As health care costs continue to rise and as new diagnostic and treatment alternatives become available, it is natural to ask which approaches work best. Comparative effectiveness research (CER) or patient-centered outcomes research aims to address this. Ideally, randomized trials could be conducted to formally test superiority or non-inferiority of alternative treatments. However, such trials are not always feasible or ethical, can be very large, of long duration, and costly, and it is not clear whose responsibility it should be to fund and conduct the trials. Many of these trials do not reflect real-world conditions of product use for either patients or health care providers. Interested groups are more often turning to observational data to address relative effectiveness questions. Unlike the primary analysis of a well-designed randomized trial, the operating characteristics of even a similarly well-designed observational study are not known. Randomized pragmatic trials with limited exclusion criteria, no blinding and observational follow-up can be an alternative but are still subject to biases. This session focuses on ways to quantify and improve the reliability and validity of observational evidence generated by CER.

Learning Objectives:
Describe the FDA Partnership in Applied Comparative Effectiveness Science (PACES) Initiative
Review the methodological challenges associated with comparative effectiveness analyses using observational research
Explain guidance for proper interpretation of the strength of evidence from such data based on quality research principles and in the context of accumulated evidence Recognize and design higher quality comparative effectiveness analyses using observational data.