Abstract
Value for money is fundamental to health insurance schemes given insurers must choose which treatments to fund. Assessing value for money ex ante is challenging, however, because costs and outcomes depend on how treatments are used. Estimates often rely on evidence from early randomized controlled trials conducted prior to regulatory approval, where provider and patient behaviors are tightly controlled. This approach ignores how different supply conditions and incentives in practice influence behaviors. This paper considers how provider and patient incentives can differ between trial and practice settings and analyses how healthcare use changed when new prostate cancer treatments were funded on the public health insurance scheme in Australia. We find evidence that doctors treated patients with worse prognosis compared to the trials, patients ceased prior treatment and switched to the new treatments earlier than expected, and treatment duration was longer than expected. These and other behavioral responses reduced value for money ex post. Our findings suggest that health insurers should carefully consider the supply conditions and incentives in practice when funding new treatments.