Nermina Ferizovic | Health Economics Expert

As argued in a previous blog post, the health economics profession should encourage or actively work to develop data sources that would be better suited for health economic needs. This is as opposed to attaching themselves to clinical trials which are not aimed at generating health economic data. The overall advantage of this would be more informed decision making with reduced uncertainty.

However, the development of such a data source would not be as simple as first appears. It would require multiple agencies and organisations working together with an agreed agenda and specification. Significantly, it would take years. It may also be that this is more of an idea suited to an idealised hypothetical world rather than to the one in which we reside. In this ideal world, what would the data source look like? Perhaps we already have data sources that can more effectively be joined together to provide us with the required data?

For example, in England the Hospital Episode Statistics, Clinical Practice Research Datalink and Office for National Statistics data potentially allow researchers in England access to primary and secondary healthcare data as well as mortality data. This data can even be linked together giving a wider specification to the patient-level data. However, currently this data is relatively difficult to obtain due to administrative burden, is expensive and may require long periods of time to acquire – something which is unhelpful to most analysts when generating their health economic evidence. Additionally, once acquired, the data sources may have large swathes of missing data, not be representative of the population being analysed or simply possess small sample sizes. The latter is particularly the case in orphan diseases where pre-existing data sources, even those that may hypothetically be relatively simple and straightforward to access, cannot be used due to non-robust sample sizes.Perhaps, then, it is not a case of developing a universal data source suited to health economic needs but more about developing a suitable framework for researchers to guide them to the appropriate data source, or type of data source that should be used depending on the research question. This may be particularly relevant given the potentially large research questions that can be posed by health economists.