Presentation on theme: "Electronic Health Records Facilitating Clinical Research"— Presentation transcript:
1 Electronic Health Records Facilitating Clinical Research Clinical Research Management WorkshopTuesday, June 22, 2010Anil Jain, MD, FACPSenior IT Executive, Information Technology, Cleveland ClinicManaging Director, eResearch, eCleveland ClinicCo-Director, Biomedical Informatics, CTSC, Case Western Reserve University11
2 Objectives Provide an overview of eResearch at the Cleveland Clinic Review our experience with simulation of clinical trial protocols using EHR dataDemonstrate our ability to enhance clinical trial recruitment with EHRsShow how we have leveraged EHR data for comparative effectiveness research
4 The Cleveland Clinic The Cleveland Clinic Founded in 1921 Not-for-profit group practice2096 Professional Staff and scientistsInnovations include:First coronary angiographyDevelopment and refinement of coronary bypass surgeryFirst minimally invasive aortic heart valve surgeryFirst successful larynx transplantDiscovery of first gene linked to coronary artery diseaseCCHS Statistics (2009):Operating Revenue: ~ 5.6 BillionClinical visits: Nearly 4 MillionAdmissions: 155,000U.S.News & World Report rankings:4th among nation’s 6,000 hospitalsNation’s #1 heart program16 specialties rankedThe Cleveland ClinicThe Cleveland Clinic
5 Cleveland Clinic and Research (2009) Funding & Scope$272 M Total Research Funding$91 M in Federal NIH AwardsOver 2,200 active projects434 Active Registries with avg. 4 – 6 cohorts per registryGeneralizability of Patient Population75% of patients came from Cleveland’s seven adjacent counties1.5 M visits in the regional medical practice sites (community-based clinics)“Centers of Excellence” for numerous diseases areas
7 EHR data is routinely used at our Institution… Health Wellness and PreventionImmunizations, Osteoporosis, Diabetes and Cancer screeningDisease-based ReportingDiabetes, Hypertension, Heart Failure, AsthmaPatient SafetyAdverse Drug Events and sentinel eventsFDA Public Health Advisories (e.g., Vioxx® and Ortho-Evra®)Local Consensus (e.g., Avandia ® notification)Public ReportingPediatric Immunizations (SIRS), Communicable Diseases, process measures and quality indicators (JCAHO, HEDIS, CMS, etc.)EHR DataReports
8 Clinical Care Domain Clinical Research Domain The Silos…Clinical CareDomainEMR w/multiple interfacesAggregates ALL avail. dataOrganized around PatientsHIPAA PolicyHL7 a commonMessaging standardNon-Structured & StructuredClinical ResearchDomainIndependent DataOrganized around a StudyWell Behaved DataBiostatisticians ManageDrug Trials21 CFR Part 11Involves Sponsor and FDA+/- CDISC Standards8
9 Our investigators want the EHR to help them! Identify current research subjects in the EHRDevelop tools to help recruit potential research subjectsAppropriately bill activity occurring in EHR-managed visits to the sponsor for research activity versus usual payor for standard of care.Develop Single Source capability with extraction to eCRFCapture rich structured data from the EHR (phenotypic) and combine with bio-informatics data (genotypic)Easily move valid data from EMR into research registriesFacilitate secure EMR access for research monitors9
10 Provides EMR-centric Resources for other groups Quality & PatientSafetyInstituteQuantitativeHealthScienceseAnalytics and Research:Research StandardizationCenter for Clinical Trials &Clinical OperationsClinical & Translational Science Consortium
11 Leveraging the EHR to capture critical data for researchers ElectronicHealthRecordPersonal Health RecordClinical EventsAdverse ReactionsPatient OutcomesClarity???DialysisCenterLocal“EMR”sRegistriesGenomicsImagingRegistrieseResearch Db11
12 Clinical Research Cycle SubjectRecruitmentClinical DataeCRFData CollectionTrialManagementSite SelectionRegulatedDatabaseseResearchServicesProtocolAssessmentResearchBillingProtocolDevelopmentDataAnalysisResearchHypothesisKnowledgeDissemination12
15 A Research Study Scenario (These parameters and numbers are purely fictional and intended only to demonstrate the scenario)InclusionDiabetes Type 2Age 18 to 65 at screeningTreatment Naïve or Oral mono-therapyExclusionUncontrolled HypertensionTriglycerides >= 1000 mg/dlLipid-lowering therapy not stable for 1 yearHistory of myocardial infarction or unstable anginaHistory of coronary artery bypass graft surgery or angioplastyHistory of insulin use (other than gestational diabetes)History of substance abuse or unlikely to finish study
16 Analysis of Protocol Criteria Inclusion and Exclusion Study CriteriaData SourceDemoVitalsDx/ProcRxLabChartDiabetes Type 2XAge 18 to 65Rx Naïve or mono-therapyNo Uncontrolled HypertensionNo Triglycerides >=1000If being treated, stable dose of lipid-lowering agentNo Hx of MI/USA, CABG/PTCANo Insulin use (except gest DM)No history of substance abuseLikely to complete study
17 Ontology & Vocabulary Information Standard Vocabulary Diagnoses ICD9 CodesIMO TermsSNOMED Mapping (limited, via IMO)MedicationsFirst Data Bank / NDDF+National Drug CodeRxNorm (limited, thru eResearch cross-walk)LaboratoryLocal CodesLOINC Mapping (limited, thru eResearch cross-walk via RELMA tool)AllergiesLocalFamily HistoryProceduresCPT™, ICD9 Procedure Codes
18 High-level Summary of the Impact of Each Criterion
19 Assess Each Criteria for Ethnic/Gender Diversity
20 e Trial Support eResearch Tools Traditional Recruitment Strategies ClinicalData RepositoryElectronicHealthRecordeResearchToolsTraditionalRecruitmentStrategiesResearchCoordinator2020
23 Physician attitudes…Embi PJ, Jain AK, Harris CM. Physicians' perceptions of an electronic health record-based clinical trial alert approach to subject recruitment: A survey. BMC Med Inform Decis Mak Apr 2;8:13.
24 What are the characteristics of the alert? False PositivesMany referrals made for each enrollee – excessive false positives…EHR may not capture key criteria.Chart review may not validate “computable” criteriaPatients may not necessarily be good candidates or willing to consentFalse NegativesDocumentation gapTime lag between presentation and documentationOnly patients who have come in for a visit??
25 Integrating study criteria with scheduling… SiteAdminProviderClinicalTrial AlertStudy Filter(patient &provider)StudyCriteriaAUTOMATEDPROCESSDURING THENIGHTDURING THEOFFICE VISITPatient Schedules(sites)SecureEHR orderenrolls Ptinto StudyPatientTrackedvia EHR25
26 EHR-data based recruitment lists work! Cleveland Clinic involved in multi-site clinical trial for safety of NIH - H1N1 vaccine among children with severe asthma.Comparison of the eResearch services to the Severe Asthma Research Program (SARP) network registry.eResearch led to higher enrollment, 93/540 (17.2%) eResearch vs. 24/109 (22%) for SARP.Performance was similar to the volunteer registry without significant increase in costly screen failuresDiversity in terms of race/ethnicity of the subjects was increased using EHR-based identificationParikh P, Jain A, et al. Recruitment and Enrollment of Asthmatics in a Phase II Clinical Trial, ATS Meeting, May 17, 2010.
27 Increasing participation and diversity… (U Pitt) Over a 22-month period, EMR-prompts for recruitment:PCPs referred 794 patients via EMR-prompts and 176 (22%) met study inclusion criteria and enrolled,8,095 patients were approached by wait room-based recruiters of whom 193 (2.4%) enrolled.Subjects enrolled by EMR-prompted PCPs were more likely to be non-white (23% vs 5%; P < 0.001), male (28% vs 18%; P = 0.03)Rollman BL et al. Comparison of electronic physician prompts versus waitroom case-finding on clinical trial enrollment. J Gen Intern Med Apr;23(4):27
28 Three recent outcomes and CER projects… Modeling cardiovascular outcomes in patients on oral hypoglycemic agentsModeling cardiovascular complication rates in patients admitted to the hospital with acute coronary syndromeIdentifying determinants of progression of kidney disease in patients with chronic kidney diseaseWhy was the EHR used?Size and scope of required electronic data was mostly already in the EHRCompetitive advantage for obtaining sponsors
29 Strategies to Overcome EHR Data Reliability Issues RELIABILITY CHALLENGESDeath is not always reliably captured in an EHR derived data set.Documentation of certain exclusions and adverse events are generally not captured as structured dataPrescription medication dispensed and taken including OTCPatients may have fragmented care with some clinical data outside institutional EHR.POTENTIAL SOLUTIONSEHR data can be linked to SSDI assuming appropriate permissionsText mining/analysis of progress notes, discharge summaries, operative reports can identify some events/adverse eventsWidespread adoption of Medication-ReconciliationLeveraging Health Information Exchanges or inclusion of payer data can mitigate the problems with fragmented care
31 Explorys Population Explorer Cleveland Clinic spin-off - Software-as-a-Service offering…Explore: Search, browse, and define cohorts based on clinically normalized dataset from multiple providers.Compare: Analyze temporal measures between cohorts.Collaborate: Safely connect and share with trust peers and sponsors.Engage: Analyze in-depth HIPAA compliant datasets or recruit across internal or distributed trusted networks.
32 The Explorys Unified Platform Healthcare CloudSecurehigh-speed VPNOLAPCompMatrixSearchEngineAlertPopulationExplorationComparativeEffectivenessBusinessIntelligencePharma-covigilanceOpenModuleQuestions:
33 “Not everything that can be counted counts, and not everything that counts can be counted.” Sign hanging on Albert Einstein’s Princeton University OfficeQuestions?Anil Jain, MD, FACPCleveland Clinic