Data Exchange Standards (Introduction to Medical Informatics) (http://www.cpmc.columbia.edu/edu/textbook) LAST REVIEWED: 15 October 1996 informatics vs. computer science focus why easier to link computers (purchased and build) one interface instead of many what form = messaging component (like schema or grammar) content = codes, vocabulary transaction model = query, update, ask encoding = low-level syntax (eg, ASN.1) low-level network stuff consensus standards volunteer-based all must approve or abstain; long process to override a "no" market forces (eg, vendor blocks standard) organizations (much competition among standards groups => duplication) ASTM: lab, Arden HL7: results, orders, ADT, some finance IEEE/Medix: same as HL7, will merge ACR/NEMA: images X12: insurance NCPDP: drugs ANSI: good housekeeping seal of approval HEALTH LEVEL SEVEN (HL7) want to communicate with purchased systems, homegrown systems, other institutions each pair of systems has its own interface (many-many) HL7 defines the exchange of medical data between computers ad hoc organization: institutions, vendors, consultants version 1.0 in 1987; version 2.1 in 1990 Rationale: there was a lack of coordination at all levels: 1. organization or structure of medical data eg, a patient has a name, age, gender, ... 2. vocabulary used to express concepts eg, "potassium" vs. "serum potassium" vs. "K" 4. what can you do with the data eg, query, update 5. high level protocol eg, solicited vs unsolicited data transfer 6. how to encode the information for transmission eg, what actually crosses the network 7. low-level communications protocols (bit level) therefore HL7 had to cover multiple levels: (supposed to be level "seven", but had to cover more) abstract messages vocabulary encoding rules lower-level protocol (eg, down to RS-232) abstract (medical) messages: admit, discharge, transfer (ADT) orders queries (some mixing of levels: ADT vs query not same level) observations finance vocabulary not fully defined; some parts encoding rules: ASCII character set (including for transmission) uses variable length fields defined by delimiters you send and receive a "message" ADT - transfer demographic, registration data QRY - ask for clinical data message is a collection of "segments" MSH - header segment defines sender, receiver, ... QRD - defines a query segment is a collection of "fields" each segment defines its fields HL7 is the most successful standard for medical data, by far communications between vendor systems use it to communicate with vendor systems also use it internally: e.g., outpatient EMR and database Problem: exact definition of fields is still settled by arbitration negotiate with vendor end up with several "versions" in one institution Problem: no consistent data schema example: same information in two places can be inconsistent may not permit multiple values may group things illogically example: alert messages some are working on a real schema to avoid inconsistencies success factors in standards eg, HL7 vs Medix (VHS vs Beta, microchannel vs AT bus) when to standardize easy early on, but not yet know what is needed late is too late to get agreement (market then decides over time) HISPP: "Healthcare Informatics Standards Planning Panel" = ANSI panel attempting cooperation rather than competition umbrella organization for ASTM, IEEE/MEDIX, ACR/NEMA corresponding EU organization = CEN TC251 ISSUES BEYOND COMMUNICATIONS matching patients two patients with the same name name changes misspell a patient's name patient ID (medical record number) mistakes Bayesian probability based on name, age, gender pattern matching algorithm for names try to account for maiden names, misspellings HL7 EXAMPLES [1] HL7 Response to a Blood Culture Query MSH!^~\&!resquery!cicsu9!bobsqry!cucis!19930330102416611!!OR F!19930330102416611!P!2.1!""! MSA!AA!19920720113142729000!RESULT LIST COMPLETED.!! QRD!19920720113142!R!I!0113142729!!!1!3131313!RES!*!! QRF!*!19900601000000!19930203000000!PDQRES2~*~*!78~56~54~PF~ 0001CM01921223144005~A~~*~*~*~*~* OBR!!CM01921223765005!19921227651300344321!2203^^L^2235^^L!N !!1992122318130000000!!!!!!!!!!!!!!!1992122819523961234!!!F! !!!!!!!!!! OBX!!TX!910^^L^315^^L!1^0!KLEBSIELLA PNEUMONIAE!!!!!!! OBX!!TX!1235^^L^1181^^L!2^1!>16 R!!!!!!! OBX!!TX!1187^^L^1181^^L!3^1!S S!!!!!!! OBX!!TX!1184^^L^1181^^L!4^1!16 R!!!!!!! OBX!!TX!1193^^L^1181^^L!5^1!>64 R!!!!!!! OBX!!TX!1223^^L^1181^^L!6^1!<1 S!!!!!!! [2] Possbile HL7 Responses to a Demographics Query What is the name, dob, sex and patient vist information for a given MRN? version 1: MSH!^~\&!pmquery!cicsu9!myappl!hostname!93041616494!!ADR!298 378!P!2.1!""! MSA!AA!MCI!PATIENT FOUND IN PM!! QRD!930416164944660!R!I!121212!!!!3131313!APN!!!T PID!!!3131313!!RUIZ^CARLA!!19240421!F!!!!!!!!!!!! PV1!!I!M6HN613101!!!!!!!ME!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 0319932137!!!!! version 2: MSH!^~\&!resquery!cicsu9!bobsqry!cucis!19930330094803603!!OR F!!P!2.1!""! MSA!AA!19920720113142729000!RESULT LIST COMPLETED.!! QRD!19920720113142!R!I!011314!!!1!3131313!RES!HNAME~HSEX~HBIRTHDT!! QRF!HPBASIC!000000000!0000000!GYDAPMP~*~*!95~95~95~*~X~~~*~* ~*~*~* OBR!!!!HPBASIC^^L!!!19930416164944!!!!!!!!!!!!!!!!!!F!!!!!!! !!!! OBX!!TX!HNAME^^L!!CARLA RUIZ!!!!!!! OBX!!TX!HSEX^^L!!F!!!!!!! OBX!!TX!HBIRTHDT^^L!!19240421000000!!!!!!! related reading: Clayton PD, Sideli RV, Sengupta S. Open architecture and intergrated information at Columbia-Presbyterian Medical Center. M.D. Computing 1992;9(5):297-303.