 54 : 39 : returning, and by day 32 the patient is shown to be in : KCAI.(P . II . ) KCAI . 912 : 1430 : : PP01( P . ·I . ) G 45 : 5 7 : remission from his leukemia and ready for discharge. :TOT CAL KCAI , l276 : I b9 1: : TIlT ~" fJ r (, 9~: 74: : :: : The use of daily plots combined with composite : Ht.~A r( ILI'G y - .. ---- .-.----: ------_: ______: ______: ______: ______.. __ : : WBC o/C U 3700 < 3000 < 1bOO * 5400 : 5000 : plots of the same variables (mean and 95070 confi­ : HCT " l].6: 34 . 0 : 3 0 . 0 : 33 . 0 : 33 . 0 : 34. 0 32. 0 : : Rf;llC'L("YTf. \ 1 . 4: 5 . 0< : 3 . ~ : dence limits for ten or more patients treated with this : PLAH,Lf. IS I~.l ijlOOO ' 8~000 - bY OOO * noae* I lbOOO< 1 05000 89000 - : Mnl.flCtH.:; \ I : :~~T~ "{(.CY1 1; I: cytotoxic regimen) help in assessing this patient's sta­ : I. 1; 10 ' 10 * 13* 9 < response to the combined experience of the total : Mu'oC'yn:s 1; Y: 11 < b : 19 > : NEIITROPIl 'I; '11: 62 : k O: : 6 4: : : CHf" 1 S,.~ t ------: ---____ : ______: ______: ______: ______: group under treatment. The availability of actual da­ : S,p " A ~f.V/L 1110 1 33 < 1 15 < 1 37 : 1 36: 13 8 136: ta to accurately describe past and current clinical ex­ :Sf.R ~ ~~ V /L 4 . 0 : 1 . 7 < 4. 2 : 4. 0 : 4. 8 : 4.9 4.6: : S;;R l'L Mf ~I /L Y5' Y5 * ys * 1 0 1: yy : 1 0 5 102: : St:R c n 2 ·'!' V/I. 29 : lO : 3 0 : l6 : 2 7: 25 24: perience is considered an important basis for rational : AN / .), G~P 11 : 12: I I : 1 4 : 15: 13 1 5 : : S~R Il~ HI "(; / IJ L 17 : ) 3 : D : 1 ~ : 14: 14 II> clinical decisions. In addition, these features of the : :;f~ CkEAI ," ';/IlL O. Y: 0 . 8 : 0 . 7 : 0 . 7 : 0 .7: O.b 0 . 3 > : SlIN/C"EAI 13 : 16 : II : ll : 20 : 2 3 3 7 : system support the Center's clinical research pro­ : SfR (,Life "(;/ I) L 144* 149 ' 1 38 ' 1 06 : 10 9< 148 154' : f 'JT Sf- P Vl v~· I Ol . 5. 2 . grams. : Sf. R AL U'IMN G- / OL 3 . 2: 2 . 9 < : TUT kll. 1 i'\G / LJ L 0 . 3 : In this example, more than a month of therapy has : O l~ "11.1 "'(, / ('\[ , 0 . 1 : 0 .1: : SGfl T l u n 26 * 3 1' been summarized in a graph to show the basic pro­ : SGPl l UlL : : : 48<: : : ------.- .... _- : ------: ------: ------: ------:------: cesses of tumor treatment, control of infectious Figure 5 - Sample flowsheet disease, and blood product support. During this

110 Johns Hopkins A PL Technical Digest B. I. Blum - JHMI Information Systems

J OH "S HOPK J kS NA" ~: UF 'kHlTf AN D PLATELET DAT A f) fril COLOGY CE~Tt : p SfflllLflC, FLlJl Ct.LL HISr: (CIJUPI!S 101.1 - 1, 000,000) DAT E : 0 4 /0 4 /7 9 0 8:35 1000000 ~ ... --- ...... ------...... - ... - ...... -- _ ..... - _ ... --- -- ... -- _ ... -- ... - - .... - ..... -- - ...... - ...... - - --...... ---- _ ... -- ... -- ... --... ----... - ...... -_ ... - --...... -- ~ 1000000

: :

p p . pp ... p .pp . ... 100000 • ...... p p pp 'pp.p : : ' Pp' p p p .PP' P :' P " ... i' P P p .p p p p p P P : • PI'" t-? F... . ' Pi" • • p. pp.pp.: : ___ •______•_ -- p_- ... -.... --p-- "' V' - - ...... --..... -----... - ...... --...... - .... ---... ------...... --...... ---... p------..... p-.. -p ...... --: : P ... P P P P : p , :

"" .""."': .'" • . .W'\II W' : lit W'''W': 1000 : _ ... ______... ______...... ____ ...... ____ .. __ ... __ ..... ______...... - lIII -- ... - ...... "w ..... ------.. ------.. --w- .. - .... ------..... - ... -: 1000 : lit '" WW.WW'." W' : \II "Iff' : : : ... O't\fro ."',.. ~ ""' ~.,..\Url ... . : C 't TOS I It:: A ~ ----~ .. - ~ - ... ~ - ---... ----...... - - -- - ... ------... ---.. - ...... - .. - .. --..... ------.... ------...... -- - .. - ...... ------.-.... !CYTOSI HE A Figure 6 -Sample plot. UAUNU" YC J "' : 0 I) U :DAUNOMYCIN bU . ~.L PRt':P ~ -- .. ------; --; --: ... - ; .. - ~ - ... ~ ... ; _ .. ; .... ; -- ~ .. - ~ ..... ; .. - ; ... ; --; ... ; .. - ;--;.-;.. -;--;--;--;--;--; .... ; .. -;--;_ ... ;-.. ; .... ;_ ...... ~ BOW EL PREP

Gt: ~ U "lC 1 1 ~ : G G <; G (, G G G G G G G G G G eGG G G G G G C G :GENTA.JnCIN CA P ijf.~CLL "' : C C C C C C :CARBENCLLN KlFl. 1N ! t.. K :KEfLIN CL pw .. .. YC~: C C C C C C C C C C C C C c e c c c C C : CLINDAiIIYCN P£ NI ClL- V : :P[NICIL-V Tl~P ~ - ; .. -;-- ~ -- ; - .. ; --:--: - .. ;-- ; .. -~-- ~ - - - ~ - .. ~ - -;--;--;-- ~ -- ; - .. :--:-.. ~ - .. ;-- ~ .. -:--~--; .. -; ... ~--~- .. ;--:.. -:--:-.. :--;--;---; TEMP PLU Tk u S : ') 5 I 5 5 1 6; 5 5 :PLAT TRA.NS wac TIU~S : t t l 1 1 1 2 1 1 : wac TRANS Ti q~~ ¥ : ~ ~ :-~ .. - ; _00 ; .. -; --; .. -; ..... ; ... - ~ -- ~ .. ; ~ -;;-;;-;; .. !::;;-;; .. ;;-;; .. ;;-; ~ .. ;;-;;-;;-;;-;;-;; .. ;;-;;-;;-~~ ... ;;-;;-;;-;;-;;-;;: IX DA. Y

OAl' f: 1') Ih 17 I~ 19 2 (} 11 22 2J 1 4 2~ 4!b 27 Lb 0 1 (12 u3 04 05 06 07 0 8 09 10 11 12 13 14 15 16 17 18 192021 22 21 DATE period, other organ systems are under stress, and ulation in order to produce a consistent set of similar plots may be used to monitor the function of data to evaluate a hypothesis. individual organs such as kidneys or liver, selected 2. Individual therapy. This involves the use of general functions such as nutrition or fluid balance, antitumor drugs generally in multi drug combi­ or specific medical complications such as hyper­ nations administered using complex time-se­ calcemia. Each day approximately 100 plots are quenced relationships. Therapy may extend for printed and distributed to the patient areas in time months or years. An example is the use of a for morning rounds or an outpatient visit. drug sequence demonstrated to be effective by a previous research protocol. Daily Care Plans 3. General support and response to life-threat­ Daily care plans were designed to assist the physi­ ening crises. These involve the use of estab­ cian who must treat many patients over long periods lished procedures to deal with anticipated reac­ of time using complex treatment modalities in both tions and problems associated with the anti­ an inpatient and an ambulatory setting. In a less tumor therapy. In many cases, these actions are complex environment, McDonald has shown27 that specified in each research protocol. Examples there is a problem of physician information overload are (a) the use of antibiotic drugs to treat infec­ that can be alleviated by use of an automated system. tions when it is known that the antitumor drug Automated pharmacy systems have illustrated how will severely lower the white blood cell count routine surveillance for drug-drug interactions causes and (b) management of fluid balance. 28 physicians to modify their therapeutic orders. ,29 4. Disease-specific continuing care. This entails Wirtschafter et al. have demonstrated that auto­ the recommended long- and short-term care mated procedures used in a community physician and patient monitoring associated with a spe­ outreach program can produce very high rates of cific disease independent of the therapy se­ compliance with a research protocol. 30 The purpose lected. These actions may also be included in of the daily care plan, therefore, was to introduce the research protocol documentation. Exam­ automated support for therapy decision-making. ples are (a) routine six-month chest X rays for In order to understand how an automated system all patients with breast cancer and (b) routine may be used in a facility such as the Oncology Cen­ three-month monitoring of serum protein elec­ ter, it is necessary to appreciate the general process of trophoresis results for certain multiple myelo­ patient care. Patients are treated for their cancer, as ma patients. well as for disease or therapy-related medical compli­ At any given time, an individual patient may be cations. Much of this therapy follows predefined treated by one or more research protocols (e.g., an protocols (clinical algorithms or treatment se­ antitumor protocol and an antibiotic protocol). The quences) that may be grouped into four categories: patient whose flow sheet is shown in Fig. 5 is being 1. Research protocols. These define a set of thera­ treated by two protocols. (The day of treatment peutic actions to be followed for a specific pop- under each is given in the heading.) These research

Vo{ume4, N umber2, 1983 111 B. I. Blum - JHMI Information Systems protocols may be supplemented by other support JOHNS HOPK I NS HISTORY NO: ONCOLOGY CENTER NAME: PLAN DATE: 08/06180 protocols (e.g., a hypercalcemia protocol to help GENERAL DA IL Y CARE PLAN WEDNES['AY manage high levels of serum calcium) and long-term 20 W F HODGKIN'S DISEASE HT 155 ADM WT 44 flAS 1.4 IDEflL WT 45 WT(OB/05/80) 43 . 2 follow-up (e.g., annual battery of tests). The situa­ FLOOD TYPE 0+ HLA PROTOCOL E1475 FLEO-MOPP-XET DAY 29 STARTED 07/09/80 tion is further complicated by the fact that (a) pro­ SECOND CYCLE - BLEO MOPP 8/6/80 tocols may have branches for preselected groups or PROVIDER LENHARD, DR R outcomes in which patients are given different doses PROTOCOL GENERATED INSTRUCTIONS of the same drug or combinations of different drugs, ******************************************************************* MAX CUMLATIVE DOSE OF BLEOMYCIN IS 36 MG / M2 ******* (b) the Oncology Center is a teaching institution and ************************************************************************ ENTER PERFORMMlCE STATUS thus subject to house-staff rotation, (c) at anyone ENTER TUMOR MEASUREMENTS MEASURE LONGEST DIAMETER & ITS PERPENDICULAR DIAMETER IN CM SEE PROTOCOL (SECTION 5.21> FOR CHENO DOSE MODIFICATION IF WBC < time, there are over 2000 patients being actively 4000 OR PLTS < 100000 SEE PROTOCOL (SECrION 5 . 12) FOR DOSE MODIFICATION IF TOTAL BLIL > treated by the Center, and (d) there are currently 1.5 more than 125 formal research protocols active in the MEASURABLE TUMOR SITES INITIAL LAST

Center. ID GRID OfT M/METHOD SIZE (CM) DATE SIZE (CM) DATE The daily care plan system operates by first break­ 9 . 7 7.4 07/09/80 9.7 7 . 4 07/09/80 ing down the protocol into logical therapy units TOTAL AREA 71.78 RECENT CLINICAL FINDINGS VALUE IIESCRIPT ION LAST FINDING STARTING THROUGH called treatment sequences. The system then provides 5.6 CUMULATIVE TOTAL BLEOMYCIN 07/16/80 07/09/80 08/07/80 44 INITIAL BODY WT AM 07/09/80 07/09/80 07/09/80 tools to allow the physician to assign these sequences 43.2 LAST BODY WT AM 08/05/80 07/10/80 08/07/80 97 PEF:CENT BODY WT AM E'ASED ON 44 08/05/80 07/10/80 08/07/80 to patients starting on a given date. This is called a 0.7 MAXIMUM SER CREAT 08/04/80 07/09/80 08/07/80 9100 MINIMUM WBC 07/16/80 07/09/80 08/07/80 standing order. Each day the standing orders are ex­ 704000 MINIMUM PL.HELETS 07/16/80 07/09/80 08/07/80 amined, redundant orders are removed, and a set of CHEMOTHERAPY DRUG HISTORY NAME DOSE GIVEN r. MSELINE recommended orders is produced. These are printed NITR MUST 8.3 MG 07/16/80 1 00 ;~ OF 6 MG/M2 PROCARB 120 MG 07/22/80 8S/. OF 140 MG VINCRIST 1.9 MG 07/16/80 95;' OF 2.0 MG as the plan that the physician uses to order tests and PREDNISONE 56 MG 07/22/80 1007. OF 40 MG/M2 procedures. The results of these orders are then en­ BLEOMYCIN 2.8 U 07/16/80 lOO/. OF 2.8 U DRUGS TO BE ORDERED ON 08/06/80 tered into the data base so that the next day's plan 2.80 NG BLEOMYCIN t . V., BASED ON 1.4 M2 x.:? MG/M2. 8.40 NG NITROGEN MUSTARD J.V., BASED ON 1.4 M2 X 6 MG/M2 can be derived from both what the standing order re­ 1.96 MG VINCRISTINE LV. MAX SINGLE DOSE 2.0 MG., BASED ON 1 . 4 M2 Xl. 4 MG i M:!. quires and what was actually done. 140 HG Pf\OCARBAZINE P.O.I BASED ON 1.4 M2 X 100 MG / M:!. The daily care plan is printed in several different rESTS AND PROCEDURES WITH DATE LAST PERFORME[I (> INDICATES ORDERED 08/06/80, * ORDERED STi:'T OB/06/80) formats. A physician plan summarizes information BLOOD AND SERUM TESTS >HCT HEMATOCRIT OB/OS/80 >DIFF w£!c DIFF 08/05/80 about the patient's status and tests to be ordered, a >PLTS PLT CNT 08/05/80 ; WBC WIle 08/05/80 >HGB HEMOGLOBIN 08/05/80 M12 SMA-12 08/01/80 general plan contains more complete information, IGSI) IG SURVEY 07/04/80 SCU SER COPPER 07/04/80 M6 SMA-6 08/04/80 >SALK S ALK PHOS 08/04/80 >SGOT SGOT OS/01/80 >S&F· T SGPT 08/01/80 and an order guide is used by the ward clerk to pro­ >TBLI TOT IfILl 08/01/80 >SUN S UREA NIT 08/04/81 > SCR SER CREAT 08/04/80 SURC S UR ACID 08/01/80 cess test requisitions and by the phlebotomist to ob­ RADIOLOGY PROCEDURES CXR CHEST XRAY 07/08/80 tain specimens. FHYSICIAN pr'OCEDURES REPEAT ANY X-R,n TO DOCUMENT RESPONSE OR PROGRESSION The general daily care plan is illustrated in Fig. 7. P. A. PROCEDURES MUnr- MUMPS SKIN 07/04/80 TRIC TRICH SUN 07/04/80 It PPD PPD SKIN 07/04/80 contains: DATA COORDINATor, WSTRUCTIONS

HAS DISEASE PROGRESSED? CHECK WITH M.D. RE: STOPPING TX SEQUENCES 1. Data about patient status. Examples include Figure 7 - Sample general daily care plan (see Note 31). height, weight, body surface area, admission weight, and blood type. 2. Therapy summary. This contains protocols with associated starting dates, names of physi­ dose computed as a function of weight, body cians, and general comments. surface area, or ideal body weight. 3. Comments generated by treatment sequence. 7. Chemotherapy history. This is a flow sheet con­ These are grouped into several categories, with taining the date and dose of the last administra­ the most important printed in a box to com­ tion of each antitumor drug. If a recommended mand attention. dose has been defined, the percent of the rec­ 4. Tumor measurements. Where there are solid ommendation is computed. tumors, a section of the plan records initial and 8. Tests and procedures. Depending on the for­ current dimensions and computes total cross­ mat, this portion of the plan will contain tests sectional area and percent of change since the and procedures to be ordered as determined by start of treatment. the active treatment sequences. STAT (im­ 5. Clinical findings. These are functions of the mediate) tests are flagged, and the date the test clinical data, such as percent of change in the was last ordered is given. patient's weight since admission, cumulative drug dose, or a maximum recorded laboratory Daily care plans have been used to order tests and value. procedures for all bone marrow transplant and leu­ 6. Chemotherapy orders. Where a treatment se­ kemia patients since April 1980. They have recently quence indicates that a drug is required, this been extended to cover all inpatients and outpatients fact is listed. The notice may also include the treated at the Center. 32,33

112 Johns Hopkins APL Technical Diges( B. I. Blum - JHMI In/ormation Systems

Evaluation of the Oncology Clinical Table 3 - Weekly utilization statistics (1981) for the Oncol· Information System ogy Clinical Information System. Three criteria are used to evaluate a clinical infor­ On-Line mation system: Printed Requests

1. Utilization. This is a measure of how and where Plots 500 50 the system is used. Clearly, extensive use indi­ Daily clinical sets 1000 cates that the system is an integral part of the Flow sheets 750 230 health care process. Furthermore, extensive use Abstracts 375 280 would suggest that the system is perceived to be Bacteriology reports 130 25 beneficial. Other reports 2. Cost. This is the cost of the system both for (searches, queries, one-time development and for operations. It special requests) 50 includes all costs associated with the informa­ tion system within the organization. *Not applicable. 3. Cost benefit. This is a measure of the net cost impact of the system, i.e., the difference be­ tween the operating costs with the Information to support the Pheresis Center enabled the Oncology system and the estimated cost of operations if Center to reduce morbidity and mortality; at the no computer system existed. same time, it produced a cost avoidance to patients of $240,000 a year. 22 Yet, in this case at least, the With respect to utilization, the system operates computer's contribution to relaized cost savings seven days a week and is the primary support for clin­ within the Center's budget cannot be isolated. ical data displays, clinic scheduling, patient admis­ sion, tumor registry operation, the pheresis program, THE CORE RECORD SYSTEM and test ordering. One aspect of use is how well the The Core Record System is a prototype automated system fits into the Oncology Center's operation; ambulatory medical record system designed for mul­ Table 3 presents some summary statistics. A second ticlinic use in outpatient departments. Like many measure of utilization is the number of new projects. other large urban hospitals, JHH frequently serves as During the past 12 months, a parallel system was es­ a primary source of health care for a large segment of tablished for Pediatric Oncology, all radiation ther­ the local population. The visit patterns of these pa­ apy activity is being processed by the system for both tients are characterized by repeated use of walk-in fa­ charge capture and reporting, a unit-dose pharmacy cilities (particularly the Emergency Department), system was implemented, and forms for starting new continuing care for the chronically ill, uncoordinated protocols have been developed. and overlapping use of multiple specialty clinics, and Support for the system and its Information Center occasional inpatient admissions. is taken from different cost centers within the Oncol­ Each year, 100,000 patients are seen in 350,000 ogy Center. With the exception of some separately visits. In an institution of this size and complexity, funded research activities, all operating costs are de­ the continuity of care for ambulatory patients is fre­ rived from patient care revenues. The Information quently complicated by the difficulty of finding and Center assessment is based on a guideline of 3 % of extracting information from the medical records. the charges for the services supported. (The 3070 fig­ There are approximately 1500 requests for medical ure includes a five-year write-off of the capital invest­ records made each day by the outpatient clinics. ment.) Units supporting the Information Center in­ While many requests are satisfied in a timely manner, clude Medical Oncology, Pediatric Oncology, Radi­ it is clear that the rapid delivery of records on short ation Oncology, the Pheresis Center, the Oncology notice presents a considerable challenge and expense. Pharmacy Center, and the Administrative Offices. In addition to the problem of medical record avail­ The measurement of cost benefit is a more difficult ability, there is also the need to coordinate patient task. In evaluating the system, it is clear that no man­ care within the institution. Duplication of tests ual system could provide the capabilities of the auto­ should be avoided, prescription of medication by mated system. Thus, a cost benefit analysis would be specialty clinics must be integrated, and patient fol­ comparing different capabilities. Furthermore, it low-up and referral must be coordinated. must be recognized that the traditional methods for The Core Record System is designed to meet many cost benefit analysis consider only those items ap­ of the information needs of an ambulatory setting at pearing in the institution's budget. Cost savings to a cost that is low enough to be acceptable by an out­ other segments of the economy usually are not tabu­ patient department unit. This is in marked contrast lated. Yet, some of the most dramatic cost savfngs to the Oncology Clinical Information System, where benefit patients and third-party payors and may have the high cost of tertiary care provides an economic a negative cost impact on the Oncology Center. By justification for a complex (and costly) information way of example, the system component that was used system.

Volume 4, Number 2, 1983 113 B. I. Blum - JHMI Information Systems Development of the System minimal record would also aid in coordinating The initial model for the Core Record System was care among clinics. the Minirecord System. This system was initiated as 2. On-line patient registration and processing of the result of a grant from the Applied Physics Labor­ charges. Terminals and printers would be in­ atory to produce a prototype demonstration of how stalled in each participating clinic. medical care could benefit from the use of informa­ 3. Operation of a clinic-oriented appointment system. Management reports would be pre­ tion systems. Begun in 1974, the Minirecord System was de­ pared and a data base would be maintained for signed to support the needs of the Hamman-Baker retrospective analysis. Medical Clinic, which provided long-term care for 4. Integration of the system with existing hospital 7000 chronically ill patients. It had already been dem­ administrative systems. Since the hospital is onstrated that the availability of patient problem lists starting to install an inpatient hospital informa­ improved follow-up and continuity of care, 34 and it tion system, the Core Record System must be capable of being integrated with it. was hypothesized that the availability of an auto­ mated problem list would have a beneficial effect on Development of the system began in 1979, and the patient care. Consequently, a system was specified system was in production use in 1981. An evaluation that would provide of the system has been completed, and the hospital is considering expanding it to support all Outpatient 1. Better treatment and follow-up of problems Department clinics. 37 ,38 identified within the Medical Clinic through the use of a problem and medication summary; Description of the System 2. Improved follow-up of problems identified and treated outside the Medical Clinic; The principal component of the Core Record 3. Immediate access to a minimal medical record System is its medical record; a sample record is to expedite medical evaluations during visits to shown in Fig. 8. The record contains four basic sets the Emergency Department or Primary Care of data: Walk -In Clinic; 4. Availability of on-line data to evaluate abnor­ 1. Standard identification and registration infor­ mal laboratory test results and unscheduled re­ mation (e.g., age, address). This is maintained quests for prescription refills; in a format compatible with the hospital ad­ 5. An appointment system for the clinic's 100 ministrative and business systems. health-care providers. 2. Active problem list, inactive problem list, and medication summary. This information is stored in free text, but there are plans to code Following the implementation of a prototype sys­ the text by a computerized coding scheme. tem on the Applied Physics Laboratory computer, a 3. Information extracted from other hospital sys­ data base was developed and the system was mod­ tems. This includes lists of outpatient visits and ified to operate on the JHH central computer. Access current appointments with any outpatient to the Minirecord was available in the Medical Clinic, clinic. the Emergency Department, and selected inpatient 4. Clinic-specific data. This includes patient in­ units. The system was well received and remained in structions, work release information, and other place until 1982, when the Hamman-Baker Medical clinic-defined data. Clinic was closed and replaced by the Johns Hopkins Internal Medicine Associates. 35 ,36 Data also are listed on a computer-printed encoun­ The Core Record System grew out of a series of ter form. This form contains the basic registration task forces organized in 1978 and 1979. These groups data, a summary of the current problem list and med­ were directed to identify areas in which automation ications, and space to record the progress note and might improve operations in the Outpatient Depart­ the tests and procedures performed. The encounter ment and the Medical Records Department. A solu­ form is either preprinted (for clinics using an ap­ tion proposed by these groups was the Core Record pointment system) or printed on demand (e.g., for System. Designed as a multiclinic expansion of the the Emergency Department). The registrar and pro­ Minirecord, the new system would support the fol­ vider each uses the encounter form to record his ac­ lowing functions: . tions. The original becomes the progress note that is stored in the medical record; a copy is used for the 1. Maintenance of a minimal automated am­ entry of the charges and to update the summary of bulatory medical record. This would be avail­ the problem list and medications. able on-line at strategically located terminals The system is currently in use in the Emergency and be printed as part of the physician's visit Department (60,000 visits/year), the Primary Care record. It would contain the minimal informa­ Walk-In Clinic (23,000 visits/year), the Orthopedic tion necessary for providing patient care when Clinics (4500 visits/ year), and the Oncology Center's a more complete record is not available. The Radiation Therapy and Outpatient units (35,000

114 Johns Hopkins APL Technical Digest B. I. Blum -JHMI Information Systems

JHH CORE RECORD EII£RCENCT DEPARTIlENT 09/08/81 Evaluation --- The Core Record was evaluated in 1982 based on RECISTRATION INFORIlATlON data from 32,500 encounters in four clinics over a -BIRTHDATE IV27f27 CLINIC nine-month period. 39 (The actual period varied BAl TII!ORE "D 21218 RACE B SEX F PHONE 301_ -MIl STAT among the clinics, depending on the date a clinic came on the system.) Four factors were considered: "EDICAID DATA medical information, process integration, new func­ HU"BER HA"E • FRO!! OV7b TO IV81 tions, and cost. flU NO ALLERC IES RECORDED IIff ----- ACTIVE PROBms ----- With respect to the medical data, it was shown that Z-A BILATERAL CARrAL TUNNELS ("IMIREC, 04111/801 the Core Record contained information about visits 4-A HYPERTENSION ("INIREC, 04/11/801 5-A "ORBID OBESITT ("INIREC, 04/11/801 to more than one clinic in over 300/0 of the cases. b-A ARTHRITIS L KMEE ("INIREC, m05/801 Such information normally would be available only II-A TORN l KNEE LIGMENT WRTH, 06/04/811 12-A DEGENERATIVE JOINT DISEASE LEFT LEG (ERIINT, 07/31/811 through the complete medical record. The availabil­ 13-A RT FLANI< PAIN (ERIINT, OB/WBll ity of Core Records was next considered. The results ----- INACTIVE AND OTHER PROBLE~ ----- were biased by the fact that there was only a limited I-l DISCH FRO" WIL -3 ON 0128S1 ("INIREC, OS /W811 (CIONTlNIl OR (QlUIT period of data collection prior to the evaluation. Nevertheless, within the Emergency Department and .JlH CORE RECORD E"ERGENCY DEPART"ENT 09/0B t81 Primary Care Walk-In Center, over one-third of all patients entering the clinic were already in the sys­ 3-1 SIP R PAROTID EXCISION ("IMIREC, OSIZb/SIl 9-1 eEN "fD CLINIC PROV: BASCO", REBECCA ("IMIREC, 06/191811 tem. (Over half the Emergency Department cases are 10-1 MEXT GEN "ED APPOIMmNT ON 08/07/81- ("INlREC, 06/19/811 walk-in patients with no current record.) Of those ----- "EDICATJONS ----- I ALDO"ET (m!REC, 04/11 / 801 identified, two-thirds had a Core Record. Z . HCTl ("INIREC, 04/11/80) An evaluation of the Core Records showed that 3 ELAYIL (PlIHIREC, 04/11180) 4 KCL ("INIREC, 04/11180) they tended to contain more than one problem and - 5 NALTON ("IHIREC, m05/S01 depending on the Outpatient Department unit-less 6 DARAYON (ERIIMT, 07/31/SI) than one medication. Physician compliance in com­ RECENT PA TlEMT OPD YIS ITS pleting the records ranged from 75% in the Emergen­ eEN PIED 09126/80 ORTH 10103 /80 PLAS SUR 10/17180 ER 10mlSO cy Department to 96% in the Orthopedic Clinic. ORTH 10130/80 ER/PCC 11105/80 ORTH IV04/80 eEN "ED 12105180 PLAS SUR IV05/80 PLAS SUR IVZ2ISO PLAS SUR OUZ3/S1 GEM PIED 0Zl27/S1 Availability of information was a function of type of ORTH 04/09/81 P. T. 05/06/81 ORTH Obl04/81 PLAS SUR 06126/81 patient and treatment; repeating patients with chron­ ERIINT 07/31/81 PLAS SUR 08/17181 ERIINT 081Zb1S1 ic problems required-and had-better documenta­ ffff NO APPOINTPIENTS fill tion. A brief comparison of the Core Record with the ------END OF CORE RECORD ------(0) lSPLAT (PIRlMT (NIEIl PATIENT complete medical record showed that the Core Rec­ ord tended to contain more summary information. Figure 8 - Sample core record. Where data were available from the medical record, of course, the documentation was more complete. The evaluation of both process integration and visits/year). Clinic-specific data have been defined by new functions was anecdotal. It was demonstrated individual clinics. For the Emergency Department, that charge capture and medical summary processing there are instructions to the patient and information could be integrated. New functions such as a patient regarding the patient's disposition. This disposition locator, appointment system, and management re­ information is particularly useful to the Emergency ports could be provided. The cost analysis focused on Department in that an automated "locator" file is the cost to operate the system. A target of $1.00 per maintained and the status of a patient can be deter­ encounter was considered reasonable. The actual cost mined by querying the system. For the Orthopedic was $0.91 per encounter plus a charge of $0.15 to Clinic, work release information and instructions to $0.20 for forms. Potential cost savings were identi­ the patient represent the clinic-specific components. fied, but realization of the savings would be depen­ In addition to the medical record, the system pro­ dent on reorganization within the hospital. vides two other services: an appointment system and Finally, a subjective evaluation by 21 clinicians, a charge capture system. The appointment system is administrators, and support staff in two clinics designed as an inter- and intra-clinic system with on­ found that line capabilities for entry, update, and display of ap­ pointments. Appointments can be displayed by pa­ tient, health care provider, or clinic. The appoint­ 1. The Core Record concept is easy to understand ment system allows scheduling of ancillary tests and (20 of 20 responding); procedures that need to be performed at the time of 2. Provider use of the system is not time-consum­ the next appointment. The charge capture function ing (providers only, 10 of 11 responding); allows each clinic to process all of its charges. The 3. Despite the added burden to users, the Core Outpatient Business Office can then process the Record System was worth the effort (19 of 20 clinic's bills and create a tape that is input to the bill­ responding); ing system on the main hospital computer. 4. If such a system were implemented, there would

Volume 4, Number2, 1983 115 B. I. Blum - JHMI Information Systems be an improvement in health care delivery (18 7R. H. Drachman, M. J. O'Neill, and K. V. Ledford, "Implementation of a Medical Information System in an Ambulatory Care Setting," in Third of 21 responding); Ann. Symp. on Computer Applications in Medical Care, IEEE 79CH 5. If the system were not continued, health care 1489-3C, pp. 225-228 (1979). delivery would be adversely affected (providers 8R. H . Drachman, M. J . O'Neill, and K. V. Ledford, "Computer Review of Ambulatory Care - The "CARE" System," in Fifth Ann. Symp. on only, 9 of 12 responding). Computer Applications in Medical Care, IEEE 81CH 1696-4, pp. 213-218 (1981). 9 J. C. Mabry et 01., "A Prototype Data Management and Analysis System CONCLUSION (CLINFO): System Description and User Experience," in MEDINFO The use of computers to manage clinical informa­ 77, North-Holland Publishing Co., New York, pp. 71-75 (1977). 10J. B. Johnson, Jr., et 01., "Five Years' Experience with the CLINFO Da­ tion systems has undergone tremendous change'in the ta Base Management and Analysis System," in Sixth Ann. Symp. on past decade. It is now clear that computers ca'n be Computer Applications in Medical Care, IEEE 82CH 1805-1, pp. 4 42 833-836 (1982) . 0,41 used to control costs and improve patient care. II D. W. Simborg, "The Development of a Ward Information-Manage­ Equipment prices are falling and our experience base ment System," Meth. Inform. Med. 12, 17-26 (1973). is growing. Five years ago, few major systems could 12D. W. Simborg et 01., "Ward-Information Management System: An Evaluation," Comput. Biomed. Res. 5,484-497 (1972). be started without external financial support; the 13H. J. Derewicz and D. D. Zellers, "The Computer-Based Unit Dose Sys­ next five years should bring into the marketplace a tem in the Johns Hopkins Hospital," Am. J. Hosp. Pharm. 30,206-212 broad spectrum of validated clinical information (1973). 14D. W. Simborg and H. J. Derewicz, "A Highly Automated Hospital systems. Medication System, Five Years' Experience and Evaluation," Ann. In­ The history of clinical information systems at tern. Med. 83,342-346 (1975). ISp . S. Wheeler and D. W. Simborg, "The Johns Hopkins Radiology Com­ Johns Hopkins reflects the progress in the field. puter Reporting System," Electromedia 2-3,104-108 (1975). Starting with some grant-supported projects in radi­ 16IBM Health Care Support/ DLIl, Patient Care System: TerminalOper­ ology and the pharmacy, systems were put into oper­ ators Guide. Program Number: 5796-ANY, SH20-1956-0 (Nov 1977). 17B. I. Blum and R. E. Lenhard, Jr., "A Prototype Clinical Patient Man­ ational use that continue to serve as integrated com­ agement System," in ACM Tech. Symp., Washington, D.C., pp. 5-12 ponents of the care process. In the mid-seventies, (Jan 1977). new systems were initiated in the clinical laboratory, 18B. I. Blum et 01., "A Clinical Information Display System," in First Symp. on Computer Applications in Medical Care, IEEE 77CH 1270-8C Pediatrics, Oncology, and the Outpatient Depart­ pp. 131-183 (1977); reprinted in V. Sondak and H. Schwartz, Computers ment. Each continues to play an essential role in the and Medicine, Artech House, Inc., Dedham, Mass. (1979). 19B. I. Blum and R. E. Lenhard, Jr., " Design of an Oncology Clinical In­ hospital's operation. Finally, in the early 1980's, the formation System," ACM 77,101-107 (Oct 1977). hospital committed itself to providing a comprehen­ 20 B. I. Blum et 01., "The Johns Hopkins Clinical Information Systems: A sive inpatient system. Study in MUMPS Productivity," in 1978 MUMPS Users' Group, Col­ lege Park, Md., pp. 6-17 (Jun 1978). Future work will most certainly involve the expan­ 21 B. I. Blum and R. E. Lenhard, J r., "An Oncology Clinical Information sion of current capabilities and networking of the System," Comput. Mag., 42-50 (Nov 1979). systems. (A model for hospital system networking 22 R. E. Lenhard, Jr., et 01., "The Johns Hopkins Oncology Clinical Infor­ mation System," in Sixth Ann. Symp. on Computer Applications in has been developed at the Applied Physics Laborato­ Medical Care, IEEE 82CHIB05-1 pp. 28-43 (1982). ry and is in use at the University of California Hos­ 23 B. I. Blum, "A Tool for Developing Information Systems," in Auto­ mated Tools for Information Systems Design, H. J. Schneider and A. I. pital in San Francisco. 43,44 ) The availability of power­ Wasserman, eds. North-Holland Publishing Co., New York, pp. 215-235 ful desk-top computers suggests that applications will (1982). proliferate. Portions of larger systems such as the 24B . I. Blum and C. W. Brunn, " Implementing an Appointment System with TEDIUM," in Fifth Ann. Symp. on Computer Applications in Oncology Clinical Information System will become Medical Care, IEEE 81CH pp. 172-181 (1981). available to support the management of other chron­ 2s B. I. Blum, "MUMPS, TEDIUM and Productivity," in First IEEE Com­ puter Society Int. ConI on Medical Computer Science/ Computational ic diseases. Finally, the ability to link and merge data Medicine (MEDCOMP),IEEE 82 CH1797-0 pp. 200-209 (Sep 1982). and algorithms among systems will dramatically im­ 26B. I. Blum and R. E. Lenhard, Jr., " Displaying Clinical Data for Deci­ prove utility. Thus, one can expect even more excit­ sion Making," 1. Clin. Eng. 8,61-68 (1983). 27c. J . McDonald, "Protocol-Based Computer Reminders, The Quality of ing developments in the application of computers to Care and the Non-Perfectability of Man," N. Engl. J. Med. 295, 24 medical care. (1976). 28 R. K. Halse et 01. , "Computerized Medication Monitoring System," Am. J. Hosp. Pharm. 33,1061-1064 (1976) . REFERENCES and NOTES 29 A. M. Weissman et 01., "Computer Support of Pharmaceutical Services for Ambulatory Patients," Am. J. Hosp. Pharm. 33, 1171-1175 (1976). I R, A. Jydstrup and M. J. Gross, "Cost of Information Handling in Hos­ 30 D. D. Wirtschafter, J . T. Carpenter, and E. Mesel, "A Consultant-Ex­ pitals," Health Servo Res. 235-371 (Winter 1966). tender System for Breast Cancer Adjuvant Chemotherapy," Ann. Intern. 2Report by the U.S. General Accounting Office, Computerized Hospital Med. 90,396-401 (1979). Information Systems Need Further Evaluation to Ensure Benefits from 31 This plan is for the first day of the second cycle of therapy. The comment Huge Investments, AFMD-81-3 (Nov 1980). in the box warns the physician that one of the drugs to be administered, 3D. A. B. Lindberg, The Growth of Medical Information Systems in the Bleomycin, has a maximum cumulative dose. The sections on measurable United States, Lexington Books, D. C. Heath and Co., Lexington, tumor sites, recent clinical findings, and chemotherapy history are auto­ Mass., p. 98 (1975). matically updated from the clinical data base each day. The actual drug 4R . E. Gibson and R. J. Johns, A Study of the Management of the Johns dose is computed on the basis of a body surface area (BSA, see top of Hopkins Medical School and the Johns Hopkins Hospital, JHU Internal plan) of 1.4. The general plan includes a list of all tests and procedures re­ Report (25 May 1971). quired for the protocol. Twelve tests are to be ordered from this plan. SR. E. Miller, G. L. Steinbach, and R. E. Dayhoff, "A Hierarchical Com­ General physicians comments are listed at the beginning of the plan. puter Network: An Alternative Approach to Clinical Laboratory Com­ Messages related to a procedure are li sted in the procedure ordering sec­ puterization in a Large Hospital," in Fourth Ann. Symp. on Computer tion. Most messages tend to be reminders. The next version of the system Application in Medical Care, IEEE 80CH 1570-1, pp. 505-513 (1980). will be able to generate warning messages based upon data trends. 6B. I. Blum, R. E. Miller, and R. E. Lenhard, Jr. , "Distributed Laboratory 32 B. I. Blum, R. E. Lenhard, Jr. , and E. E. McColligan, " Protocol Di­ Data Processing in a Large Hospital: The Johns Hopkins Experience," in rected Patient Care Using a Computer, in Fourth Ann. Symp. on Com­ Informatics in Health Facilities: Vol. I. Computerization and Automa­ puter Applications in Medical Care, IEEE 80CH 1570-1 pp. 753 -761 tion, M. Rubin, ed., CRC Press, Inc., Boca Raton, Fla. (in press). (1980) .

116 Johns Hopkins APL Technical Digest B. I. Blum - JHMI In/ormation Systems

33E. E. McColligan et al., "The Human Element in Computer Generated 43S. G. Tolchin et al., "Progress and Experience in the Implementation of Patient Management Plans," in Tenth Ann. Conf. of the Society for a Hospital Local Area Network at UCSF,"in IEEE MEDCOMP82, pp. Computer Medicine, San Diego (Sep 1980); reprinted in 1. Med. Syst. 6, 291-298 (Sep 1982). 265-276 (Jun 1982). 44S . G. Tolchin et al., "A Distributed Hospital Information System," 34c. J. Johns, B. I. Blum, and D. W. Simborg, "The Minirecord Approach Johns Hopkins APL Tech . Dig. 3,342-354 (1982). to Continuity of Care for a Large Population of Ambulatory Patients," in Third Iffinois Conf. on Medical Information Systems, pp. 121 -132 (1976). 35c. J. Johns et al., "A Minirecord: An Aid to Continuity of Care," Johns Hopkins Med. J. 140,277-284 (1977). ACKNOWLEDGMENTS- The author served as the director of the 36B. I. Blum et al. , "A Low-Cost Ambulatory Medical Information Sys­ Clinical Information Systems Division (CISD) of the Department of Bio­ tem," J. Cfin. Eng. 4, 372-378 (1979). medical Engineering from 1975 to 1983 . Working essentially full-time in the 37B. I. Blum, R. J . Johns, and E . E. McColligan, " An Approach to Ambu­ School of Medicine, it is natural that my contributions are closely linked to latory Care Medical Records," in Int. Conf. on Systems Science in those of the people at JHMI with whom I worked. Now that I am returning Health Care, Montreal, Canada, pp. 1025-1033 (1980) . to APL, I would like to express my appreciation to all those in JHMI who 38E. E. McColligan, B. I. Blum, and C. W . Brunn, "An Automated Core were so helpful to me. Medical Record System for Ambulatory Care," in 1981 SAMS/ SCM Richard J. Johns, is both a friend and a teacher. Dick helped me transfer Joint Conf., Washington, D.C., pp. 72-77 (1981); to be published in J. my experience to this new application area, found the financial support I Med. Syst. needed, and taught me to be uncompromising with respect to my standards. 39E. E. McColligan and B. I. Blum, "Evaluating an Automated Medical My other departmental director, Albert H. Owens, was equally inspiring Record System for Ambulatory Care," in American Medical Informatics and supportive. He provided a secure, stable environment in which sound Assoc. (AMIA), San Francisco, pp. 206-215 (May 1982). accomplishments flourished . 40R. J. Johns and B. I. Blum, "The Use of Clinical Information Systems to In the technical area, I have had special relationships with two people. Control Cost as Well as to Improve Care," Trans. Am. Cfin. Cfimatol. Raymond Lenhard, Jr., provided the perfect complement in the develop­ Assoc. 90, pp. 140-152 (1978). ment of the Oncology Clinical Information System (OCIS). By identifying 41B. I. Blum and R. J. Johns, "Computer Technology and Medical the requirements and testing the results in a clinical setting, Ray helped Costs," in Information Processing 80, S. H . Lavington, ed., North­ transform a collection of programs and people into a useful system. Holland Publishing Co, New York, IFIP, pp. 903-906 (1980). Elizabeth McColligan joined CISD in 1977 and was instrumental in making 42B. I. Blum, "Review of the Established Benefits of Automated Infor­ both the Minirecord System and the OCIS care plans effective clinical tools. mation Systems in Patient Care," Applications of Computers in Medi­ Liz manages the Core Record project and has succeeded me as director of cine, M. D. Schwartz, ed ., IEEE Catalog No. TH0095-0, Vol. I, pp. CISD. There are, of course, many others who contributed greatly to this 234-240 (1982). work. Some of them are listed as authors in the references.

Volume 4, Number 2, 1983 117