CDC's Advisory Committee on Immunization Practices (ACIP) provides annual recommendations for the prevention and control of influenza. The ACIP Influenza Work Group* meets monthly throughout the year to discuss newly published studies, review current guidelines, and consider potential revisions to the recommendations. As they review the annual recommendations for consideration of the full ACIP, members of the Work Group consider a variety of issues, including burden of influenza illness, vaccine efficacy and effectiveness, safety and coverage in groups recommended for vaccination, feasibility, cost-effectiveness, and anticipated vaccine supply. Work group members also request periodic updates on antiviral production, supply, safety, efficacy, and effectiveness from clinician researchers, regulatory agencies, public health epidemiologists, and manufacturers and review influenza surveillance and antiviral resistance data obtained from CDC's Influenza Division.
Published, peer-reviewed studies are the primary source of data used by ACIP in making recommendations for the prevention and control of influenza, but unpublished data that are relevant to issues under discussion also are considered. The best evidence for antiviral efficacy comes from randomized, controlled trials that assess laboratory-confirmed influenza virus infection as an outcome measure. However, randomized, placebo-controlled trials might be difficult to perform in populations for which antiviral treatment already is recommended. Observational studies that assess outcomes associated with laboratory-confirmed influenza virus infection can provide important antiviral effectiveness data but are more subject to biases and confounding that can affect validity and the size of effects measured. Randomized, placebo-controlled clinical trials are the best source of antiviral safety data for common adverse events; however, such studies do not have the power to identify rare but potentially serious adverse events. In cited studies that included statistical comparisons, a difference was considered to be statistically significant if the p-value was