School of Economics and Management LIUC University, Castellanza, Italy

Transcription

1 Stefano Capri School of Economics and Management LIUC University, Castellanza, Italy ISPOR Issue Panel, Dublin SESSION IV, Wednesday, 6 November 2013 What should health economists, from industry or academia, do to make models more trustworthy? Health economists have just to apply scientific methods and knowledge at their best The role of models and the institutional framework has to be changed The following issues illustrate these points Stefano Capri 2

2 Industry-sponsored studies generally considered as biased HTA-commissioned models maybe biased too? Imagine a government who takes decisions concerning the political economy based on econometric models provided by an industrial company, or by a bank or by a private financial institution. It would be a nonsense. Governments take decision by using models built internally (Central Bank, Ministry of Economy) and they look also to models provided by international organisations (IMF, OECD) and created by Universities and centres of research. Stefano Capri 3 Pricing & Reimbursement decisions concerns the health policy, therefore in a similar way the models should be provided by public institutions. However a comparison with manufacturer s models and also the building of collaborative models should be encouraged. Stefano Capri 4

3 A reference model, e.g. for each specific disease, within the HTA will reduce duplication and effort (Afzali et al. Med Decis Making April 2013) Useful at national level, but not for an European reference model. A reference model would need a single European HTA agency Problems of who does the model, who and when is updated, with which data, etc. Different models, different forecast (like IMF, OECD, etc. models about the macroeconomic forecast). Stefano Capri 5 Trasparency [ clearly describing the model structure, equations, parameter values, and assumptions to enable interested parties to understand the model (Eddy D et al. Med Decis Making 2012) and to replicate: Publicly funded models are easily shared, but problems with privately developed models with strong intellectual property (limit disclosure of the their methods). For an HTA agency is not acceptable this limitation any longer: the same does not occur with clinical trials Stefano Capri 6

4 HTA agencies has to give clear instructions about the crucial issues of every model: The choice of comparators Parameters selections Time horizon Appropriate techniques related to specific disease problem (e.g. Markov is appropriate for recurrent health states without the need of memory of the patient, discrete choice-individual based state transition- appropriate with high number of states and influence of memory of the patient, decision tree is appropriate for short term modelling) Electronic copy of the model, transparent and operable Stefano Capri 7 Like with economic models used to take heavy decisions for the economy (interest rate, public expenditure, investments, etc.) HTA models have to be updated and reassessed frequently Limits: P&R decision might change (price revision, high uncertainty for companies). Possibly to revise the models at fixed point in the time (after first year, second year of commercialization). Real world data might be some input of the revised models. Stefano Capri 8

5 Whatever the models are built, and independently by the authors, they need to be updated periodically (new data, new hypothesis, new instruments, new constraints) which implies a regularly new appraisal new P&R decisions (every two years?). Limits: uncertainty for manufacturer (new price, new restrictions, etc.) industrial planning?; work and costs for the public agencies. Strengths: more solid evidences, best decisions, better resource allocation, benefits for the patients. Stefano Capri 9 HTA authority requires resources for reviewing models and for building internal models. HTA agencies, in some countries (e.g. Italy) have not the internal skill in order to provide models and more limited resources for reviewing external models. National HTA vs. Regional HTA Stefano Capri 10

6 A model has to be predictive (uncertainty) [ the best use of predictive validation is to simulate a clinical trial (Eddy D et al. Med Decis Making 2012)] For HTA agencies and payers this is not enough: a model has to predict the economic impact. It is not only a surrogate of the RCT. Stefano Capri 11 More on prediction: Type I errors (approving bad products bad ICER) Type II errors (rejecting good products good ICER) ( good products may be delayed, but in the long run will always be accepted eventually) When the model is accepted, the Authority will approve a drug only when the benefits of approval exceed both the costs of approval and the benefits of waiting further. Stefano Capri 12

7 Thank you for your attention Stefano Capri 13 Stefano Capri 14

8 The cost-effectiveness model submitted by the manufacturer of memantine reported that treatment with memantine dominated best supportive care. The Assessment Group s model, however, estimated an ICER of 32,100 per QALY gained. One of the main NICE technology appraisal guidance differences between the manufacturer s model and the Assessment Group s model was that treatment was assumed to continue in institutionalisation by the Assessment Group but not the manufacturer. (NICE technology appraisal guidance 217: Donepezil, galantamine, rivastigmine and memantine for the treatment of Alzheimer s disease, 2011) Stefano Capri 15