Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Overview'of'EHR$Systems$and'Clinical$Data$
Roberto'A.'Rocha,'MD,'PhD,'FACMI'Clinical'Informa9cs'Director,'
Partners'eCare,'Partners'Healthcare'System'
Assistant'Professor'of'Medicine'Division'of'General'Internal'Medicine'and'Primary'Care,'Department'of'Medicine,'Brigham'and'Women’s'Hospital,'Harvard'Medical'School'
Big%Data%in%Healthcare:%The%Impact%on%Healthcare%Quality,%Cost%and%Access%in%China%=%April%18=19,%2016%=%%
Overview$
• Background'– Clinical'Systems,'EHR'Systems,'Data'Capture'
• Types'of'clinical'data'– Unstructured,'Structured,'Coded''
• Structured'and'coded'clinical'data'– Consistency,'level'of'detail,'seman9cs,'standards'
• Conclusions'– Current'efforts,'methodology,'par9cipa9on'
Clinical$System$
• �…'an'automated'system'with'a'long'term'
database'containing'clinical'informa9on'used'
for'pa9ent'care.�'
– Bruce'Blum,'1986'
• Support'(automa9on)'for'one'or'mul9ple'
clinical%(pa9ent'care)'func9ons'• Electronic*Health/Medical*Record*system'is'
an'integrated'suite%of'clinical'systems'
Separate$“clinical”$systems$
Clinical*Research*Systems*
Medical*Record*Systems*
EHRs/EMRs:*Electronic'Health/
Medical'Record'Systems'
PHRs:'Pa9ent'Health'
Records'
EDCs:*Electronic'Data'Capture'
Systems'
eCRFs:*Electronic'Case'
Report'Forms'
EHR$systems$
• Electronic'Health'Record'(EHR)'– “electronic'version'of'a'pa<ents*medical*history,'…'maintained'by'the'
provider'over'9me,'…'clinical'data*relevant*to'that'persons*care*…'including'demographics,'progress'notes,'problems,'medica<ons,'vital*signs,'past*medical*history,'immuniza<ons,'laboratory*…'radiology”'
– (hZps://www.cms.gov/medicare/e[health/ehealthrecords/index.html)'
• US'market'
– Commercial'EHRs'dominate;'small'number'of'vendors'
– Designed'for'data'storage'&'communica9on:'human*users*
– High'tolerance*for'incomplete,'incorrect,'ambiguous'data'
– Limited*capability'for'computerCassisted*decision*making*
– Emphasis'on'data'sharing*(government'incen9ves)*
Outpatient$EHR$@$Partners$
Documenta<on,*ordering,*results*review,*messaging,*etc.*
Historical$perspective:$1991$(1/2)$
• Percent'of'pa9ent's'paper'chart'contained'in'the'EHR?'– “Using'the'number'of'pages'in'the'record'as'the'criteria,'the'answer'is'
about'26%*overall,'but'between'35%'and'40%'for'pa9ents'in'nursing'divisions'where'computerized'nurse'char9ng'is'used.”'
Kuperman'GL,'Maack'BB,'Bauer'K,'Gardner'RM.'The'impact'of'the'HELP'computer'system'on'
the'LDS'Hospital'paper'medical'record.'Top%Health%Rec%Manage.'1991'Aug;12(1):1[9.'
Historical$perspective:$1991$(2/2)$(C)'Computerized:'EHR'na9ve'data'
(F)'Form:'pre[printed'material'
(H)'HandwriKen:'handwriZen'notes'(T)'Typed:'typed'notes'
(O)'Other'
• 130'different'categories*of*data*iden9fied'
• Notes*alone'represent'over'26%'of'the'record'
Kuperman'GL,'Maack'BB,'Bauer'K,'Gardner'RM.'The'impact'of'the'HELP'computer'system'on'
the'LDS'Hospital'paper'medical'record.'Top%Health%Rec%Manage.'1991'Aug;12(1):1[9.'
Data$capture$methods$
• typing*(keyboard)'• forms*and'flowsheets*(with'or'without'voice'recogni9on)'
• text'macros*and'pre[defined'note'templates'
• dicta<on*(with'or'without'voice'recogni9on)'• digital'handwri9ng'• document'scanning*(with'or'without'OCR)'
• data'extrac9on'(natural'language'processing)'• automated'data'acquisi9on'(biomedical'devices)'
TYPES$OF$CLINICAL$DATA$
Clinical$data$
• Highly'complex'
– Diverse'types'of'data'! structured'and'unstructured,'images,'sounds,'etc.'
– Dynamic'(changing)'nature'
! flexible'and'extensible'underlying'models'
• Large'quan99es'– High'performance'database'environment'
! response'9me'is'the'cri9cal'factor'
• Confiden9ality'and'privacy'(security)'
Types$of$clinical$data$
• Unstructured'– “Mr.'Jones'has'Appendici<s”'– No'structure'or'codes'
• Structured'– Diagnosis:'“Appendici<s”'– Ques9on'is'defined'(coded)'but'answer'(value)'is'free[text'
• Structured'&'Coded'– Diagnosis:'K35*(Appendici<s)'– Ques9on'&'answer'are'defined'(coded)'
Unstructured$(narrative)$data$
• Significant'por9on'of'the'medical'record'is'
available'as'narra9ve'data'(70[95%)'! medical'history,'physical'exam,'progress'notes,'discharge'summary,'radiology'reports,'opera9ve'notes'
– advantages:'widespread,'comprehensive,'
convenient,'expressive,'natural*
– disadvantages:'ambiguous,'complex,'different'
styles,'redundant,'embedded'errors,'loose'
structure'
Structured$&$Coded$data$
• Clinical'and'epidemiological'research'
• Clinical'event'monitoring'(decision'support)'
• Audi9ng'(clinical'and/or'administra9ve)'
• Quality'control'(computerized)'
• Indexing'and'retrieval'(images,'sounds,'etc.)'
• Portability'(language'independence)'• Text'genera9on'• Advanced'analy9cs'
STRUCTURED$AND$CODED$CLINICAL$DATA$
Standard$data$definitions$
LOINC'
8339[4:' Body'weight^at'birth'
' Mass;'Pt;'^Pa9ent;'Qn;'Measured'
SNOMED'CT'(or'UCUM)'
258681007:'Units'of'mass'(SI)'
SNOMED'CT'
258682000:'gram,'g'
Data*Element'(aZribute):'numeric'measurement'
with'unit'
Topic*
Value*set*
Value'(concept)'
Standard*data*defini<ons:*data'elements'+'data'values'
Birth*Weight:*<number><units>*
Coded$data:$variation$
AZribute' Value'
Pain* Severe*
AZribute' Value'
Severe*Pain* Yes'
AZribute' Value'
Severe*Pain* 02[01[2001'
AZribute' Value'
Sys'BP' 180*mmHg*
AZribute' Value'
Sys'BP' Elevated*
AZribute' Value'
Sys'BP' Abnormal*
AZribute' Value'
Problem' Severe*Pain* AZribute' Value'
Finding' Elevated*Sys*BP*
Clinical$data$representation$(1/3)$
• Not'sufficient'to'catalog*all'possible'combina9ons'of'
data*elements*and'data'values**– Cataloging'is'the'state'of'the'prac9ce'for'US'EHR'systems'
• Essen9al'to'combine'data'points'and'obtain'
meaningful'clinical'informa9on:'informa<on*models'– Focus'on'high*priority*clinical*domains*given'the'significant'effort'to'develop'and'implement'models'
– Leverage'and'contribute'to'ongoing*efforts*to'validate'and'share'clinical'informa9cs'models'
! HL7'Clinical'Informa9on'Modeling'Ini9a9ve'(CIMI)''
Coded$data:$information$model$
• Acute*Pain*model:'1. Pain'Quality'(Character)'2. Pain'Primary'Loca9on'
3. Pain'Onset'4. Speed'of'Pain'Onset'5. Pain'Periodicity'6. Pain'Temporal'PaZern'
7. Pain'Allevia9ng'Factors'8. Pain'Aggrava9ng'Factors'9. Pain'Dura9on'(Hours,'Minutes)'
10. Associated'Signs'and'Symptoms'
11. Pa9ent'Severity'Score'12. Pain'Course'13. Rela9ve'Temporal'Context'
14. Pa9ent'Stated'Goal''
Gesner'E,'Collins'SA,'Rocha'R.'Pain'Documenta9on:'Valida9on'of'a'
Reference'Model.'Stud%Health%Technol%Inform.'2015;216:805[9.'
Clinical$data$representation$(2/3)$
• Not'sufficient'to'develop'informa<on*models*for'high'priority'data'domains'
– Models'have'to'be'shared*to'enable'data*exchange*
• Need'to'also'capture'detailed'provenance*to'understand'how'and'when'data'was'acquired,'
and'also'who'provided'the'data'
Coded$data:$provenance$
Oniki'TA,'Zhuo'N,'Beebe'CE,'Liu'H,'Coyle'JF,'Parker'CG,'Solbrig'HR,'
Marchant'K,'Kaggal'VC,'Chute'CG,'Huff'SM.'Clinical'element'models'in'the'
SHARPn'consor9um.'J%Am%Med%Inform%Assoc.'2016'Mar;23(2):248[56.'
Clinical$data$representation$(3/3)$
• Not'sufficient'to'develop'informa9on'models'
with'detailed'provenance*– Missing'important'rela<onships*between'data'instances'
– Cri9cal'for'understanding'the'pa9ent'medical'
history'and'the'evolu9on'of'clinical'condi9ons'
• Need'to'represent'seman<c*and'temporal*linkages*between'models'(data'instances)'
Coded$data:$linkages$
Ontology*is*the*source*
Puleston'C,'Parsia'B,'Cunningham'J,'Rector'A.'Integra9ng'Object[Oriented'and'Ontological'Representa9ons:'A'Case'Study'in'Java'and'OWL'Export.'The%SemanJc%Web%=%ISWC%2008%(2008),'pp.'130[145.'
CONCLUSIONS$
Known$consequences$
Medical*Record*Systems*
Clinical*Research*Systems*
Manual'data'entry'
Data're[entry'
Missing'data'
Misinterpreta9ons'
Limited'contextual'details'
Disparate'data'defini9ons'
Laborious'data'extrac9on'
Overlapping'standards'
…'
Data$capture:$guiding$principles$
1. Be'clinically*per<nent,'pa<entCcentric,'and'represent'an'individual’s'life9me'
health'and'healthcare.'
2. Support'capture'of'high'quality'informa9on'that'is'accurate,'relevant,'confiden<al,'reliable,'valid,'complete,'and'secure.'
3. Be'efficient*and'usable'while'enhancing'the'healthcare'organiza9on’s'and'the'care'team’s'overall'efficiency,'effec9veness'and'produc9vity.'
4. Support'mul9ple'downstream*uses*as'a'byproduct'of'the'recording'of'care'delivery'including'quality'measurement,'performance'improvement,'popula9on'
health'care'delivery,'policymaking,'research,'educa9on,'and'reimbursement.'
5. Enable'joint*pa<entCprovider*decision*making,'team'collabora9on,'care'process'
management,'and'advanced'clinical'decision'support.'
6. Enable'collec9on'of'data'and'interpreta9on'of'informa9on'from'mul<ple*sources*as'appropriate'and'necessary,'including'nuanced'medical'discourse,'structured'items,'and'data'captured'in'other'systems'and'devices.'
7. Automa9on'of'data'capture'and'documenta9on'should'be'op9mized'whenever'
appropriate,'allowing'human'beings'to'focus'on'gathering'and'entering'data'that'
cannot*be*effec<vely*collected*by*automated*tools.'
Cusack'CM'et'al.'The'future'state'of'clinical'data'capture'and'documenta9on:'a'report'from'AMIA's'2011'Policy'Mee9ng.'J%Am%Med%Inform%Assoc.'2013'Jan'1;20(1):134[40.'
Methodology$@$Partners$
Collins'SA,'Gesner'E,'Morgan'S,'Mar'P,'Maviglia'S,'Colburn'D,'Tierney'D,'Rocha'R.'
A'Prac9cal'Approach'to'Governance'and'Op9miza9on'of'Structured'Data'Elements.'
Stud%Health%Technol%Inform.'2015;216:7[11.'
Example:$SHARPn$
• Strategic'Health'IT'Advanced'Research'Project'area'four'(SHARPn)'
– Open[source'tools'to'normalize*EHR'data'for'secondary'use'–'e.g.'high'throughput'phenotyping'
! Sources:'HL7'messages,'NLP'output,'and'C[CDA'documents'
– Standards[conformant'structured*and'coded*data*suitable'for'secondary'uses'
– 28'generic'models:'diagnosis,'procedure,'allergy,'findings,'
medica9ons,'laboratory'results,'vital'signs,'plus'linkages,'collec9on,'
and'detailed'provenance'
! Values'defined'using'standard'terminologies:'RxNorm,'HL7,'SNOMED'CT,'LOINC'
– Reliance'on'specialized'resources'that'are'not'commonly'available'
– Recommenda9ons:'training,'beZer'tools,'model'library'
Oniki'TA,'Zhuo'N,'Beebe'CE,'Liu'H,'Coyle'JF,'Parker'CG,'Solbrig'HR,'Marchant'K,'Kaggal'VC,'Chute'CG,'Huff'SM.'Clinical'element'models'in'the'SHARPn'consor9um.'J%Am%Med%Inform%Assoc.'2016'Mar;23(2):248[56.'
Opportunity$to$participate$
• The'Clinical*Informa<on*Modeling*Ini<a<ve*(CIMI)'is'
an'HL7'Work'Group'that'is'producing'detailed'clinical'informa9on'models'to'enable'interoperability'of'
health'care'informa9on'systems'
• CIMI'models'are'free'for'use'for'all'purposes'
• See'hZp://www.opencimi.org/'for'more'details'
Acknowledgements*
Perry'Mar'
Sarah'Collins'
Emily'Gesner'
Saverio'Maviglia'
Charles'Lagor'
Li'Zhou'
Beatriz'Rocha'
Other'members'of'the'Clinical*Informa<cs*Team*at'Partners'
Stanley'Huff'
Members'of'the'Modeling*and*Terminology*Teams*at'Intermountain'
Thank*you!*
Roberto A. Rocha, MD, [email protected]
http://scholar.harvard.edu/rarocha
This'work'by'Roberto'A.'Rocha'is'licensed'under'a'Crea9ve'Commons'AZribu9on[
NonCommercial[ShareAlike'4.0'Interna9onal'License'