Transcription
Real world evidence (RWE) – anintroduction; how is it relevant for themedicines regulatory system?London, EMA, April 2018Hans-Georg EichlerSenior Medical OfficerAn agency of the European Union
Why do we need RWE?Case study 1: Patients with unprovoked venousthromboembolism (VTE) after 3–6 months on anticoagulants:recurrence risk of 5–10% per year. Who should/should notreceive life-long anticoagulants?Derivation study to define prediction model: Prospective; 929patients; approximately 17 years, cost 12-15 millionValidation study: secondary analysis of pre-existing (trials) data; 6 months, cost 100.000Could a study like this be based on pre-existing e-HRs (i.e.“RWE”), only faster and cheaper?1Eichinger et al, Circulation 2010; Marcucci et al, J Thromb Hemost 2015
Why do we need RWE?Case study 2: Gene therapy for thalassemia holds promise foronce-only administration, initial conditional approval after a fewyears observation in clinical trial plausible.Limited information on duration of the effect, safety concern overvector-based gene therapies, i.e. insertional mutagenesis leadingto oncogenesis; several years for tumours to develop, risk level islikely to be low decade-long (or life-long) surveillance of allpatients unrealistic in the interventional research settingReliable RWE (from eHRs) would be a key enabler for thedevelopment, licensing, reimbursement and safe use2
Why do we need RWE?Case study 3: Oncology: combination therapy offers greatestpotential for most patients; a plethora of potential molecules andpathways to target are available. huge “combinatorial complexity”: dose selection, drugcombination partners, sequence of treatments, washout periods,changing tumour characteristics, patient-/tumour-relatedstratification biomarkers the full potential and optimal use ofnew agents cannot be characterised before routine use.RWE is our only hope to come to grips with combinatorialcomplexity3
A look over the fence - to other industriesE-controls, sensors real-time analysis keep plane safe in the air and inform next-gen product design4
Sensors for real-time monitoring; geocoded maps; soil, weatherconditions raise agricultural productivity and inform next-genproduct and services design5
Across the fence everyday use (‘practice’) and R&D are not two separateactivities;future learning and rapid feed-back loops between use and R&Dare built into the system from scratch ‘data-driven innovation’.How about healthcare?research settings(‘learning’)6everyday clinical practice(‘using’)
Going beyond the research-practice divideGoing beyond Population focus precision (personalised) medicine Monotherapies complex (combination) regimens Short/mid time horizons decade/life-long horizon (Randomised controlled) trials full spectrum of methods Pre-licensing knowledge generation lifespan approach Silos (insurers, regulators, developers) common approach Research-practice divide learning health care system7
E-health records:the linchpin of a learning healthcare systemAre we ready?Organisation for Economic Co-operation and Development(OECD) conducts (repeat) surveys of member countries report on the:“Readiness of electronic health record (eHR) systems tocontribute to national health information and research” [OECDwebsite]Note: routinely collected data (“longitudinal record .”), nottrial data, not registries, not wearables, 8
Are we ready?9Source: OECD website (2017) “New health technologies: Managing access, value and sustainability”
Why not learn from learning airplanes,learning harvesters?Bottlenecks: Technology Data ownership, politics, .“We are different ” Informed consent Patient data protectionWith adequate personal data protection and against undesirableuse - most patients will support a learning healthcare system10
Can we accelerate the implementation of aRWE-learning/ learning healthcare system?Giving the Bandwagon a big push at the level of. Political/public debate Best practice Implementation Methodology development11
Political/public debate (1)(All!) OECD Health Ministers 2017 agreed “that governmentsestablish a national health data governance framework toencourage the availability and use of personal health datato serve health-related public interest purposes while promotingthe protection of privacy, personal health data and data security.”to “encourage common data elements and formats; qualityassurance; data interoperability standards; common policies thatminimise barriers to sharing data for health systemmanagement, statistics, research and other health-relatedpurposes that serve the public interest.”12
Political/public debate (2) Reinforce the urgency that opportunities for patients arelost; science progresses faster than the “system”, impedingthe development and best use of novel treatment options. Myth-busting: patient-data protection and secondary datause are not a trade-off; both can be achieved at the sametime. Shift the debate: “Analysing personal health data is a risk toindividuals” “Not analysing personal health data is a riskto individuals”.13
Sharing best practice(examples of ‘building blocks’) New Zealand: public consultation in 2015 national eHR,single longitudinal view accessible to consumers, carers anddecision-makers; support precision medicine, Australia: legislation from an opt-in to an opt-out patientconsent model 13 countries: offer financial incentives to encourage healthcare providers to adopt eHRs that conform to natl. standards EMA: ENCePP collaboration Code of Conduct to facilitatecooperation between the private pharmaceutical sector andhealthcare systems14
Implementation(examples)“People Who Say It Cannot Be Done Should Not InterruptThose Who Are Doing It”. Kaiser Permanente: integrated eHR system and applied BigData analytics algorithm to predict likelihood of sepsis innew-borns, reduce unnecessary use of antibiotics;adherence and glycaemic and blood pressure control improve disease management FDA Sentinel initiative: “share answers not data” successfully address drug safety concerns15
Methodology?Algorithms and (statistical) methods to extract, analyse, andinterpret eHR data are in place and broadly acceptable for anumber of research questions.Achilles' heel: conducting relative effectiveness studies; inherentrisk of bias and confounding due to the non-randomised nature ofthe comparisonAre we making progress?Can real world data studies (based on eHRs) match the results ofan RCT? (predict versus confirm!)16
17Slide courtesy of S. Schneeweiss, submitted for publication
ConclusionsScience progresses now faster than the “system”, impeding thedevelopment and best use of novel treatment options.Leveraging RWE is a need – and an achievable goal.Accelerating the use of RWE requires a concerted effort and thenecessary upfront investments.The good news: this is not a zero-sum game, all players in thepharmaceutical ecosystem stand to gain.18
Thank youfor listening!European Medicines Agency30 Churchill PlaceLondon E14 [email protected]
interpret eHR data are in place and broadly acceptable for a number of research questions. Achilles' heel: conducting relative effectiveness studies; inherent risk of bias and confounding due to the non-randomised nature of the comparison . Are we making progress? Can real world
