There is a high risk of biased comparisons from observational (non-randomised) data

During medicine development, observational data are generally not available on effectiveness of the new therapy. They are however used for a number of purposes: to describe the natural history of disease, disease burden and treatment patterns, or the relationship between surrogate and final endpoints, or to validate new study endpoints or provide information on the effectiveness of comparator therapies. Some of this information may be used indirectly in estimating the effectiveness of the new therapy, for example through predictive modelling of long-term outcomes. Assessors will be particularly interested in the generalisability of the study population and the statistical methods used to control for bias.

In situations of conditional reimbursement, potentially within an adaptive pathway for a new medicine, observational data (for example, from registries) on the effectiveness of new medicine may be presented and assessed at HTA reviews. Comparisons between therapy alternatives of health outcomes based on non-randomised studies are particularly subject to bias: careful study design is required to minimise the bias. A variety of analytical techniques are available to adjust for imbalances observed between study groups that may affect the comparison, although this is unlikely to fully eliminate bias.