VALIDATING HEALTH ECONOMIC MODELS- DEVELOPING A METHODOLOGICAL APPROACH
Author(s)
Hart WM1, Chandler F21EcoStat Consulting Ltd, Sidestrand, Norfolk, United Kingdom, 2GlaxoSmithKline, Uxbridge, Middlesex , United Kingdom
Models developed in Microsoft EXCEL of increasing complexity are being used as an integral part of the submission process for new pharmaceutical products and for technical assessments produced by national agencies. The credibility of the health economic models utilised depends on their validity. Firstly, in terms of reflecting the appropriate clinical process using the correct data and assumptions and secondly, in terms of ensuring a correct functioning of the model based on the relationships between variables and formulas embedded in the model. In this research, the authors concentrate on the latter aspect of validating health economic models. While recognising that models should be validated, there is a paucity of detailed techniques available in the literature that show researchers and users of health economic models how to validate them. Hence, we have attempted to develop a methodological framework that may be helpful to both those reviewing complex models and to model developers themselves who could incorporate validation processes while they develop their own models. The authors have reviewed the literature, looked at other disciplines (e.g. finance) where modelling plays a central role, investigated different software options in addition to the inbuilt validation tools in EXCEL. The evolving methodology has been applied to a number of existing real-life models in order to develop a consistent approach. By applying the developed methodology, errors have been identified at the design/review stage in a number of budget impact and cost-effectiveness models of varying degrees of complexity. A consistent approach to validation is a useful tool to test the often highly complex processes and relationships with thousands of formulas in health economic models. It may also encourage modellers to take a more disciplined and organised approach in the development process and give increased confidence to end-users that the models they are using can be relied upon.
Conference/Value in Health Info
2012-11, ISPOR Europe 2012, Berlin, Germany
Value in Health, Vol. 15, No. 7 (November 2012)
Code
PRM165
Topic
Methodological & Statistical Research
Topic Subcategory
Confounding, Selection Bias Correction, Causal Inference
Disease
Multiple Diseases