HL 7 -UK 2003 Conference Healthcare Interoperability Past

HL 7 -UK 2003 Conference Healthcare Interoperability Past Efforts-Present Benefits-Future Prospects Clement J. Mc. Donald, M. D. December 11, 2003 1

Work of many Regenstrief faculty • • Mike Barnes, M. D. Paul Dexter, M. D. Burke Mamlin, M. D. Mick Murray, Pharm. D. Marc Overhage, M. D. , Ph. D. Susan Perkins, Ph. D. Gunther Schadow, M. D. Bill Tierney, M. D. 2

Where we were in early 80’s • Building our own systems to collect data – Lab – Pharmacy – Scheduling system • Horrible batch and cobbled interfaces to collect data from other institutions systems 3

What to do? • Impossible except for nutty people • We needed an easier way • Standards… 4

Plea for standards help • 1981 • 1983 (Jan. ) • 1983 (Nov. ) • 1984 (Nov. ) 1 st rejection of grocer’s paper Grocer paper published 1 Plea for standards at fall SCAMC meeting 2 A discussion of the draft proposal for data exchange standards for clinical laboratory results. 3 (1) Mc. Donald CJ, Park BH, Blevins L. Grocers, physicians, and electronic data processing. AMA Cont Med Ed Newsletter 1983; 12: 5‑ 8. (2) Mc. Donald CJ. Standards for the transmission of diagnostic results from laboratory computers to office practice computers ‑‑ an initiative. Proc Annu Symp Comput Appl Med Care 1983; 123‑ 24. (3) Mc. Donald CJ, Wiederhold G, Simborg DW. A discussion of the draft proposal for data exchange standards for clinical laboratory results. Proc Annu Symp Comput Appl Med Care 1984; 406‑ 13. 5

1984 SCAMC Fall Symposium • Panel Members: – Clement Mc. Donald, M. D. – Gio Wiederhold, Ph. D. – Donald W. Simborg, M. D. – Ed Hammond, Ph. D. – Fredrick R. Jelovsek, M. D. – Ken Schneider, M. D. 6

1984 Proposal- Familiar? FIGURE 1 Patient 1 (general information about patient) Battery 1 (information about first battery reported/requested) Result 1 (information about the first result of battery 1) Result 2 (information about the second result of battery 1) Result n (information about the last result of battery 1) Battery 2 (information about battery 2) Result 1 (information about the first result of battery 2) Result 2 (information about the second result of battery 2) Result n (information about the last result of battery 2) Battery n (information about the last battery for the first patient) Result 1 (information about the first result of the last battery) Patient 2 (all of the structure repeats) Patient N 7

Pleas answered • 1988 • 1989 • 1990 ASTM 1238 -88 4 HL 7 published version 2. 0 5 SCAMC – Progress, Promises and the Conductors Wand 6 (4) Standard Specification for Transferring Clinical Laboratory Data Messages Between Independent Computer Systems. Annual Book of ASTM Standards; Philadelphia, PA, 1988. (5) Health Level Seven, Version 2. 0. An application protocol for electronic data exchange in healthcare environments. Ann Arbor, Michigan: Health Level Seven, 1989. (6) Mc. Donald CJ. Standards for the electronic transfer of clinical data: progress, promises, and the conductor's wand. Proc Ann Symp Comput Appl Med Care. 1990; 14: 9 -14. 8

Music & words 1990 • MIDI to orchestra what HL 7 is to EMR 9

1995 -2003 • Messages fairly good • Could get HL 7 result messages working in 1 -3 weeks • Vendors knew the standard • Most “BAD” messages due to conscious placement of right info in “wrong” field 13

Example bad messages • Value, units, normal ranges, flags, and performance site put ALL in OBX-5 • Value and units both jammed into OBX-5 • OBX-5 says “see comment” - everything jammed into following NTE • Whole report (many test results) jammed into single OBX-5 14

Life good but not perfect • Codes for the same question (observation) different every where • Need most-Standard question (observation) codes – If answer is text- all’s well – If answer is numeric - all’s well – If answer is a code (but come with text representation) – not perfect, OK 15

LOINC standard for codes • First meeting – 1994 – Select group of interested parties • Mayo medical laboratories • CORNING Met. Path (since merged with SHF to become Quest) • University of Washington • Indiana University/Regenstrief Institute • University of Utah (Intermountain Healthcare) • Veterans Administration • University of Ghent Belgium 16

LOINC Actual • • Identifies the observation (OBX 3 and OBR 4) The question not the answer First file distributed April 1995 (6000 terms) Now 34, 000 terms – – Battery structures Hierarchy Synonym Free Mapping and browsing program • Free for all uses – url: http: //www. loinc. org/ Mc. Donald CJ, Huff SM, Suico JG, Hill G, Leavelle D, Aller R, et al. LOINC, A Universal Standard For Identifying Laboratory Observations: 17 A 5 -Year Update. Clin Chem 49: 4; 624 -633

Medical Records Without Standards 18 [G]

The present medical records with standards 19

Wishard the oldest and the broadest coverage • Data beginning in 1972 • Since early 80’s – – – All laboratory data All pharmacy (in patient and out) All dictation All diagnostic studies ( radiology, cardiac, etc) Al encounters ( with DX's and procedures etc) Coded clinical information from largest clinics ( OB and Medicine and med subspecialty • Since late 80’s – All- physician entered orders • Recently – Inpatient vitals, MD note entry, etc. 20

Standards that support EMRs Display: courtesy of HTML and IP standards 21

22

23

24

25

VISIT INFORMATION From registration system Courtesy of HL 7 VITAL SIGNS DATA FROM AUTOMATIC BP- PULSE – TEMPERATURE courtesy of HL 7 26

27

Chest x-ray 12 bit, 400 x 400 resolution 28

Chest x-ray 12 bit, x 1247 resolution 1024 29

Chest x-ray 12 bit, 2048 x 2494 (full) resolution 30

Radiology & images courtesy DICOM, JPEG, and JPEG 2000 31

Trade offs between JPEG and JPEG 2000 • JPEG decodes much faster (5 X) • JPEG Compresses “small” images (e. g. CT and MRI as well as JPEG 200 • JPEG 2000 compresses large films (chest x -rays) (2000 x 2700) 12 bit depth by 5 x better • Use JPEG for “small” CT and MRI images and JPEG 2000 for large plain films. 32

33

34

Echo reports courtesy of HL 7 35

36

37

• Echo images Courtesy of Motion JPEG 38

39

40

Scanned images • Courtesy of TIFF 41

Telemedicine Courtesy of: • H 232 • IPSEC • MPEG 1 42

43

IT is ALL in the Standards !HL 7 (most) • DICOM (to capture uncompressed images from PACS • NCDP (Prescriptions from outpatient pharmacy • JPEG , JPEG 2000 for storing clinical images in Clinical Repository • IP (Everywhere) • JPEG, MPEG 1 • Code Standards Currently Used: – – CPT ICD LOINC CDO-3 44

Guard Rails • Improve preventive care from 30% to 400% over control state (1) • Improve drug dosing and reduce errors in drug prescribing • Increase required corollary orders 5%-64% (e. g. test Vancomycin levels when giving Vancomycin) 1. Mc. Donald CJ, Hui SL, Smith DM, Tierney WM, Cohen SJ, Weinberger M. Reminders to physicians from an introspective computer medical record. Annals of Internal Medicine 1984; 100: 130‑ 138. 2. Overhage JM, Tierney WM, Zhou XH, Mc. Donald CJ. A Randomized Trial of “Corollary Orders” to Prevent Errors of Omission. JAMIA 1997; 4: 364 -375. 45

Percent Response Preventive Reminders Mc. Donald, et al. Reminders to Physicians from an Introspective Computer Medical Record. Annals of Internal Medicine, 1984; 100: 130 -138. 46 [RG]

Physician Order entry- (POE) 47

Computer based POE Gopher order entry • Were first to succeed with order entry – 20 million orders at Wishard since 1986 • POE provides better options for delivering influencing patient care • Even non fixed (not patient specific) content can have an effect 48

Gopher physician Users 49

Cost Savings CHF example 50

51

Preventive POE inpatient POE reminders 2001 (Dexter et al. ) Dexter PR, Perkins S, Overhage JM et al. A Computerized Reminder System to Increase the Use of Preventive Care for Hospitalized Patients. N Engl J Med 2001; 345: 965 -970. 52

City Wide RMRS Functions • CARE deliver clinical information when needed for emergency care (NLM supported study) • Public health seek out reportable conditions from laboratory data evidence, equivalence duplicate cases, organize and deliver to public health • Research molecular biologic epidemiology through Pathology samples ( National 53 Cancer institute

Whole city • • About 1. 3 million people Five hospital systems and their clinical facilities Fourteen separate hospital facilities All four homeless care clinics 450 K emergency room visits per year State and county public health Three uses of community data repository – Direct patient care – Public health – Research 54

County Health Department Quest State Board of Health Lab. Corp 55

INPC all 5 city institutions # HL 7 message streams 149 # NCPDP streams #DICOM streams 1 # rows/observations stored 2 (Agfa PACs, GE PACs) 436 million # text reports stored 11, million # CPOE orders stored 22 million (est) # radiology images stored Studies= 1 million images=50 million (est) 480 thousand # EKG tracings stored 56

Clinical use • At Wishard and Clarian total of 2. 5 million accesses per month by more than 4000 different clinical users • Emergency Department physicians can look at records from all institution as one merged record when patients present at the Emergency department 58

Monthly volumes # HL 7 messages per month 5, 491, 882 # OBX results per month 5, 803, 541 + ___ 59

Uses in public health (all based on HL 7 messages) • Automatically scan HL 7 lab messages for reportable disease – – Look at test code Look at normal flag * Look at result Send those that =>reportable condition • Faster by 8 days, more complete by 2 fold • Immunization reporting • Tumor registry 60

Reportable disease example 61

Outcomes • Reliable – Real time delivery – 100%received (for participants) • Reporting completeness (capture/recapture) – Found nearly twice as many cases from the index hospitals as found by the usual methods • Reporting timeliness (versus result date) – 8. 4± 15. 4 days faster than HD – case finding – 1. 4± 2. 0 days faster than hospital case finding 62

INPC Use in research • Help in writing and planning grant applications – How many patients of a given kind are available in the database – What are their characteristics • Help in recruiting patients to clinical trials (see next slide) 63

Use in research – grist for epidemiologic studies • Mahon BE, Rosenman MB, Kleiman MB. • Maternal and infant use of erythromycin and other macrolide antibiotics as risk Factors for infantile hypertrophic pyloric stenosis. J Pediatr. 2001 Sep; 139(3): 380 -4. 65

Research – Shared Pathology Informatics Network (SPIN) • Pathology departments store tissue as paraffin blocks for 10 -15 years • Protein, DNA and even MRNA can be identified in these specimens • Clinical repository can identify cancer patients with specific characteristics – E. g. short and long survivors of small cell lung cancer • Can get from the record to pathology report and the tissue blocks to answer biologic questions 66

Research SPIN • Auto coding of pathology reports • Free text searching of pathology reports • Peer-to-peer links between many nodes – Boston – Pittsburgh – UCLA – All hospitals in Indianapolis 67

Future • Don’t think we will see monoliths every where. Rather, message coupled systems • For every system pushed into monolith 1 -2 new ones pop up at our institution. – – – Prosolve cardiac echo Witt cath lab system _____Endoscopy system Otter transplant system ______ ICU system 68

Future • Strong common/replicated master files – Patient master ( Easy in GB) – Provider master ( may be described in dec 03) – Orderables master – Result master – Geographic location master – Printer locations 69

Evidence based coding • Selected structured data collection • Hopefully – evidence based • We need to find out which variables are predictive or decisive – Stiels Ottawa Ankle Rule • Narrative continues to reign 70

Leap in amount of non-structured non text data • Digital voice – Speech understanding ( has to be successful some time. • Photos • Video – record the whole encounter (Charlie Safran) • Scanned documents (To get totally electronic record ) 71

Predictions • Preempt disease rather than perfect desperation care – Empty the hospitals – A major disease, e. g diabetes will be cured • Community-based clinical systems 72

Wireless- everybody, everywhere to everyplace - (WEEE) • Clinical data and info services as available as the air – Real time alerts and updates to Provider carried systems – Hi-grade video conferencing between provider and patient. • More care in lower acuity settings, Home health care replaces other forms if more care shifted to lower acuity settings – Wireless patient monitoring direct to provider systems – Image reading (radiology, pathology) from afar 73

Evidence based coding • Hopefully – evidence based data collection • First determine which variables are predictive or decisive in guidelines • Stiels Ottawa Ankle Rule • Miles to go before mostly coding 74

Leap in amount of non-structural data • Digital voice – Speech • Photo and video – record the whole encounter (Charlie Safran) • Scanned data 75

Some Observations 76

The Keyboard Lives • For order entry and note writing is the fastest entry mechanism available – Faster than mousing – Faster than hand writing recognition – Faster than thumb typing (but it is faster than hand writing recognition • All new MDs can type. 77

78

Paper lives • Input: you can’t avoid scanning documents if you want a complete record • Output: 1 trillion laser printed pages n 2001, printer vendors predict 6 trillion pages in 2006 79

Paper lives • Paper generated by computer may be with us forever • Paper is Light weight and portable • Extra-ordinary resolution – Best UXGA 1600 x 1200 – Page (@ 2400 dpi) 18, 000 x 24, 000 • Printers getting cheaper and better all of the time • Easy to annotate I 80

The prefect PDA or “carried computer” not here yet • Screen too small • Lacks a keyboard – the king of input • Most clinical applications need wireless connection • But battery life too short (or box too big when battery adequate ) 81

Our wireless invention: The Popcorn Man’s “PDA” 82 [RRInput]

Radio Link Tablet Computers 83

Tomorrows Cyborg wireless 84

Ski mask wearable computer 85

Cool wearable computer 86