OBJECTIVES: Patient randomization within clinical trials ensures the generalizability of results to the entire patient population. Trial results are, therefore, based on the outcomes achieved by the average patient. However, once a therapy is marketed, physicians are more likely to recommend that the patients performing well with the current therapy remain on it, and worse performing patients switch to an alternative. Informed switching therefore introduces a selection bias that may positively affect the average treatment effect of therapies. If only average treatment effects are considered in the decision-making process, economic evaluations may systematically undervalue new therapies. This study aims to explore the implication of selection biases on economic evaluation results.

METHODS: In the context of a cost-effectiveness analysis, a treatment’s clinical efficacy was varied to account for selection bias. The incremental cost-effectiveness ratio (ICER) was estimated based on the comparator’s mean efficacy and a range of lower efficacy values, to reflect informed switching. The impact of informed switching was measured as the difference between the ICERs.

RESULTS: When accounting for informed switching the ICER decreases compared to a randomised selection process. The degree of selection bias was tested over a range of efficacy values and the ICERs were plotted to show the impact over a range of possible clinical benefits.

CONCLUSIONS: Physicians play a critical role in the decision of switching a patient’s treatment. If economic evaluations do not account for this, new treatments entering the market may be undervalued. The health gains for patients switching treatments rise, as they had a below average prognosis on the comparator, and the average treatment effect of those remaining on the comparator will also rise. Informed selection improves the average health outcomes for all treatments when new therapies joins the market by ensuring patients are allocated to the best performing treatment for them.