Transcription
AMIA 2018 – San Francisco, CAW14: Analyzing Large Drug Prescription Datasets - Principles, Tools and Examples(sponsored by Pharmacoinformatics Working Group)November 3, 2018Analyzing Large Drug PrescriptionDatasetsPrinciples, Tools and ExamplesOlivier Bodenreider, MD, PhD, NLMVojtech Huser, MD, PhD, NLMChristian Reich, MD, PhD, IQVIA
DisclosureOB, VH: No relationships with commercial interestThe views and opinions expressed do not necessarily state orreflect those of the U.S. Government, and they may not be used foradvertising or product endorsement purposes.CR: IQVIA has many clients in the pharmaceutical industry, whoseproducts are the subject of this Tutorial. However, all products areintended to only be mentioned for the purposes of this Tutorial andwithout any recommendation or judgment of their use.2
Learning ObjectivesTo become familiar with large prescription datasetsTo gain knowledge about tools available for analyzing prescriptiondatasetsTo gain knowledge about clinical data models, such as OMOP3
OverviewPart 1 – Resources and use cases (OB) Prescription datasets RxNorm and NLM drug APIs; drug classification systems Common use casesPart 2 – Drug data processing in practice (VH) Implementing use cases with RxMix and RPart 3 – Experience with OHDSI (CR) Clinical data models Handling international drugs4
AMIA 2018 – San Francisco, CAW14: Analyzing Large Drug Prescription Datasets - Principles, Tools and Examples(sponsored by Pharmacoinformatics Working Group)November 3, 2018Part 1 – Resources and use casesOlivier Bodenreider, MD, PhD, NLMVojtech Huser, MD, PhD, NLMChristian Reich, MD, PhD, IQVIA
Part 1 overviewTypes of drug entities Drugs Drug classesPrescription datasets Structure SourcesDrug data processing Mapping drugs to standards Aggregating drugs (by ingredient, by class)Resources for processing drug datasets RxNorm Drug classification systems NLM drug APIsCommon use cases Pharmaco-epidemiology Assess exposure to drugs by ingredient or class Identify potentially inappropriate medications6
Types of drug entitiesPart 1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics7
Types of drug entities (drugs I)Ingredient AzithromycinDrug form Ingredient dose form Azithromycin Oral TabletClinical drug (unit of prescription) Ingredient dose form strength Azithromycin 250 MG Oral TabletPack (packaging level) Collection of clinical drugs (unit of dispensation) Z-PAK({6 (Azithromycin 250 MG Oral Tablet [Zithromax]) })Collection of clinical drugs (bulk) Manufacturer: Apotex Corp.; pack size: 500 in 1 BOTTLE8
Types of drug entities (drugs II)Generic vs. Brand G: Azithromycin 250 MG Oral Tablet B: Zithromax 250 MG Oral TabletSingle vs. multi-ingredient drug S: Amoxicillin 250 MG Oral Capsule M: Amoxicillin 250 MG / Clavulanate 125 MG Oral TabletSystemic vs. topical drugs S: Azithromycin 250 MG Oral Tablet T: Azithromycin 10 MG/ML Ophthalmic SolutionBase vs. salt/ester (as basis of strength substance) B: Erythromycin 250 MG Oral Tablet S: Erythromycin Ethylsuccinate 400 MG Oral Tablet9
Types of drug entities (drug classes)AtorvastatinTherapeutic class Anticholesteremic AgentChemical structure n/aMechanism of action HMG-CoA Reductase InhibitorPhysiologic effect Decreased Cholesterol Synthesis10
Types of drug entities (drug classes)11
Prescription datasetsPart 1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics12
StructureTransactions captured at dispensation time (pharmacy) Pharmaceutical claims data sent to payersCommon elements Prescription IDPatient identifierDate (prescription)Product ID (NDC)Total quantity dispensedDays supplyCost dataDrug nameStrengthGeneric namePrescriber ID99999912345620161112000710156233030-LIPITORTAB 20MG20Atorvastatin789NDCs arenot uniqueidentifiersfor clinicaldrugsRequiredOptionalDrug namesare notstandardized13
Product Name from Medicaid ClaimsProduct NameCOUMADINCOUMADIN TABFrequency27013COUMADIN TABLET215JANTOVEN130JANTOVEN TABJANTOVEN TABLETWARFARINWARFARIN TABWARFARIN SODIUMWARFARIN SODIUM TAWARFARIN SODIUM TABWARFARIN SODIUM TABLWARFARIN SODIUM TABLET242621,0937638,71795516587Source: Dan Malone8,55514
Product Identification: NDCsNational Drug Codes Product identification system Three components Manufacturer Product PackagingIntroduced in 1972 by FDAOnly format permitted by NCPDPMandated by HIPAA regulations for drug transactionsSource: Dan Malone15
NDC Elements: 3 6
NDC FormsWarfarin Sodium 1 MG Oral TabletXXXX-XXXX-XX (4-4-2)0555-0831-02(Teva Pharmaceuticals USA, Inc.; 100 in 1 BOTTLE)XXXXX-XXX-XX (5-3-2)21695-672-30 00555083102 XXXXX 0XXX XX(Rebel Distributors Corp; 30 in 1 BOTTLE)XXXXX-XXXX-X (5-4-1)50090-1213-0 0XXXX XXXX XX 21695067230 XXXXX XXXX 0X(A-S Medication Solutions; 30 in 1 BOTTLE) 5009012130017
NDC Characteristics11 Digit code (leading zero for 4-4-2 format)Hyphens between segments are missing in claims transmission (Field 407 inNCPDP claim format)NDC codes set by the manufacturer/labelerHigh turnover compared to other drug IDsProduct codes are unique to manufacturer – not to the chemical entityPackage codes are unique to the manufacturer and product – there is nostandardization for packaging codesSource: Dan Malone18
Source of prescription datasetsSurescripts transactionsData from payers Medicaid Medicare Part DCommercial health analytics companies Truven (120M patients) PharMetrics Plus (100M patients) Ambulatory EMR (35M patients) Open Claims (250M patients)Reagan-Udall Foundation for the FDA IMEDS Research Lab (temporarily suspended)19
Drug data processingPart 1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics20
Mapping and aggregationMapping to standard resources (e.g., RxNorm) Standard names and codes Standard set of relations among drug entities Link to drug classification systems E.g., NDC clinical drugAggregation “roll up” to the appropriate level of granularity for analytics (use case-dependent) E.g., branded drug clinical drug ingredient drug classTemporal aggregation Aggregate individual prescriptions into longer spans (“drug eras” in OHDSI)21
Esomeprazole (283742)Esomeprazole (A02BC05)Esomeprazole 40 MG DelayedRelease Oral Capsule (606730)Esomeprazole 40 MG Delayed Release OralCapsule [Nexium] (606731)001865040310186-5040-31
Resources for processingdrug datasets – RxNormPart 1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics23
/Terminology integration system Structured Product Labels, First DataBank, Micromedex, Multum, MeSH, SNOMED CT, MED-RT,ATC, Scope Drug names and codes Drugs available on the U.S. marketDeveloper: National Library of MedicinePublicly available*Monthly updatesSize: 10k ingredients; 19k clinical drugsUses: e-prescription, information exchange, analytics
Normalization Lexical levelNameWARFARIN (COUMADIN) NA 1MG TABwarfarin 1 mg oral tabletWARFARIN NA 1MG RFARIN NA 1MG TAB,UD [VA Product]N0000161787NDFRTWARFARIN SODIUM 1 mg ORAL TABLET14198NDDFWARFARIN SODIUM 1 mg ORAL TABLET60429-784Warfarin Sodium 1 MG Oral Tablet104045MTHSPLMMXWARFARIN SODIUM 1 mg ORAL TABLET63629-4017MTHSPLWARFARIN SODIUM 1 mg ORAL TABLET [Warfarin Sodium]53808-0985MTHSPLWarfarin Sodium 1 MILLIGRAM In 1 TABLET ORAL TABLET15330-100MTHSPL281572MTHFDAWARFARIN SODIUM 1.09 MG ORAL TABLETWarfarin Sodium 1mg Oral tabletWarfarin sodium 1mg tablet (product)933319733000GSSNOMEDCT USWarfarin Sodium Tab 1 MG6749MDDBWarfarin Sodium, 1 mg oral tablet3617MMSLWARFARIN [email protected] [email protected]@TABLET14198NDDF[.]Warfarin Sodium 1 MG Oral Tablet (855288)
Normalized formStrengthIngredientDose form1 MGWarfarin SodiumOral TabletStrengthIngredientWarfarin Sodium 1 MG (855287)Clinical drug componentClinical drug formIngredientDose formWarfarin Oral Tablet (374319)StrengthIngredientWarfarin Sodium 1 MG Oral Tablet (855288)Clinical drugDose form
Relations among drug entitiesIngredientBrand NameAzithromycinZithromaxC. Drug Comp.C. Drug FormAzithromycin 250 MG Azithromycin Oral TabletB. Drug Comp.B. Drug FormAzithromycin 250 MG Azithromycin Oral Tablet[Zithromax][Zithromax]C. DrugB. DrugAzithromycin 250 MG Oral TabletZithromax 250 MG Oral TabletG. PackB. Pack{6 (Azithromycin 250 MG Oral Tablet) } PackZ-PAK
RxNav – RxNorm browserhttps://mor.nlm.nih.gov/RxNav/28
Resources for processing drugdatasets – Drug classificationsystemsPart1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics29
ATC/DDD Indexhttp://www.whocc.no/atc ddd index/Origin World Health Organization (WHO) Collaborating Centre for Drug Statistics Methodology (Norway)Purpose For drug utilization research / pharmaco-epidemiology 1300 classes (1-4)Organization Drug classification on 4 levels Anatomical Therapeutic ChemicalDrugs (5th level)
Established Pharmacologic Classes (EPCs)Origin Veterans Health Administration’s Medication Reference Terminology (MED-RT) For use by the U.S. Food and Drug Administration (FDA)Purpose For drug classification in the Structured Product Labels 600 classesNo hierarchical organizationExamples Macrolide antibacterial (e.g., Azithromycin)
Mechanism of action (MoA)Physiologic effect (PE)Chemical structure (Chem)Origin Veterans Administration’s Medication Reference Terminology (MED-RT)Purpose For drug classification in the Structured Product Labels For drug classification at the VANumber of classes MoA: 600; PE: 1900; Chem: 10,000Hierarchical organizationExamples MoA: HMG-CoA Reductase Inhibitor (e.g., atorvastatin) PE: Decreased Blood Pressure (e.g., enalapril) Chem: Penicillins (e.g., amoxicillin)
VA ClassesOrigin Veterans Administration’s National Drug FilePurpose For drug classification in the VA formulary 500 classesShallow hierarchical organization (3 levels)Examples L1: ANTIMICROBIALS L2: PENICILLINS AND BETA-LACTAM ANTIMICROBIALS L3: QUINOLONES (e.g., Ofloxacin 200 MG Oral Tablet)NB: links to clinical drugs rather than ingredients
RxClass – Drug class browserhttps://mor.nlm.nih.gov/RxClass/
Resources for processing drugdatasets – NLM drug APIsPart 1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics35
NLM drug APIsExpose the content of RxNorm, RxTermsand MED-RT (and other resources) Logical structure, not storage format Up-to-date information (monthly updates of RxNorm) Additional features Normalized and approximate matching; spelling correction Drug-drug interactions checking (from DrugBank) Link to drug classes (from ATC, DailyMed, MeSH, MED-RT) Archive of NDCs since 2007Optimized graph traversal (pre-computed)For use in applications Web services SOAP, REST (XML, JSON) Independent of any programming language
https://rxnav.nlm.nih.gov/API documentation and examples37
Esomeprazole (283742)Esomeprazole (A02BC05)Esomeprazole 40 MG DelayedRelease Oral Capsule (606730)Esomeprazole 40 MG Delayed Release OralCapsule [Nexium] (606731)rxnorm:findRxcuiById(“NDC”, “0186-5040-31”, 0) 606731001865040310186-5040-31
Esomeprazole (283742)Esomeprazole (A02BC05)Esomeprazole 40 MG DelayedRelease Oral Capsule (606730)Esomeprazole 40 MG Delayed Release OralCapsule [Nexium] (606731)rxnorm:getRelatedByType( “606731”, “IN”) 283742001865040310186-5040-31
Esomeprazole (283742)Esomeprazole (A02BC05)Esomeprazole 40 MG DelayedRelease Oral Capsule (606730)Esomeprazole 40 MG Delayed Release OralCapsule [Nexium] (606731)rxclass:getClassByRxNormDrugId( “283742”, “ATC”, “ALL”) A02BC, Proton pumpinhibitors001865040310186-5040-31
Esomeprazole (283742)Esomeprazole (A02BC05)40 MG Delayed Release Capsule [Nexium] (606731) A02B, DRUGS FOR PEPTIC ULCER AND GASTRO-Esomeprazole 40 MG DelayedRelease Oral Capsule (606730)OESOPHAGEAL REFLUX DISEASE (GORD);00186504031A02, DRUGS FOR ACID RELATED DISORDERS;A, ALIMENTARY TRACT AND METABOLISM0186-5040-31
RxMixhttps://mor.nlm.nih.gov/RxMix/Graphical interfaceto the drug APIs RxNorm, MED-RT, RxTerms, RxImage, Interactions, RxClass, MedEx, DailyMedHandles interoperability between functionsHelps users compose complex queries Find all the NDC codes for a given allergy class (e.g., barbiturates)Supports batch execution
Common use casesPart 1 – Resources and use [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics43
Common use casesPharmaco-epidemiology Assess exposure to drugs (by ingredient or class) Assess prescribed daily doseIdentify potentially inappropriate medications Elderly (Beers) Pregnant women (Briggs)44
Use case #1Pharmaco-epidemiology[Bodenreider, AMIA, 2014]
Prescribed vs. defined daily doseDataset Surescripts feedAll prescriptions to ER patientsFor 3 months in 2011 in a Bethesda hospitalReference for defined daily dose: ATCMethods RxNorm clinical drug RxNorm ingredient ATC ingredient ATC defined daily dose prescribed dailydoseRestricted to systemic drugs (based on dose form)Findings Confirmed feasibility25% of the prescriptions exactly match the ATC DDD50% of the prescriptions within 66-150% of the ATC DDD75% of the prescriptions within 50-200% of the ATC DDD
Methods ExampleRxNormAmoxicillin500 MG OralCapsule(308191)ATC/DDD IndexAmoxicillin (723)amoxicillin(J01CA04)Oral CapsuleO1gSurescriptsAmoxicillin500 MG OralCapsule(308191)40capsules10 days40 x 500mg / 10 2g
Results Prescription classificationN pentinQuetiapineOxycodoneFluoxetineDuloxetine
Deviation of the prescribed daily dose (PDD) in Surescriptsfrom the defined daily dose (DDD) in ATCfor 68,462 oral solid dose form prescriptionsPDD/DDD(log scale)3x2x3.5% 300%of the ATC DDD86.1% of the prescriptions are within 33%-300% of the ATC DDD76.1% of the prescriptions are within 50%-200% of the ATC DDD1.5x49.5% of the prescriptions are within 66%-150% of the ATC DDD28.6% of the prescriptionsexactly match the ATC DDD2/31/21/310.4% 33%of the ATC DDD
Use case #2Identifying potentially inappropriate medications for elderly patients[Mundkur, AMIA, 2016]
PIMs for elderly patientsDataset Medicare Part D 1M beneficiaries 65 All prescriptions for one year (2009)Reference list of PIMs: Beers listMethods NDC RxNorm clinical drug ingredient Beers Restricted to systemic drugs (based on dose form)Findings 47% of all beneficiaries were prescribed at least 1 PIM Top PIMs: zolpidem (6.3%), nitrofurantoin (4.5%)
Methods ExampleRxNorm55111047901Zolpidemtartrate 10 MGOral Tablet(854873)BeerszolpidemzolpidemOral TabletOral PillMedicare55111047901Demographic data470,523 prescriptionsDFG filter
Use case #3Identifying potential risk in drug prescriptions during pregnancy[Dhombres, AMIA, 2016]
Potential risk during pregnancyDataset Large prescription dataset from private insurer (150M patients) 3.7M pregnant women; 19M prescriptions (2003-2014) OMOP common data modelReference list for risk during pregnancy: Briggs textbookMethods RxNorm clinical drug ingredient Briggs drug fetal risk Restricted to systemic drugs (based on dose form)Findings 41.2% compatible with pregnancy or probably compatible 55.6% potential risk 3.29% high risk or contraindicated
Specific challengeObsolete identifiers NDC drug manufacturer packaging information 250,000 active NDCs 300,000 obsolete NDCs in the past 10 years 220.000 “alien” NDCs (not curated by RxNorm) Obsolete NDCs Removed from RxNorm (e-prescribing use case) Needed for analytics (longitudinal datasets) RxNorm API provides access to obsolete NDCs Mapping obsolete NDCs to active drugs rxnorm:getNDCStatus( ndc, startDate, endDate, option )List of all NDCs – active or obsolete – for a given drug rxnorm:getAllHistoricalNDCs( rxcui, history )
Other challengesReuse of identifiers NDCs (time-indexed)Insufficient coverage in RxNorm International drugs Over-the-counter drugsGranularity of knowledge Ingredient-class vs. clinical drug-classHeterogeneity of drug classification Different use cases
Impact assessment1B prescriptions from Medicare analyzed Over a 10-year period (2005-2014)Vast majority of NDCs can be resolved with the RxNorm API functions Minor issues Start/End date do not match prescription date Ambiguous mapping (multiple RxCUIs; often clinically insignificant – generic vs. brand) 5% unmapped NDCs (mostly supplies; OTCs)57
AMIA 2018 – San Francisco, CAW14: Analyzing Large Drug Prescription Datasets - Principles, Tools and Examples(sponsored by Pharmacoinformatics Working Group)November 3, 2018Part 2 – Drug data processing in practiceOlivier Bodenreider, MD, PhD, NLMVojtech Huser, MD, PhD, NLMChristian Reich, MD, PhD, IQVIA
Part 2 overviewDetailed look at the APITry it yourself (Follow-along examples with RxMix) and get help (if you hit a problem) 5-10 minR code examples59
/master/rxnormhttps://github.com/mpancia/RxNormR (not used in this session)
Detailed look at the APIPart 2 – Drug data processing in [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics61
API and RxMix nih.gov/RxMix/
Esomeprazole(283742)Esomeprazole 40MG DelayedRelease OralCapsule (606730)Esomeprazole (A02BC05)Esomeprazole 40 MGDelayed Release OralCapsule [Nexium] (606731)0018650403101865040-31
Functionsrxnorm:findRxcuiById Parameters id string: NDC AllSourcesFlag: 0 Input: 00186504031 or 0186-5040-31 Output: 606731rxnorm:getRelatedByType Parameters: term type: IN Input: 606731 Output: Esomeprazole 283742
Try it yourself (Follow-alongexamples with RxMix)Part 2 – Drug data processing in [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics65
RxMix
Library of pre-built workflows
Lyrica 150 MG Oral CapsuleNCD: 00071101668
Resolving NDC 00071101668 in RxNav
Getting properties for NDC 00071101668
Getting all other NDCs for Lyrica 150 MG Oral Capsule
Getting all RxNorm entities related toLyrica 150 MG Oral Capsule
Selecting parameters
Getting ATC classes for Lyrica 150 MG Oral Capsule
Getting drug-drug interactions forLyrica 150 MG Oral Capsule
Nonsensical query returning no results
Output options - JSON
TXT output
Saving your workflow
R code examplesPart 2 – Drug data processing in [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics80
AMIA 2018 – San Francisco, CAW14: Analyzing Large Drug Prescription Datasets - Principles, Tools and Examples(sponsored by Pharmacoinformatics Working Group)November 3, 2018Part 3 – Experience with OHDSIOlivier Bodenreider, MD, PhD, NLMVojtech Huser, MD, PhD, NLMChristian Reich, MD, PhD, IQVIA
Part 3 overviewClinical data modelsHandling international drugs88
Clinical data modelsPart 3 – Experience with [email protected]@AMIAinformaticsOfficial Group of [email protected]#WhyInformatics89
FDA Regulatory Action over TimeNumber of FDA-caused s2000ies
FDAAA calls for establishing Risk Identification andAnalysis SystemRisk Identification and Analysis System:a systematic and reproducible process toefficiently generate evidence to supportthe characterization of the potentialeffects of medical products from across anetwork of disparate observationalhealthcare data sources
OMOP Experiment 1 (2009-2010) Open-source Standards-basedOMOP Methods LibraryInceptioncohortCase controlCommon Data Model 14 methods Epidemiology designs Statistical approachesadapted for longitudinaldata 10 data sources Claims and EHRs 200M livesAplastic AnemiaAcute Liver InjuryBleedingHip FractureHospitalizationMyocardial InfarctionMortality after MIRenal FailureGI Ulcer HospitalizationBntibs u iolfo ticna s:m erid ytAes h rnt, t omieetc a pilra yc ic y nsrb e pa m ticl ,inaz pal hosen pdr hoon nat ateeTrs:icycl cregression
OMOP Experiment 2 (2011-2012)Methods Case-Control New User Cohort Disproportionality methods ICTPD LGPS Self-Controlled Cohort SCCSObservational data4 claims databases1 ambulatory EMRDrug-outcome pairsPositives NegativesTotal165234Myocardial Infarction3666Upper GI Bleed2467Acute Liver Injury8137Acute Renal Failure2464
European OMOP ExperimentObservational dataAarhusPedianetARSIPCIHSPHARMOMethods Case-Control New User Cohort Disproportionality methods ICTPD LGPS Self-Controlled Cohort SCCSDrug-outcome pairsPositives NegativesTotal165234Myocardial Infarction3666Upper GI Bleed2467Acute Liver Injury8137Acute Renal Failure2464
Data Used in European ExperimentNameDescriptionPopulationAarhusDanish national health registry, covering the Aarhus region.Includes inhabitant registry, drug dispensations, hospitalclaims, lab values, and death registry.2MARSItalian record linkage system covering the Tuscany region,including inhabitant registry, drug dispensations, hospitalclaims, and death registry4MHealthSearchItalian general practice database (no children)1MIPCIDutch general practice database0.75 MPedianetItalian general practice pediatric database0.14 MPHARMODutch record linkage system. Includes inhabitant registry,drug dispensations, hospital claims, and lab values.1.28 M
ResultsOBSERVATIONALMEDICALOUTCOMESPARTNERSHIP
Heterogeneity in estimates due to choice of database Heterogeneity in estimates due to analysis choices Except little heterogeneity due to outcome definitions Good performance (AUC 0.7) in distinguishing positive fromnegative controls for optimal methods when stratifying byoutcome and restricting to powered test cases Self controlled methods perform best for all outcomes
Observational Health Data Sciences andInformatics (OHDSI)Plans and Ambitions
Letter SoupOMOP: ended in 2013 with SymposiumIMEDS: Program at Reagan-Udall Foundation of the FDA Methodological research to inform Industry and Agency Research LabOHDSI: Open Research Collaborative started by OMOP Pis and coordinatedthrough Columbia University Multiple stakeholders: academia, government, industry Multiple geographies: US, Europe, Asia-Pacific Multiple disciplines: Statistics, epidemiology, informatics, clinical sciences Maintains OMOP CDM and Vocabularies99
OHDSI’s visionOHDSI collaborators access a network of 1 billion patients to generateevidence about all aspects of healthcare. Patients and clinicians and otherdecision-makers around the world use OHDSI tools and evidence every day.Join us on the journeyhttp://ohdsi.org100
OHDSI: a global community101OHDSI Collaborators: 220 researchers in academia,industry and government 21 countriesOHDSI Data Network: 114 databases from 19 countries 1.9 billion patients records (duplicates) 222 million non-US patients
OutcomeTreatmentDiseaseA caricature of the patient journeyConditionsDrugsProceduresMeasurementsPerson timeBaseline time0Follow-up time
OutcomeTreatmentWhich treatment didpatients choose afterdiagnosis?DiseaseQuestions asked across the patient journeyConditionsWhichDrugs patientschose whichProcedurestreatments?Does one treatmentcause the outcome morethan an alternative?MeasurementsHow many patientsexperienced the outcomeafter treatment?Person timeBaseline timeWhat is theprobability Iwill develop the disease?Does treatment causeoutcome?Follow-up time0 What is the probability Iwill experience theoutcome?
Classifying questions across the patient journeyClinical characterization: What happened to them? What treatment did they choose after diagnosis? Which patients chose which treatments? How many patients experienced the outcome after treatment?Patient-level prediction: What will happen to me? What is the probability that I will develop the disease? What is the probability that I will experience the outcome?Population-level effect estimation: What are the causal effects? Does treatment cause outcome? Does one treatment cause the outcome more than an alternative?
OMOP Common Data ModelOBSERVATIONALMEDICALOUTCOMESPARTNERSHIP
OMOP Common Data ModelPersonObservation periodSpecimenStandardized health system dataLocationCare siteStandardized clinical dataProcedure costDrug costDevice exposureDevice costConcept relationshipRelationshipConcept synonymConcept ancestorSource to concept mapDrug strengthCohort definitionNoteObservationFact relationshipVocabularyStandardized vocabularyVisit costDrug exposureCondition occurrenceMeasurementConceptPayer plan periodStandardized healtheconomicsProcedure occurrenceCDM sourceProviderDeathVisit occurrenceStandardized meta-dataCohortStandardized derived elementsDrug eraDose eraCondition era
Drug HierarchyDrugClassesClassificationsVA ClassCVXNDFRTNDFRT IndATCFDB IndETCSPLSNOMEDVA ClassCVXNDFRTNDFRT IndATCFDB IndETCSPLSNOMEDDrugsIngredientsRxNorm RxNorm ExtensionDrug Forms and ComponentsRxNorm RxNorm ExtensionDrug productsRxNorm RxNorm ExtensionStandardDrugVocabulary:Source iptGenseqnoDrugCodesEU Productdm dDPDAMISBDPMProcedure DrugsHCPCSCPT4
Simple Use CaseGive me all patients who takePlavix 75 mg Tablets108
OMOP VocabularyBrandedDrugclopidogrel 75 MGOral Tablet [Plavix]
OMOP VocabularyBrandedDrugclopidogrel 75 MGOral Tablet [Plavix]Plavix
OMOP VocabularyClinical Drugclopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]
OMOP VocabularyDrug Formclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]
OMOP VocabularyDrug Formclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]Tablet
OMOP Vocabulary RelationshipsDrugComponentclopidogrel 75 MGclopidogrel 75 MG[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]
OMOP Vocabulary RelationshipsDrugComponentclopidogrel 75 MG75 MGclopidogrel 75 MG[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]
OMOP VocabularyIngredientclopidogrelclopidogrel 75 MGclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]clopidogrel 75 MG[Plavix]
OMOP VocabularyIngredientclopidogrelclopidogrel 75 MGclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MG[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]RxNorm WWWWWWWWWWWWWWWWW
OMOP VocabularyPhysiological LMCOATEDAggregationFILMCOATEDFILM COATEDAlterationETCPlateletAggregationInhibitors ThienopyridinesATCIndicationPlatelet aggregationinhibitors ILMCOATEDORALFILMTABLET,COATEDFILM COATEDFILM COATEDclopidogrelclopidogrel 75 MGclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]clopidogrel 75 MG[Plavix]
OMOP T,FILMCOATEDAggregationFILMCOATEDFILM LTABLET,FILMCOATEDAggregationFILMCOATEDFILM COATEDAlterationPlateletAggregationInhibitors &CombinationsANTITHROMBOTIC AGENTSPlateletAggregationInhibitors ThienopyridinesPlatelet aggregationinhibitors COATEDORALFILMTABLET,COATEDFILM COATEDFILM COATEDclopidogrelclopidogrel 75 MGclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]clopidogrel 75 MG[Plavix]
OMOP T,FILMCOATEDAggregationFILMCOATEDFILM LTABLET,FILMCOATEDAggregationFILMCOATEDFILM COATEDAlterationPlateletAggregationInhibitors &CombinationsANTITHROMBOTIC AGENTSPlateletAggregationInhibitors ThienopyridinesPlatelet aggregationinhibitors COATEDORALFILMTABLET,COATEDFILM COATEDFILM COATEDclopidogrelclopidogrel 75 MGclopidogrel Tabletclopidogrel Tablet[Plavix]clopidogrel 75 MG[Plavix]clopidogrel 75 MGOral Tabletclopidogrel 75 MGOral Tablet [Plavix]GPIClopidogrelBisulfateTab 75 ix]COATED[plavix][plavix][plavix]CLOPIDOGREL 75 MGOral sBDPMFranceSPL
FILMCOATEDAggregationFILMCOATEDFILM COATEDAlte
Nov 03, 2018 · W14: Analyzing Large Drug Prescription Datasets - Principles, Tools and Examples . Part 2 – Drug data processing in practice (VH) Implementing use cases with RxMix and R. Part 3 – Experience with OHDSI (CR) . B. Drug Form. B. Drug. G
