GUIDELINES FOR ENSURING DEGREES OF VALIDITY IN HEALTH ECONOMIC MODELS
Author(s)
Joshua A Ray, MSc, Health Economic Modeller1, Pierre Ducournau, MSc, Statistician1, M Gyldmark, Manager, Head of Modelling, Outcomes Research and Scientific Experts1, Andrew Briggs, DPhil, Professor21F. Hoffmann - La Roche, Basel, Switzerland; 2 University of Glasgow, Glasgow, United Kingdom
OBJECTIVES: Greater emphasis for economic evaluations in healthcare reimbursement decision-making has brought about the need for models that are accurate and methodologically sound. Although guidelines exist suggesting appropriate methods for model validation, there remains a paucity of practical guidance outlining steps to ensure that outcomes have been appropriately validated. The objective was to establish a process outlining a series of practical tests to ensure consistent model validation. METHODS: A systematic search of the literature was conducted to identify modeling guidelines and published modeling studies discussing validation techniques. Results from the literature search were used to develop practical steps to ensure consistent model validation. RESULTS: A list of validation tests were compiled and collated following a review of 15 guideline publications on model validation. Model validation should not only encompass checking for calculation errors or structural inconsistencies, but also include a complete synthesis of the evidence-based data used to produce results. Initially, the model structure should be reviewed for face validity, assessing the economic and clinical assumptions that include consulting scientific experts, external to the model development process. Secondly, convergent validity should be conducted where a series of tests are performed comparing if results from other independently developed models draw similar conclusions. Internal validity involves a series of procedures to ensure the basic model framework and mathematical calculations are consistent. Test-runs should be performed including a replication test, and extensive sensitivity analyses, whereby a series of extreme/hypothetical assumptions are explored. Further assurances of internal validity include the replication of the model in another software. CONCLUSION: Approaches to model validation should be included as part of any publication. Budgetary and time allocation should take model validation into account given the increased importance placed on the outcomes of health economic models.
Conference/Value in Health Info
2007-10, ISPOR Europe 2007, Dublin, Ireland
Value in Health, Vol. 10, No. 6 (November/December 2007)
Code
PMC33
Topic
Methodological & Statistical Research
Topic Subcategory
Modeling and simulation
Disease
Multiple Diseases