Cross-European Research Projects: Developing and validating disease management evaluation methods for European health care systems (DISMEVAL) Jack Needleman UCLA School of Public Health AcademyHealth June2009 Goals for the Presentation • • • • • Project background and goals Project scope Consortium members and individual projects Synergies Comments and reflections Project background Scope of current DM initiatives • DM response to extensive, expensive chronic illness Expand patient education, support & services outside traditional health plan to facilitate patient self-management Education, support and reimbursement for providers to effectively manage patients • Many models, inputs, strategies • Many diseases, although Diabetes major focus • Many health care systems in Europe and elsewhere are beginning to embrace this strategy E.g., Germany encourages sickness funds to offer DM Project background State of DM evaluation • • While conceptually attractive, ability of DM to reduce cost and improve care has not been empirically demonstrated. No universally accepted evaluation methods to measure performance in a scientifically sound fashion that are practicable for routine operations. Selection, with randomization difficult or undone Variations in program content – How are variations in assessment of effectiveness associated with differences in program organization & content Different measures, metrics for evaluating program performance – Costs – Utilization – Clinical indicators DISMEVAL goals • Provide overview of approaches to chronic care, DM methods, and DM evaluations across Europe • Test and validate possible evaluation approaches Identify best practices Develop evidence-based recommendations for policymakers, program operators and researchers on which evaluation approach is most useful in a given context • Promote and support the networking and coordination of research and innovation activities on aspects related to scientific knowledge and policy development in this area Build on existing work carried out in Member States and at the wider European level Consortium members RAND Europe Cambridge Ltd RAND UK London School of Hygiene and Tropical Medicine UK Paracelsus Medizinische Privatuniversitat Salzburg Austria KØbenhavns Universitet UCPH Denmark Denmark Johann Wolfgang Goethe Universitaet - Frankfurt Am Main Germany Universite Paris XII - Val de Marne UPVM France France Universiteit Maastricht UM The Netherlands Instituto de Salud Carlos III ISCIII Spain Centre Anticancereux Leon Berard CLB France Review current state of DM programs and evaluation strategies in Europe • Objectives Review approaches to managing (chronic) conditions that have been developed and/or implemented by different countries in Europe Assess whether and how countries evaluate the approaches to (chronic) disease management • Update existing information for 5 countries/new info for 9 Common template Obtain documentation/grey literature • Products Description of programs Descriptions of evaluation approaches Synthesis of evaluation approaches – lessons/best practices Examples of questions to be addressed • • Indicators Which domains (e.g. cost, quality, patient satisfaction) should be used to measure the effect of disease management programmes What are appropriate and established measures for these domains, and what is the evidence behind them? Methods Which non-experimental attribution strategies have been well enough validated to consider them viable alternatives to experimental designs? How do effect size estimates from non-experimental designs track those from experimental designs How can you account for confounding factors in experimental designs? How do contextual factors influence the choice of evaluation design? Implement evaluation of 6 DM programs Design in coordination Austria Diabetes Denmark Diabetes & COPD Germany Diabetes France Diabetes / cancer Netherlands Diabetes Spain CVD prev. Cluster RCT. Intervention then introduced into control group as open study. Some observational data from same DMP introduced into two other regions Pre-Post observational data during ‘project period’, some follow up data. Nested RCT of hospital vs community care. Limited comparative data from national diabetes database going back 2-3 years (hospital patients only) Longitudinal case control study, some earlier data. Possibly limited data from wider sick fund data. Interest in testing different matching procedures. Pre-Post observational data (patient and network level). Heterogeneous networks (some option of analysis of intensity of intervention). Data for matched controls from claims. Pre-Post observational data for 10 pilots (and some data for 10 additional ones). Some intermediate outcomes for the rest of the Netherlands Observational data on risk factor reduction. Comparison of enrolled vs nonenrolled patients (patient choice). Starting Dec 08, systematic differences in intervention by region. Synergies: the value of a consortium: Designing in coordination offers… • Peer consultation on design Additional methodological advice Encouragement to push beyond original design, e.g. – Germany on case-control – France on control group Applying lessons from one evaluation to others Clever expansions to allow methods to be contrasted • Opportunities for pooling & comparing/contrasting results: By disease By DM strategy Measures of dose-response, or more effective models Examples of discussion from first consortium meeting • What is the range of things you can measure? Knowing that the audiences of programme evaluations differ, what outcome measures should have the higher interest? • What are the selection issues and with what confidence can we use tools for solving selection bias (e.g. regression of the mean as a selection problem)? That is, do results differ when we apply different tools and examine the same database in alternative ways? Which of these tools is most easy to apply? • Dose-response issue. In theory, disease management should work but in practice evaluations have been disappointing. What are the formative components that help a programme work? How can you measure delivered dose? How can you set up your evaluation so that you can learn about the formative components that work? Consortiums & EU sponsorship may require new recordkeeping and vehicles for administrative and program coordination • EU expectations How expenses reported Timekeeping • Consortium administrative and substantive issues Joint authorship Data use/sharing/acknowledgment agreements Intra-consortium communication – Listserves, websites, intranet Common dissemination strategy and “boilerplate” language, templates Personal reflections from initial work • Frustration with additional administrative demands but… • Strong teams with access to unique data • Enthusiasm and curiosity about what consortium can mean for own research • Willingness to be open about weaknesses and limitations of data, design • Jury is out on whether, as pressure to complete work grows, promise of collaboration, exploration and stretching is achieved, but I am optimistic.