Should a submission use an existing economic model or commission a new one?
You’re a pharma company with a new license for your product in an area like rheumatoid arthritis (RA), breast cancer or diabetes, disease areas where there has been lots of previous economics work carried out. As you plan your HTA submission do you look for an existing model or commission a new one?
I was thinking about this issue as I read the HTA monograph “Newer Agents for Blood Glucose Control in Type 2 Diabetes” by Norman Waugh, Ewen Cummins and colleagues. Building from the UKPDS model, these researchers adapted the model to include the direct effect of weight changes on utility, costs and utility loss from severe hypoglycaemic attacks, and utility loss from fear of hypoglycaemic attacks.
Given these adaptations, the UK-specific nature of this work, the selection of this model by an independent research group, and the transparency of their report in terms of methods and results, why would a UK-based HTA organisation accept a different modelling approach when reviewing a future diabetes submission?
More to the point, why would a pharma company commission a different approach, incurring costs in the process? As an SMC reviewer myself, I am not an expert in modelling diabetes, breast cancer or RA so what I want is to be reassured that independent people who are cleverer and know more about this topic than I do are happy with the method – that’s by no means a perfect test but short of organising a validation exercise for different economic models, I don’t know of a better one. (Of course I can check to see that the model predicts the observed outcomes of the RCT used for licensing but since this often involves HbA1c changes over 6 months and the model is over a lifetime it’s unclear how useful a check this is.)
Of course, as a reviewer I would listen to the company’s view and if they had a good reason for not using the approach of Waugh et al I would consider what they had to say. There could be some innovation in the model, or some benefit from the new product that was not captured by the model, or perhaps new data supersedes the inputs previously used. However, my starting presumption would be that Waugh et al’s approach was the preferred one and each departure would have to be explained to me. The best place for companies to debate the merits of different models is through articles in published, peer-reviewed journals.
The bigger issue, of course, is whether HTA organisations should start publishing lists of preferred economics models and data inputs for companies planning submissions. This has the potential to save companies money and give consistency in results.
Footnote: for transparency, I am working on a paper on NICE and SMC decisions with Norman Waugh, and Ewen Cummins is a fellow reviewer for SMC.