1、Designation: E 2171 02An American National StandardStandard Practice forRating-Scale Measures Relevant to the Electronic HealthRecord1This standard is issued under the fixed designation E 2171; the number immediately following the designation indicates the year oforiginal adoption or, in the case of
2、 revision, the year of last revision. A number in parentheses indicates the year of last reapproval. Asuperscript epsilon (e) indicates an editorial change since the last revision or reapproval.1. Scope1.1 This standard addresses the identification of data ele-ments from the EHR definitions in Guide
3、 E 1384 that haveordinal scale value sets and which can be further defined tohave scale-free measurement properties. It is applicable to datarecorded for the Electronic Health Record and its papercounterparts. It is also applicable to abstracted data from thepatient record that originates from these
4、 same data elements. Itis applicable to identifying the location within the EHR wherethe observed measurements shall be stored and what is themeaning of the stored data. It does not address either the usesor the interpretations of the stored measurements.2. Referenced Documents2.1 ASTM Standards:E 1
5、77 Practice for Use of the Terms Precision and Bias inASTM Test Methods2E 456 Terminology Related to Quality and Statistics2E 691 Practice for Conducting an Interlaboratory Study toDetermine the Precision of a Test Method2E 1169 Guide for Conducting Ruggedness Tests2E 1384 Guide for Content and Stru
6、cture of the Computer-Based Patient Record33. Terminology3.1 DefinitionsFull definitions and discussion of Scale-Free Measurement Terms are given in Annex A1.3.2 Definitions of Terms Specific to This Standard:3.2.1 adaptive measurementadvantage of measurement toaccount for missing data.3.2.2 additiv
7、ityrating scale adherence to associativity andcommutability.3.2.3 bias analysisinvestigation of considerations relativeto subject or area of performance.3.2.4 calibrationprocess of establishing additivity andreproducability of a data set.3.2.5 concatenationprocess of measurement uses enumer-ated phy
8、sical unit quantities equal to the magnitude of themeasured item.3.2.6 constructname of the conceptual domain measured.3.2.7 convergenceclosing of the differences in sequentialmeasure estimates.3.2.8 countingbasic activity upon which measurement isbased and utilizes enumeration.3.2.9 dataobservation
9、 made in such a way that they leadto generalization.3.2.10 data quality/ statistical consistency/ model fitestablishment of whether the measuring instrument is affectedby the object of measurement.3.2.11 determinismmeasurement model that requirescounts to be sufficient for reproducing the pattern of
10、 theresponses over the length of the instrument.3.2.12 dimensionalityproperty of having multiple compo-nents of a measured value.3.2.13 equality/cocalibrationprocess of ensuring that dif-ferent instruments measure the same property.3.2.14 erroruncertainty of measured properties.3.2.15 estimation alg
11、orithmsmathematical specificationof an observational framework.3.2.16 incommensurable/commensurablemeasure valueof the same quantity does/does not depend upon rating/responses of the rating construct and does not/does remainconstant.3.2.17 instrumentsensing device having a defined scale.3.2.18 intra
12、 and inter-laboratory testingvariability testingusing the same setting/measure/operator as opposed to differentsetting/measure/operators.3.2.19 item response/latent trait theoryanalytic modelsthat forego prescriptive parameter separation, sufficiency andscale and sample free data standards for addit
13、ional descriptiveparameters.3.2.20 items/item-bankpart of survey statements/testquestions for adaptive administration.3.2.21 levels of measurementnature of scale of measure-ment.1This practice is under the jurisdiction of ASTM Committee E31 on HealthcareInformatics and is the direct responsibility o
14、f Subcommittee E31.25 on HealthcareData Management, Security, Confidentiality, and Privacy.Current edition approved Dec. 10, 2002. Published February 2003.2Annual Book of ASTM Standards, Vol 14.02.3Annual Book of ASTM Standards, Vol 14.01.1Copyright ASTM International, 100 Barr Harbor Drive, PO Box
15、C700, West Conshohocken, PA 19428-2959, United States.3.2.22 logitscale unit using logarithms of odds ratios.(P/1P).3.2.23 mathematical entitiesconcepts that can be taughtor learned through what is already known.3.2.24 measurementdetermining in units the value of aproperty in a scale having magnitud
16、e (that is, ratio or differ-ence).3.2.25 metaphor in measurementsuspension of disbeliefof some areas or properties in the name of estimating magni-tude.3.2.26 metricmeasure of a property in defined units.3.2.27 missing datause of uncalibrated data in instru-ments with varying numbers of items.3.2.28
17、 multi-faceted measurementuse of measurementmodels that have more than two basic parameters.3.2.29 ordinal dataone scale for measurement.3.2.30 populationuniverse of elements relevant to mea-surement of a particular construct.3.2.31 probabilistic conjoint measurementframework fordemonstrating data q
18、uality, statistical consistency, and modelfit of non-deterministic measures with a stable order of facets.3.2.32 quantificationcocalibration of different constructswith respect to the same property (variable) in a commonmetric.3.2.33 Rasch analysis measurement and modelsanalyticmodel specifying the
19、observational framework and data qualitymeasures for quantification.3.2.34 raw scoresum of ratings or count of direct re-sponses in a given measurement event.3.2.35 reliabilityratio of variation to error or signal tonoise.3.2.36 repeatabilityvariability of measurements in asingle setting by a single
20、 operator using the same measuringinstrument.3.2.37 reproducibilityvariability of measurements in dif-ferent settings.3.2.38 root mean square errormathematical algorithm fordetermining the variation due to error of the estimates.3.2.39 samplesubset of measured population.3.2.40 sample sizemagnitude
21、of the measured population.3.2.41 scale-free/scale-dependentmeasures not affectedby the instrument employed as opposed to measures that are soaffected.3.2.42 separability theorem/parameter separationabilityof measures to be independent of the instrument selected andability of the instruments item ca
22、librations to be independentof the sample measured.3.2.43 softwarepackages of machine code used for dataanalysis.3.2.44 specific objectivitydata satisfying the separabilitytheorem.3.2.45 standardizedcommon conventions for instruments,reference measurement material, scales and units of measurefor a m
23、easurement process.3.2.46 suffciencystatistics that extract all available infor-mation from the data.3.2.47 targetinglack of floor an/or ceiling effects in mea-surement.3.2.48 transparencyability to “look through” raw scoresto the composite ratings producing that score (see also suff-ciency).3.2.49
24、unit of measurementcommon conventions for theappropriate smallest basic measures for a given construct.3.2.50 validity/construct/contentboth content and con-struct must make sound theoretical sense to be consideredvalid.3.2.51 variableattribute of the property being measured.4. Significance and Use4
25、.1 The simplicity and practicality of Raschs probabilisticscale-free measurement models have brought within reachuniversal metrics for educational and psychological tests, andfor rating scale-based instruments in general. There are at least3 implications to the application of Raschs models to thehea
26、lth-related calibration of universal metrics for each of thevariables relevant to the Electronic Health Record (EHR) thatare typically measured using rating scale instruments.4.1.1 First, establishing a single metric standard with adefined range and unit will arrest the burgeoning proliferationof ne
27、w scale-dependent metrics.4.1.2 Second, the communication of the information per-taining to patient status represented by these measures (physi-cal, cognitive, and psychosocial health status, quality of life,satisfaction with services, etc.) will be simplified.4.1.3 Third, common standards of data q
28、uality will be usedto evaluate and improve instrument performance. The vastmajority of test and survey data quality is currently almostcompletely unknown, and when quality is evaluated, it is viamany different methods that are often insufficient to the task,misapplied, misinterpreted, or even contra
29、dictory in their aims.4.1.4 Fourth, currently unavailable economic benefits willaccrue from the implementation of measurement methodsbased on quality-assessed data and widely accepted referencestandard metrics. The potential magnitude of these benefits canbe seen in an assessment of 12 different met
30、rological improve-ment studies conducted by the National Science and Technol-ogy Council (Subcommittee on Research, 1996). The averagereturn on investment associated with these twelve studies was147 %. Is there any reason to suppose that similar instrumentimprovement efforts in the psychosocial scie
31、nces will result inmarkedly lower returns?4.2 Until now, it has been assumed that the Guide E 1384would necessarily have to stipulate fields for the EHR thatwould contain summary scores from commonly used func-tional assessment, health status, quality of life, and satisfactioninstruments. This is be
32、cause standards for rating scale instru-ments to date have been entirely content-based. Those whohave sought “gold” or criterion standards that would commanduniversal respect and relevance have been stymied by theimpossibility of identifying content (survey questions andrating categories) capable of
33、 satisfying all users needs. Com-munication of patient statistics between managers and clini-cians, or payors and providers, may require one kind ofinformation; between providers and referral sources, otherkinds; between providers and accreditors, yet another; amongclinicians themselves, still anoth
34、er; and even more kinds ofinformation may be required for research applications.E21710224.2.1 For instance, payors may want to know outcomeinformation that tells them what percentage of patients dis-charged can function independently at home. A hospital man-ager, referral source, or accreditor might
35、 want to know moredetail, such as percentages of patients discharged who candress, bathe, walk, and eat independently. Clinicians will wantto know still more detail about amounts of independence, suchas whether there are safety issues, needs for assistive devices,or specific areas in which functiona
36、lity could be improved.Researchers may seek even more detail yet, as they evaluatedifferences in outcomes across treatment programs, diagnosticgroups, facilities, levels of care, etc.4.2.1.1 Members of each of these groups have, at sometime, felt that their particular information needs have not been
37、met by the tools designed and developed by members ofanother group. Despite the fact that the information providedby these different tools appears in many different forms and atdifferent levels of detail, to the extent that they can be shownto measure the same thing, they can do so in the same metri
38、c.This is the primary result of the introduction of Raschsprobabilistic scale-free measurement models. The differentpurposes guiding the design of the instruments will stillcontinue to impact the two fundamental statistics associatedwith every measure: the error and model fit. More general, andalso
39、less well-designed instruments, will measure with moreerror than those that make more detailed and consistentdistinctions. Data consistency is the key to scale-free measure-ment.4.3 The remainder of this document (1) identifies, in Section5, the fields in the current Guide E 1384 targeted for change
40、from a scale-dependent to a scale-free measurement orienta-tion; (2) lists referenced ASTM documents; (3) defines scale-free measurement terms, often contrasting them with theirscale-dependent counterparts; (4) addresses the significanceand use of scale-free measures in the context of the EHR; (5)li
41、sts, inAnnexA2, scientific publications documenting relevantinstrument calibrations; (6) briefly presents some basic opera-tional considerations; (7) lists minimum and comprehensivearrays of EHR database fields; and (8) lists, in Annex A3, thereferences made in presentation of the measurement theory
42、,estimation methods, etc.4.4 Publications of calibration studies referencing this prac-tice and the associated standard practice should require:4.4.1 The use of measures, not scores, in all capture of datafrom the EHR for statistical comparisons;4.4.2 The reporting of both the traditional reliabilit
43、y statis-tics (Cronbachs alpha or the KR20) and the additive, linearseparation statistics (Wright 4.4.3 Aqualitative elaboration of the variable defined by theorder of the survey questions or test items on the measurementcontinuum, preferably in association with a figure displayingthe variable;4.4.4
44、 Reporting of means and standard deviations for eachof the three essential measurement statistics, the measure, theerror, and the model fit;4.4.5 Statement of the full text of at least a significantsample of the questions included on the instrument;4.4.6 Specification of the mathematical model emplo
45、yed,with a justification for its use;4.4.7 Specification of the error estimation and model fitestimation algorithms employed, with mathematical details andjustification provided when they differ from those routinelyused;4.4.8 Evaluation of overall model fit, elaborated in a reporton the details of o
46、ne or more of the least and most consistentresponse patterns observed;4.4.9 Graphical comparison of at least two calibrations ofnew instruments from different samples of the same populationto establish the invariance of the item calibration order acrosssamples;4.4.10 Graphical comparison of measures
47、 produced by atleast two subsets of items on new instruments to establish theinvariance of the person measure order across scales (collec-tions of items);4.4.11 Graphical comparison of new instrument calibrationswith the calibrations produced by other instruments intended tomeasure the same variable
48、 in the same population, to establishthe potential for sample-free equating of the instruments andestablishment of reference standards;4.4.12 At least a useable prototype of the instrument em-ployed, with the worksheet laid out to produce informativequantitative measures (not summed scores) as soon
49、as it isfilled out; and4.4.13 Graphical presentation of the treatment and controlgroups measurement distributions, for the purpose of facili-tating a substantive interpretations of differences significance.5. Applicable Data Elements5.1 The data elements in Guide E 1384 which are affectedby the suggestions for measurement standardization made hereinclude the following:PHYSICAL EXAM SEGMENT09001.16 Patient Health Status Measure Name09001.17. Patient Health Status Measure Total Value09001.19. Patient Health Status Measure Elemen
