Comparative effectiveness research methods: symposium overview and summary.

BACKGROUND: The Agency for Healthcare Research and Quality, in expanding its efforts to advance methods for comparative effectiveness research (CER), convened a second invitational symposium in June 2009 through its Developing Evidence to Inform Decisions About Effectiveness (DEcIDE) network. A distinguished Planning Committee set the agenda, reviewed more than 70 abstracts, and invited almost 30 speakers; the program emphasized issues of clinical heterogeneity and longitudinal outcomes, including patient-reported measures.
RESULTS: Conference papers covered a varied set of points about 3 major areas of CER: study design and data collection issues; statistical and analytic techniques; and applications to policy and clinical decision-making. Both primary data collection and analyses of databases (including electronic health records and distributed networks) are crucial for the CER agenda. Methods advances enable investigators to pay greater attention to important population subgroups, including persons of low literacy, non-English speaking patients, or the frail elderly. Both established and newer statistical techniques--eg, propensity scoring and various types of modeling, including Bayesian approaches--offer analysts improved ways to address issues such as confounding and measurement bias. Finally, some articles provided guidance for and examples of extending CER into newer realms, such as evaluations of medical devices or surgical procedures, and providing better information for decision-makers, clinicians, and patients.