HTA agencies facing model biases: the case of type 2 diabetes

When evaluating new drugs or treatments eligible for reimbursement, health technology assessment (HTA) agencies are repeatedly faced with cost-effectiveness analyses that evidence lack of adequate data and modeling biases. The case of type 2 diabetes illustrates this difficulty. In spite of its high disease burden, type 2 diabetes is poorly documented through existing cost-effectiveness analyses. We support this statement by an exhaustive literature review that enables us to precisely pinpoint the limitations of models used for the assessment of newly marketed (and expensive) drugs. We find that models are mostly restricted to surrogate endpoints and based on non-inferiority clinical trial data; they also show biases in the choice of comparators and inclusion criteria. Such limitations undermine the scope and applicability of HTA practice guidelines based on cost-effectiveness evidence. Nevertheless, cost-effectiveness models remain an opportunity to better inform decision makers and to reduce the uncertainty surrounding their decisions. HTA agencies are best placed to provide incentives for companies to improve the quality of the cost-effectiveness studies submitted for pricing and reimbursement decisions. One such incentive is to include stages of discussion between the company and the health authority during the evaluation process.