Background and Aim of Study:
An increasing number of data sources are relevant for populating decision analytic models. Detailed methodological advice on which type of data is to be used for what type of model parameter is lacking. We aim at identifying standards on appropriate data sources and the correct handling of such data for modelling purposes.
Method:
We searched for information in Health Technology Assessment manuals and economic (modelling) guidelines that were retrieved via a call among members of the International Network of Agencies for Health Technology Assessment (INAHTA) and by hand search. Documents from Europe, USA, Canada, Australia and New Zealand that are written in English or German were included. Information on appropriate data sources for modelling, their advantages and disadvantages, on data identification methods and on the correct handling of them was narratively summarised.
Results:
Overall, more than 30 relevant data sources for populating models are mentioned in the included documents. They comprise secondary data from published (clinical) studies, secondary data from databases (routine data, compiled statistics) and further sources such as expert opinions. Valid sources for defining parameters on efficacy/effectiveness and safety, on the natural history of disease, on resource use/costs, health state values and on demographic characteristics have been identified. Methodological details on how to further handle the data in the modelling process are hardly specified. Detailed model-type specific methodological advice is lacking.
Conclusions:
There is consensus across the documents on the type of data sources that may be appropriate for populating decision analytic models. Further work is required on how to statistically handle such data in different types of models.