A Potential Outcomes Approach to Documenting the Public Health Impact of the Introduction of PCV13 for the Prevention of Invasive Pneumococcal Disease

• Activity Number: 155
• Type: Topic Contributed
• Date/Time: Monday, August 4, 2014 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Elizabeth Zell*+
• Companies: Stat-Epi Associates
• Keywords: causal inference ; potential outcomes ; counterfactual ; multiple imputation

Abstract

Recommendations are often made to improve the health, education, well-being, etc. of individuals or communities. Estimating the impacts of such recommendations is important. Often costs are associated with their implementation and cost-benefit analyses should be considered. After a new recommendation has been implemented is NOT the time to "look for" baseline data from the pre-implementation period. Planning in advance is important to the evaluation process. Here we illustrate a potential outcomes approach to estimate IPD (Invasive Pneumococcal Disease) outcomes in the counterfactual world in which a new vaccine, PCV13, had not been introduced. This effort was possible because the CDC had baseline active, population-based IPD surveillance data prior to the introduction of PCV13. This approach allowed the assessment of disease reduction attributed to the new vaccine for those recommend for vaccination (direct effect) and for those not recommended for vaccination (indirect effect). Comparing the PCV13 rates in counterfactual world, estimated using time series models, to the rates that actually occurred estimates the number of IPD cases prevented due to the introduction of the vaccine