165-2010: Predictions of Infection and Their Accuracy

Given the recent occurrence of swine flu, many predictions have been made concerning future occurrence. In particular, a strong Fall 2009 outbreak was predicted, motivating the development of a swine flu vaccine to be distributed prior to this predicted onslaught. This is reminiscent of a swine flu scare that occurred in 1976, again predicted as a pandemic. Approximately 25% of the population was vaccinated when there were approximately 200 cases of swine flu in a relatively restricted geographic area. As it turned out, there were some serious side effects from the vaccine, especially Guillain-Barre syndrome that ultimately halted the distribution of the vaccine. No prediction can be accurate if the basic model assumptions and probability of risk are not accurate. In particular, the initial rate of growth of the infection determines the predicted spread of the epidemic. We will demonstrate how data and SAS® can be used to make accurate predictions concerning infections and pandemics. In particular, we will discuss the best way to estimate the initial rate of infection growth.