To attain the anticipated benefits of increased investments in comparative effectiveness research, the results must translate into improved medical practice and policy. In this article we present an analysis of a case study of percutaneous vertebroplasty, a widely used invasive procedure to treat painful vertebral fractures by injecting bone cement into the spine. In August 2009, results from a pair of rigorous double-blind randomized controlled trials were published and reported that vertebroplasty provided no better pain relief than a sham procedure in which needles were introduced into the back without injecting cement. More than two years after publication of the two studies, insurers' coverage of the procedure continues unchanged. This raises serious questions about the policy mechanisms that exist in the United States to interpret and act upon "negative" research findings from studies of popular health care interventions.