why cumul.pmd
Contents
1. immunofixat lipid Lipid metabolic markers hematin Hematinics Subclasses for Allergy pmic Pneumallergens microorganisms pdus Pneumallergens acarians dusts pgra Pneumallergens grass pollens pwee Pneumallergens weed pollens ptre Pneumallergens tree pollens ppsc Pneumallergens pollen screens pani Pneumallergens animal residues pveg Pneumallergens vegetal residues ppro Pneumallergens ocupational exposure tveg Trophallergens vegetal tani Trophallergens animal iven Inoculated absorbed allergens toxics ipar Inoculated absorbed allergens parasites idru Inoculated absorbed allergens drugs imisc Inoculated absorbed allergens miscellan Subclasses for Microbiology sbac Bacterial circulating antibodies antigens smyc Fungal circulating antibodies antigens spar Parasitic circulating antibodies antigens svir Viral circulating antibodies antigens direct Direct determination of infectious agent cult Culture Subclass for Urinalysis strip Dipstick chemical testing pressed in litre noted L to avoid confusing with the number one 1 number concentration expressed with out entity symbol as the n power of 10 noted 10En L the letter u to symbolise the greek letter m not always available on current character sets In most cases the unit expresses the measured property which allows omitting the latter Where no unit is present the result may be a serological titre titre another dimensionless value n2. the exact significance of which is lost making the confusion even worse The reporting issue The above decision to develop short test names that could be used for both mapping and reporting deeply changed the scope of the Cumul project It was then realised that with only a few more steps the new reportable test names could be made to auto arrange into a meaningful report without the need of a specific sort routine With such a page 3 feature in the Cumul layer any string of LOINC numbered patient test results would become autoreportable i e Cumul would provide the appropriate ordering sequence and hier archical structure to produce an acceptable layout The required steps were based on following assumptions e Two levels of test grouping e g chapter and subchap ter are sufficient for an acceptable report layout the subchapter level being that below which the test order becomes arbitrary Within the above hierarchy some tests still need to pre sented as ordered subsets e g electrophoresis fractions erythrocyte indices etc The first feature was achieved by carrying over and build ing upon the LOINC class The second feature was achieved by setting appopriate rules for the name construction The class subclass issue LOINC distributes tests into classes or conventional chap ters to narrow down the portion of the database which the user must browse when looking for a specific test This is not an index
3. Commisssion on Nomenclature Properties and Units to be obtained by different method for the same compo nent in the same report as sometimes occurs in infec tious serology A small number of widely acknowledged abbreviations are authorised in English and in order to avoid dupli cation of work LOINC abbreviations will be taken over as in Table 3 below Tabl 3 Abbreviations for methods AGGL agglutination CFR complement fixation reaction CIE _counter immunoelectrophoresis COAG coagulometric CZE capillary zone electrophoresis ECL _ electrochemiluminescence EIA enzyme immunoassay ELP electrophoresis ENZ enzymatic FPIA fluorescence polarisation immunoassay GC gas phase chromatography GCMS gas chromatography mass spectrometry GEN genomics w wo sugnal target amplification HA hemagglutination HAI hemagglutination inhibition HPLC high performance liquid chromatography IB immunoblot IE immunoelectrophoresis IEF immunoelectrofocusing IF immunofluorescence IFX immunofixation IMM immunological IR infrared spectrometry ISAGA immunosorbent agglutination assay LATX latex microparticle agglutination MSCOP optical microscopy RIA radioimmunoassay RIBA recombinant immunobinding assay RID radial immunodiffusion STICK dipstick TLC thin layer chromatography Chapter and subchapter The LOINC Class a mandatory field field 9 is retained as a major sort criterion for the presentation of chapters L
4. interface mask and the User s Manual of RELMA could obviously be trans lated As already mentioned translated coding tables could be prepared for each part and subpart of the LOINC names including the Component terms This being done one could think of adaptating the pre mapping utility to scan for anything that does not belong to this translated LOINC vocabulary and the RELMA engine to look for translated information when comparing LMOF terms with LOINC terms But much of RELMA s ergonomy rests on the ambiguity which LOINC can main tain between component codes and names in English The character string that RELMA brings forward for the user to make a decision remains a concatenation of LOINC codes In other words translating and adapting RELMA would be a considerable effort with little outcome Adding a layer to LOINC Local Master Observation File The fact that it cannot in any useful way be translated into another language was soon perceived as a major obstacle to the promotion of LOINC in Switzerland a country with four national languages To the the CUMUL group self entrusted with this promotion the only way out seemed to try and provide LOINC records with some sort of language specific tags that would allow indexing the table for other than English speaking users This amounted to naming tests once again an issue which very much looked like re inventing the wheel see part 1 The scope and conditions of the
5. no table for them They serve as both code and information If these are to be translated the trans lations will need to be acccomodated into a newly appended information column with the english term being retained in a code column Then if the coincidence of code and information is lost and the reader s language is remote from English the LOINC table will look unfamiliar and present many traps In other words a translated LOINC alone would require the user to continually refer to the coding tables i e the product would not be any friendlier than the native LOINC Translating RELMA The RELMA utility provides a rather complex multifunctional Windows based interface in line with two source databases e LMOF which contains relevant information imported from the local laboratory test file and e LOINC The interface essentially allows to pull records from either side with a view to relate each of the LMOF entry to an appropriate LOINC entry using elaborate search filter se lect features The package includes a utility which is to be applied prior to mapping and intended to spot in LMOF those terms which are not part of the LOINC vocabulary and to allow either replacing them with LOINC compatible terms or remov ing them This pre mapping utility is intended mainly for getting rid of locally coined abbreviations and acronyms Could RELMA be translated and applied to an LMOF writ ten in another than English language The
6. when this is an excretion g d umol d etc e Frequent values are abbreviated the choice of abbre viations was oriented by NPU rather than LOINC policy abbreviations will be taken over as in Table 1 below yet with only the initial capitalized e The value SER PLAS is substituted for S Table 1 Abbreviations for systems Amf amniotic fluid Mec meconium Asp aspirate Pt patient Bar arterial blood Pcf pericardial fluid Bca capillary blood Ptf peritoneal fluid Bve venous blood P plasma Bro bronchial aspirate PIf pleural fluid Csf cerebr spin fluid Sal saliva Cvm cervical mucus Sep seminal plasma Diaf dialysate S serum Duf duodenal fluid Spt sputum Rbc erythrocytes F stool Gas gas Sw sweat Gast gastric contents Tre thrombocytes Lke leukocytes U urine Lyc lymphocytes Vom vomitus Mar marrow bone B whole blood Component The term Component complies with the following rules e When comprising several keywords these will be ar ranged in order of decreasing weight separated by commas unless this is the natural order of the lan guage e g S Amylase pancreatic e New keywords may be added ahead of the name to en sure a structured layout and the integrity of grouped results sets as e g in Bld erythrocytes indices mean volume e Terseness takes priority over systematics as long as the meaning remains clear e g the term virus is omitted when this generates no ambi
7. Ay wy ELM Cumul Project position papers 2 Why CUMUL Jack Bierens de Haan Summary LOINC an English language product cannot in any useful way be translated into another language as it intricately mixes code and information Neither can RELMA the LOINC mapping assistant utility be adapted to operate on other than English local files The native form of LOINC therefore cannot be expected to be widely adopted in other than English speaking environments for the mapping of local test master files To overcome this limitation sets of index names were developed in several languages under the name CUMUL These names point to LOINC numbers and allow bypassing the LOINC name in most cases Selected units of measurements were added to help keep names short subclass tags were nested under LOINC classes and the syntax was adapted such that CUMUL names can also be used for patient test reporting Introduction This is the second paper in the ELM Cumul project papers series and aims at answering the question if LOINC is the only currently available candidate for a universal test directory then why not just recommend that laboratories implement LOINC numbers in their local test master files and move on to more thrilling communication issues LOINC is written in English The reason is that such a bare recommendation would meet with little success in other than English speaking environ ments that is in most Europea
8. OINC values are ordered in a conventional sequence as in table 4 below The CUMUL Subclass is optionally posted It can take the values ordered in a conventional sequence within each class as in Tables 4 Unit LOINC Field 29 ipec_units is meant to receive in the fu ture a C NPU recommended unit Until this occurs CUMUL provides a field that may or may not be distin guished from the latter and holding e Aconsensus unit which will mostly match IUPAC IFCC recommendations yet sometimes depart from strict com pliance as e g mass fraction or number fraction ex page 5 Table 4 main LOINC classes HEM Haematology BC Cell counts CELLMARK Cell markers COAG Haemostasis BLDBK Immunohaematology CHEM Chemistry CHAL Function tests SERO Autoimmune serology ALLERGY Allergens CHALSKIN Skin tests FERT Fertility DRUGDOSE Drug dose DRUG Therapeutic drug monitoring TOX Toxicology UA Urinalysis MICRO Microbiology ABXBACT Antibiotic susceptibility CYTO Cytology PATH Pathology SURGPATH Histopathology HLA Tissue typing MOLPATH Genetic anomaly Table 5 CUMUL subclasses Subclasses for Cell Counts difa Automated leucocyte differentiation difm Manual leucocyte differentiation Subclasses for Chemistry elec Electrolytes and blood gases ppt Precipitates metab Metabolites enzy Enzymes enzel Isoenzyme electrophoresis prot Proteins protel Protein electrophoresis protim Protein immunoelectrophor
9. guity e A small number of widely acknowledged abbreviations are authorised in English and in order to avoid dupli cation of work LOINC abbreviations will be taken over as in table 2 below Table 2 Abbreviations for component parts ADP adenosine diphosphate ATP adenosine triphosphate DNS deoxyribonucleic acid ENA extractable nuclear antigen HDL high density lipoprotein HIV human immunodeficieny virus HLA histocompatibility leucocyte antigen HTLV human lymphotropic virus Ig immunoglobulin IGF insulin like growth factor INR international normalised ratio LDL low density lipoprotein RNP ribonucleoprotein RNS ribonucleic acid TRH thyrotropin releasing hormone VLDL very low density lipoprotein Property The term Property calls for two remarks e It is set beween quotes because the term is taken in a wider and more flexible acceptation than the standard term more particularly to include the reference compo nent when the test component is expressed as a fraction or a ratio e g hemoglobin fraction ratio to total CK ratio to creatinine etc e The term is posted optionally i e only when there is no European consensus and it is not implied by the unit Method The term Method calls for two remarks e Itis posted only when deemed useful for clinical inter pretation LOINC rule or to distinguish results likely NPU in this context stands for the project no7 10 18 87 of the IUPAC IFCC
10. key field because tests can be uniquely iden tified without resorting to the class Besides the class a concept which more or less overlaps that of scientific dis cipline from an undefined mix of analytical biochemical and clinical points of views is not sufficiently objective to allow for a consensus definition and for undisputed as signation of individual tests On the other hand patient test reports are always arranged into chapters following a similar if not identical ration ale e g most reporting schemes will combine BC HEM and CELLMARK classes under Haematology Using LOINC classes to organize the patient test report into mean ingful subsets may thus look awkward but definitely not absurd provided that the order of such subsets is not ran dom Cumul therefore endorses the LOINC class but sets the sequence in which these are to be reported While some LOINC classes would better be combined for reporting others are definitely too broad and need further subdivision Hence the introduction of a new Cumul sub class tag for some of the classes The subclass is an arbi trary group whithin which the order of tests layout is no longer significant i e it can just as well be alphabetical Subclasses were developed for five classes see tables 2 6 There is not necessarily a subclass for each test in a class i e some tests may remain not subclassified Cumul syntax The syntax must be such that tests in a repo
11. lems as it will in fact restrict its use and deprive the same test expressed in alternative units from a deserved LOINC reference For example assigning the unit g L toa component with the property mass concentration would leave a test expressed in mg 100mL without CUMUL ref erence or stated otherwise if a substance concentration result can be expressed either as 0 00x mmol L or as 00x nmol L CUMUL will impose one of the two alternatives leaving the other one without a CUMUL reference This illustrates perhaps the reason why LOINC dropped the unit altogether but it also points to a weakness of LOINC which can deliver correctly identifiable tests yet stops short of delivering readily comparable results Considering the fairly homogeneous European context the Cumul working group decided to take the risk and post the unit thereby departing from the more neutral fully prag matic attitude of LOINC The abbreviation issue Abbreviations in the form of capitalised initials of com plex terms are extremely common in laboratory medicine The policy of LOINC is very restrictive about their use for component names and rightfully so because this would defeat the readability feature of the database Abbrevia tions would require a specific coding table CUMUL fol lowed the same policy for the same reason and also be cause the set of common abbreviations in other than Eng lish languages will most often be a mix of national and English acronyms
12. n countries Implementing LOINC numbers involves finding an equivalent fully speci fied LOINC name to each local test name and assigning the LOINC number to that test an operation which is re see ELM cumul project position papers 1 LOINC vs NPU Logical Observation Identifier Names and Codes 3 Regenstrief LOINC Mapping Assistant cumul secretariat jackbdh vixnet ch tel fax 41 22 8400341 2 y o post address centre suisse de contr le de qualit 2 chemin du petit Bel Air CH 1225 Ch ne Bourg ferred to as mapping The exercise assumes a good knowl edge of LOINC syntax and semantics and an ability to read between lines of the local file that is equating appar ently different terms and making up for missing terms in other words interpreting based on expert knowledge but it is first of all based upon just reading the lines and in any case does not assume any multilingual skill Even so mapping is not a straightforward task because a directory with exhaustive pretension will tend to list all existing tests regardless of their relative relevance or fre quency and will therefore be ten to twenty times as large as the average local file often leaving the user hesitating between two like entries or wondering whether he should submit a new one The issue was recognised from the out set by the promoters of LOINC and tentatively addressed by their provision of RELMA a browsing mapping pro gram which will take the use
13. o dim or it may be posted on a discrete ord or nominal nom scale It may not be sound practice to use the same field to ex press such different LOINC concepts as the unit which is a reference property with the value 1 the property for the serological titre and the scale But it is just a fact that in the above case the three concepts are mutually exclusive and the resulting expression is unambiguous and concise Discussion An index of CUMUL names in the user s language is an invaluable productivity tool for anyone engaged in the LOINC mapping of local test master files CUMUL names are meant to look familiar to the average user of laboratory tests a feature that was deliberately rel egated by LOINC designers and NPU experts as well both insisting that fully specified names are not meant for the end user and arguing that this is the way it should be The designers of CUMUL on the contrary realised that the conditions set for test mapping could be very close to those set for test reporting They borrowed from NPU no menclature until the test name became obscured by stand ard syntax and semantics after which they made choices that may look arbitrary but were always guided by the re quirement of readability page 6
14. r by the hand and lead him stepwise towards mutual recognition between LOINC and local jargon enabling him to assign LOINC numbers to his local tests in a tidy way Because the 29 000 LOINC entries include 18 300 distinct words in a 368 200 English word index officially recom mending LOINC is likely to remain a verbal exercise if nothing is done to overcome the language barrier where this exists Translating LOINC As explained previously see part 1 LOINC is a multi column table exclusively populated with coded variables where code stands for information e g SER stands for Serum From this point the LOINC table will be called a database to distinguish it from other tables Indeed indi vidual codes paired with their corresponding information entity are listed in two column coding tables e g for Sytem Scale Method etc The information in the coding tables can be translated but not the code Translating a code is pointless it amounts to re coding that is ignoring the very standard Translating the information in the coding tables on the other hand can be an option but of limited interest in this case because the readability of the LOINC database owes much to the information being evoked by the mne monic code such that in practice the user seldom resorts to the coding tables In this respect the Component terms in the first part and subparts of the fully specified name pose a special prob lem because there is
15. rt message when sorted by name within subclass and class and with the unit posted will become laid out as a readily under standable logically structured meaningful report Names should be as short as possible yet with scarce use of abbre viations The approach is freely adapted from C NPU principles i e departing only from these to avoid e dispersing tests that ought to remain grouped e producing redundancy e g with the unit when this is posted expressing self evidence or a wide consensus As the above criteria are rather subjective and entail much simplification the resulting name cannot be as specific as the fully specified LOINC name which therefore remains the ultimate reference for file mapping in case of doubt or ambiguity The CUMUL name follows the format System spec Component spec Property Method where spec stands for specification and Property is pur posely written between quotes the separators semicolon and comma are followed by a space for improved readability The terms System Component Property and Method are further detailed below System The term System complies with the following rules e The primary specimen comes first optionally followed by a bracketed subsystem as in U Sed Bld Whc etc e For excretions the patient is not considered a system such notations as Pt U or Pt F are substituted for U and F respectively as the unit will tell
16. st often referred to as e g on a patient test report or in conversation Lack of standards in this respect is the very source of the need for such directories as LOINC And the structure of LOINC names sufficiently demonstrates the difficulty to fully identify tests in text On the other hand the average patient test report is based on names which do not exceed 30 40 characters yet seem good enough for the target user if this is a human Now good enough is maybe not optimal Physicians in general and a majority of labo gt CUMUL is not an abbreviation it is the name of a medical informatics project initiated in 1997 in Geneva Switzerland see http www cumul ch ratories would welcome more homogeneity i e some sort of guideline for naming tests Although the working group did not feel entrusted with any standard setting mission it decided to seize the oppor tunity and propose a test naming system that could poten tially be used for both mapping and reporting purposes The unit issue One reason why shorter names are acceptable in routine work is because the unit will often tell about the kind of property the scale and the time aspect The unit is a highly informative item Because the unit is not part of the LOINC name it was decided to add this to the CUMUL layer and drop any redundant information from the CUMUL name to help keep it as short as possible Admittedly assigning a unit to a LOINC entry can raise prob
17. task however made this considerably simpler e the entities to be renamed were 6 part LOINC entries rather than real world tests e the new names were to appear on the top of the LOINC directory leaving the original record available for scru tiny by any user with sufficient basic English notions Then to meet the indexing purpose in a mapping context the tags had to be unique though not necessarily fully specified e as short as possible for ready identification and ma nipulation e highly readable i e they had to look familiar to the person in charge The idea was to build in each language an index of names that could be directly related to LOINC numbers yet for most purposes bypassing fully specified LOINC names Hence the concept of layering a set of multilingual in dexes on the top of LOINC to facilitate access to and use of the database The reportable name issue So far Cumul efforts had aimed at opening up the LOINC mapping of local test files to other than English speaking users The objective of mapping it must be reminded is to allow computers to recognise the same test behind names which are spelled differently not to impose a standard spell ing It was soon realised however that with the conditions set as above for terseness and legibility the newly coined names became close to a longed for species the standard report able name The reportable name of a test is that by which it is mo
Download Pdf Manuals
Related Search