Pii: s1386-5056(98)00090-2

International Journal of Medical Informatics 51 (1998) 59 - 68 MERIT-9: a patient information exchange guideline using Michio Kimura a,*, Kazuhiko Ohe b, Hiroyuki Yoshihara c, Yutaka Ando d, Fumiaki Kawamata e, Fumito Tsuchiya f, Hiroyuki Furukawa g, Shingo Horiguchi b, Takaya Sakusabe h, Shigeki Tani a, Masanori Akiyama i a Department of Medical Informatics, School of Medicine, Hamamatsu Uni6ersity, 3600 Handa Hamamatsu 431-3192, Japan b Hospital Computer Center, The Uni6ersity of Tokyo, Tokyo, Japan c Di6ision of Medical Information, Miyazaki Medical College, Miyazaki, Japan d Department of Radiology, Keio Uni6ersity Hospital, Tokyo, Japan e Otsuka Pharmaceutical Company Limited, Tokyo, Japan f Di6ision of Pharmacy, Teikyo Uni6ersity Ichihara Hospital, Chiba, Japan g Di6ision of Pharmacy, Kanazawa Uni6ersity Hospital, Kanazawa, Japan h Weathernews Incorporated, Makuhari, Chiba, Japan i International Medical Center of Japan, Tokyo, Japan Received 20 February 1998; received in revised form 20 March 1998; accepted 28 March 1998 Abstract
To realize clinical data exchange between healthcare providers, there must be many standards in many layers.
Terms and codes should be standardized, syntax to wrap the data must be mutually parsable, then transfer protocolor exchange media should be agreed. Among many standards for the syntax, HL7 and DICOM are most successful.
However, everything could not be handled by HL7 solely. DICOM is good for radiology images, but, other clinicalimages are already handled by other 'lighter' data formats like JPEG, TIFF. So, it is not realistic to use only onestandard for every area of clinical information. For description of medical records, especially for narrativeinformation, an standard generalized mark-up language, document type definition (SGML DTD) for medicalinformation, called MML (medical markup language) had been created in Japan. It is already implemented in morethan ten healthcare providers. However, it is again not realistic to use MML solely for clinical information in variouslevel of detail. Therefore, we proposed a guideline for use of available medical standards to facilitate clinicalinformation exchange between healthcare providers. It is called MERIT-9 (MEdical Records, Images, Texts,-Information eXchange). A typical use is HL7 messages, DICOM files, referred from an MML file in a patient record,as external entities. Both MML and MERIT-9 have been research projects of Japanese Ministry of Health and * Corresponding author. Tel.: + 81 53 4352770; fax: + 81 53 4352769; e-mail: kimura@hama-med.ac.jp 1386-5056/98/$19.00 1998 Elsevier Science Ireland Ltd. All rights reserved.
M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 Welfare and the purpose is to facilitate clinical data exchanges. They are becoming to be used in technicalspecifications for new hospital information systems in Japan. 1998 Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Medical data standard; Hospital information systems; Computerized medical record systems; Communitynetwork; Telemedicine 1. Introduction
2. What we can use now and what we still
need?
It is needless to say that exchanging pa- tients' clinical information between health- 2.1. What we can use, for example, for care facilities is very important. It is required clinical obser6ation reporting in many situations. Patient referrals betweenclinic and hospitals are the most frequent.
For syntaxes to wrap the observation data, Then hospital may transfer information to/ from special examination laboratories, or to [4], UN/EDIFACT [5], etc. A good compari- other related healthcare service providers.
son between them is reported by S. From and Any other facilities within allied healthcare E. Brox [6]. After this survey, done by CEN message format ENV 1613, for this purpose [7]. In the United States, HL7 is now most change, however, are different in each case.
widely and successfully used for this purpose.
Patient's episode history is sent from clinics HL7 now has six international affiliates.
to hospitals, while chemical laboratory com- pany sends massive amount of lab test results candidates include CPT-4 [8], and LOINC. In Japan, The Japan Society of Clinical Pathol- change, there must be a lot of agreementsbetween both sides. Fig. 1 shows what areneeded. First, a data transfer protocol, ormedia for off-line, should be agreed, to be atleast readable. Then, the syntax of a data fileshould be parsable for both of them, to knowwhat is where within the data file. In the caseof Fig. 1, HL7 [1] is used. Finally, the termsand codes in the expression should be under-standable. In Fig. 1, the left side usesJapanese lab test code while the counterpartuses LOINC [2]. Fortunately, thanks to amapping table between these codes, patient'sdata of gamma-GTP was transferred success- Fig. 1. Clinical observation 'gamma-GTP is 120' is transferred from database to database. For this, four medical information, especially for the upper layers of mutual understandings are needed. Item code, syntax, data format and transfer method.
M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 Fig. 2. Hierarchical structure of MML. 'Clinical encounter section' is described in detail. 'Abbr.' means this elementcould be abbreviated. 'Rep.' means multiplicity of this element. 'MERIT-9' in 'Data type' means this element is forexternal entity reference, described later.
mated data. Even though the syntax is now LOINC. Thus, the code is mutually convert- being brushed up by HL7 working group, it is not the way that most ECG data areexchanged, until now. ECG vendors' propri- 2.2. Wider co6erage is needed etary data standards are mostly used, whichare deeply concerned with each vendor's data [10] for the part of clinical observation re- porting. HL7 also covers orders of prescrip- patient, which are captured by clinical en- tions, nursing and intervention, etc. Looking counters. It is not, however, the usual way to describe and exchange this kind of informa- orders and reports are what HL7 is mostly tion between healthcare providers. The syn- successful for descriptions of laboratory tests, M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 prescription, etc. But, it is not suitable for radiology examination orders are much more descriptions of clinical signs, which are usu- in detail, compared with the data varieties ally described in texts of free format. We offered by HL7. [11] Therefore, many locally cannot solve everything by HL7 solely.
described elements had to be added, in orderto fetch usual physicians' image examination 2.3. Variety of information granularity is tions in each requirements for image quality.
HL7 has descriptions for radiology exami- DICOM [12] is the most successful in this nation orders. At Tokyo University Hospital, HL7 is being used to transmit orders includ- ing radiology examinations. Their implemen- minor modifications on Supplement 8 Image tation experiences made clear that, in Japan, Storage Service Class and Supplement 10 Ba-sic Worklist Management Service Class [13].
In Japan, JIRA (Japan Industries associationfor Radiation Apparatus) translated DICOMdocuments in Japanese. It is often used andreferred in technical specifications for openbids of medical imaging systems in Japanesehospitals.
Even DICOM is so successful in radiology, however, images other than radiological arecarried and archived by other image stan-dards, such as JPEG [14], TIFF [15], etc.
even though DICOM committee has recentlyreleased a draft for specifications for visiblelight images [16]. Requirements for imagequality are different in many situations, froma 4000 × 4000 full 24 bit color to a 256 × 2568 bit grayscale. It is realistic for each imple-menter to select the best suitable solutionamong available candidates, which are men-tioned above.
Fig. 3. Original presumed case of uses of MERIT-9.
2.4. Meta information to help understanding Case 1 is patient data transfer between hospital and the episode sequentially is needed clinic. It could be by floppy disk or on-line. In case 2,laboratory test company reports massive amounts oftest results to hospital. In this case, MML document Most of the successful standards including could be omitted. Then, the report consists of massive HL7 files. In case 3 laboratory test controlling center information messages. Therefore, they are for collects lab test results of this area from many test sites.
clinical information captured in a certain Case 4 shows that a patient wants to cumulate his/her time. For HL7, it usually is one order or examination history from many healthcare providers.
In case 5, hospitals may like to use this for clinical case result of examinations. For DICOM, it usu- ally is for images of one image examination.
M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 Fig. 4. An MML file refers an HL7 message as an external entity by a file name 'A0001.HL7'. A DICOM file isreferred by a file name 'D0002.DCM'.
As a result, it is very difficult to put one Use of a hierarchical description for clinical patient's narrative episode which involves MEDIX committee [19]. The idea is indispens- diagnostic reports, bunch of prescriptions, able, but, no detailed description of massive data elements for clinical information ex-change has been published. CEN TC251 isnow also proposing, but not yet being unveiled 3. Medical markup language - an SGML
a standard for message of patient referral and DTD for medical record information
As a consequence of the requirements men- tioned in the succeeding chapter, a medical prepared for now-realizing method of object request brokering between healthcare infor- in 1995, as a Japanese Ministry of Health and mation systems. DICOM is already a standard Welfare's research project for electronic med- based on objects concerning image examina- ical records [17]. It is an SGML DTD (stan- tions and HL7 version 3 is going to be on the dard generalized mark-up language, document same track. Therefore, this method of data type definition) [18] for medical record infor- description will be used for the basis of data mation. As shown in Fig. 2, it is a hierarchy transactions between them. As what is impor- of descriptions of clinical information.
tant is not the data definition syntax, but the What is in each data element, is described structural definition of information, MML by data types, which are similar to HISPP- may not be using 'mark-up' descriptions of information in future. Even then, it will still be ing data types are used: ST; TS; ID; DT; and used as a basis of clinical information descrip- M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 Fig. 5. A DTD of an MML document which refers external HL7 messages designated as URI. File names are'ca1001.hl7' and 'ca1002.hl7', both under directory 'pid001'. Minimal information (File type ( = HL72.3) andExamination date) from HL7 message contents are on DTD tag, in order to let DTD parser know what is in thereferred entity.
The research project of the Ministry is now which are crisp and already successful stan- in its third year. There are already more than dards, for detailed specialty of information.
10 healthcare providers, from outpatient clin-ics to university hospitals, which are doingthe evaluations of this standard.
4. Patient information exchange guideline:
Total list of the data elements, a part of which is shown in Fig. 2, is not very large,considering the required wide coverage. It is possible to make a very precise and detailed descriptions of the data elements for every year later. It is to facilitate clinical informa- type of orders, observations, etc. But it tion exchange between healthcare providers.
makes the standard enormous and the enor- Among many kinds of clinical information, mous standard is very hard to be kept up-to- laboratory test results, diagnostic images, date in all part. Furthermore, there are many prescriptions, are ready to be electronically kinds of non-textual data required in clinical information exchange. The typical example is providers' databases. It is because we have all image data. Therefore, it is realistic to use the necessities described in the Introduction MML for narrative clinical information and of this paper, for these three kinds of infor- mation. Treatments, interventions will follow, HL7 messages, DICOM files, TIFF files, etc.
then, diagnostic disease names. But, signs M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 captured at clinical encounters are very hard “ Tags for examination information associ- to exchange in a electronically-recognizable way. So the project concentrated on the top- IS&C, because all others are just image standards and have no information of clin- “ Detailed understanding and usage of HL7, change) is for exchanging lab test results, prescriptions, images, as well as patient iden- tification information, making full use of “ External entity referring method. We de- be referred from technical specifications for cided on indirect pointing from directory open bid of healthcare information systems.
Original presumed case of uses are shown in Usually, by MERIT-9, a patient's narrative change media, transfer protocols, encrypting episode is described by MML, with detailed lab test results, prescriptions and diagnostic uations in mind so that the best solution TIFF files, etc. referred from MML file as tion. Moreover, it is difficult to keep these external entities. Recently, HL7 formed a contents up-to-date. How external entities are working group for SGML. It is preliminarily intended to use SGML for an encoding rule of HL7 messages. Thus, it is SGML withinHL7, while MERIT-9 proposes HL7 withinSGML. Fig. 4 shows the idea.
5. Differences of medical practice, illustrated
To realize clinical data exchange, the fol- by using foreign standards
stated in the MERIT-9 standard.
“ Listing of external entity repertoire: For images; DICOM; IS&C [21]; JPEG; TIFF; in countries outside Japan, except for MML.
presumed case of real use. Therefore, any standard reflects customs of medical practice “ Tags to be on the MML file representing of the countries where it was created, more or less. This time, by trying to use a foreign- made standard, differences of medical prac- tice between countries are clearly illustrated.
what is in the external entity, though not We are reporting these things, thinking that total details of inside. Therefore, minimal they are helpful for future international stan- dardization, as well as they are interesting.
and time, etc. should be on MML tags.
Generally speaking, doctors in Japan place “ Detailed description of MML tags, keep- (pharmacy, radiology, clinical laboratory) M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 placing image examinations. Examination de-tails such as beam directions, contrast agent, etc. are to be ordered in Japan, while they are at the discretion of radiologists in the US.
Therefore, there are no fields for these data in found one thing very common in Japan, but is hard to be described by them. Single orders which are to be reported as multiple individ-ual results, such as 'Electrolytes', 'FBS (free blood sugar) Glucose challenge test', are han-dled as 'batteries'. The variety of results are determined by the examining laboratory. One laboratory may report the latter as FBS be- prescription in Japan. For example, they are fore Glucose intake, 60 min after and 120 min after, while another laboratory may re- prescription groups. In Japan, about 15% of port it as before, 30, 60, 90 and 120 min. By total prescriptions are powders. At Teikyo HL7, when order placing, expected detail of University Ichihara hospital, total number of results cannot be described by them, unless prescriptions in 1 day (1997/12/1) was 3745.
results of each timing were ordered individu- ally. In Japan, however, these expected de- usually have variety of different concentra- tions. We had to find a way to specify this information explicitly clear. Next, in Japan, tee to give us a way to describe expected macy, medicines to be taken at the sametiming (three times a day, two times a day, one in the morning only, etc.) are grouped ina bag within the whole bag. So we had to put information of prescription groups in order- are not exclusive for medical images. This ing messages. These small things were solved means they are not reflecting each country's by 'broader' interpretation of existing fields medical practice. From the beginning, when international collaboration. As a result, DI- segments. In the discussion to use HL7 for COM is prepared to meet requirements at the prescription in Japan, however, we thought US, Europe and Japan, at the same time.
This is quite desirable and natural. It is be- which are to be used for encoded order mes- cause vendors of each side have certain mar- 'Total daily dose' while RXO doesn't. In Shimadzu are selling their products in US Japan, prescription order is based on total and Europe, while in Japan, we can buy GE, daily dose. We write 'Adalat-L three tabs, after every meal, 7 days' meaning 'one tab for each time, total 21 tabs'. Total is not 63 tablets. And this style of Total-daily-dose- M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 based prescription is mandatory by law, for Acknowledgements
our valid prescription orders. We requestedHL7 committee to add Total daily dose in tion to following personnel for fruitful com- By a survey at Teikyo University Ichihara Hospital, total number of oral medications in CareNet Inc., Dr Shigekoto Kaihara of Na- tional Okura Hospital, Dr Hirozo Ueda and Among them, 3100 had instruction of timing Dr Yoshiyuki Matsumoto of the Ministry of not associated with meals. This means 93% of Health and Welfare, Professor Clem J. Mc- total prescriptions has timings related with Donald of Indiana University, Professor W.
meals. Usually, they are 'after meals', but 'before meals' (mostly Chinese traditional medicine) 'between meals' 'with meals' are not uncommon. Even for 'after meals', some- sity, Dr Teruyoshi Hishiki of the University times it is 'after breakfast and dinner', or 'after breakfast and lunch'. In HL7 v2.3, however, there is no description in this pre- Chiba University, Dr Koichi Ishikawa of Na- ciseness. We requested these detailed descrip- tional Cancer Center, Mr Mitsuhiro Aizawa of Arcobel Graphics Japan Inc., Mr HaruoKarasawa of Konica Corp., Mr Koichi Kitaof Toshiba Corp.
6. Final remarks
In the second year of the ministry project, References
MERIT-9 is now being evaluated for its im- [1] Health Level Seven, An application protocol for plementations. It is adopted for technical spe- electronic data exchange in health care environ- cifications of national university hospital ments. Version 2.3 Ann Arbor, MI, Health Level information systems for clinical trail manage- ment. Future expansions of applicable spe- [2] A.W. Forrey, C.J. McDonald, G. De Moor, S.M.
cialties of information will be carried on Huff, Logical observation identifier names andcodes (LOINC) database: a public use set of codes according to the opinion described early in and names for electronic reporting of clinical labo- this thesis. However, these information like ratory test results, Clin. Chem. 42 (1) (1996) 81 - signs obtained by the clinical encounters could be described in free text form, because [3] ISO Standard, ASN.1. ISO 8824, 1987.
MML can have them in that form. Of course [4] Euclides Consortium, Euclides Syntax Manual.
it is not transferable between databases in [5] B.J. Love, UN EDIFACT-an EDI standard for information retrievable manner, human clini- healthcare, in: G.J.E. De Moor, C.J. McDonald cians can read and understand the free text.
(Eds.), Progress in Standardization in Health Care English version of information concerning Informatics, ISO Press, 1993, pp. 156 - 161.
[6] S. From, E. Brox, Investigation of syntaxes for existing interchange formats to be used in health- the Japan Association of Medical Informat- care, in: G.J.E. De Moor, C.J. McDonald (Eds.), ics, research committee for MML/MERIT-9.
Progress in Standardization in Health Care Infor- matics, ISO Press, 1993, pp. 118 - 126.
M. Kimura et al. / International Journal of Medical Informatics 51 (1998) 59 - 68 [7] Comite Europeen de Normalisation. CEN ENV [16] Digital Imaging and Communications in Medicine 1613: Messages for Exchange of Clinical Labora- (DICOM), NEMA Publications PS 3. Supplement tory Information. Brussels: CEN, 1994.
15: Visible Light Image and Anatomic Frame of [8] CPT-4, Clinical Procedures and Terminologies, Reference for Endoscopy, Microscopy and Pho- American Medical Association, Chicago, II, 1987.
tography. Draft for Public Review. The National [9] Committee of Laboratory Codes and Laboratory electrical Manufacturers Association, Rosslyn VA, Instruments, The Japan Society of Clinical Pathol- ogy, Classification and Coding for Clinical Labo- [17] T. Hishiki, K. Ohe, H. Yoshihara, M. Kimura, Y.
ratory Tests (JLAC9 release 2). Rinsho Byori Ando, T. Sakusabe, F. Kawamata, R. Yamamoto, S. Yamazaki, Y. Hirose, K. Ohashi, K. Matsui, K.
[10] American Society for Testing and Materials, Minagawa, Medical Markup Language: A simple method for clinical data exchange, Proceedings of Transferring Clinical Observations Between Inde- The Third International Conference on the Medi-cal Aspects of Telemedicine, 1997, pp. 117.
pendent Computer Systems, Philadelphia, PA, [18] N. Bradley, The Concise SGML Companion, Ad- [11] K. Ohe, S. Kaihara, Implementation of HL7 to [19] J.J. Harrington, T.J.R. Benson, A.L. Spector, client - server hospital information system (HIS) in IEEE P1157 Medical Data Interchange (MEDIX) the University of Tokyo Hospital, J. Med. Syst. 20 Committee: Overview and Status Report, Proceed- ings of the Fourteenth Symposium on Computer [12] Digital Imaging and Communications in Medicine (DICOM), NEMA Publications PS 3.1-3.12, The [20] M. Kimura, Telemedicine Information Standards National electrical Manufacturers Association, in Japan, Proceedings of the Third International [13] H. Gnoyke, Standards for the integration of modalities with information system and IMAC [21] M. Kimura, N. Ohyama, K. Inamura, Y. Ando, N. Shigemura, Y. Shima, T. Saito, IS&C (Image ments, Proceedings of Medical Imaging 1996, Save and Carry) Standard: Standardized Exchange PACS Design and Evaluation, vol. 2711, SPIE, Media for Medical Information, Proceedings of the Eighth World Congress on Medical Informat- [14] C-Cube Microsystems, JFIF: JPEG File Inter- ics (MEDINFO-95), 1995, pp. 207 - 211.
change Format version 1.02, Milipitas, CA, 1992.
[22] ISO/IEC JTC1/ SC29, MPEG: Information Tech- [15] Aldus Developers Desk, TIFF Revision 6.0 Final, nology-Generic Coding of Moving Pictures and Aldus Corporation, 441 First Ave. South, Seattle,

Source: http://lob.kuhp.kyoto-u.ac.jp/paper/merit9.pdf