No abstract

Despite the availability of vaccines, every year 40,000 individuals die due to the direct effect of Invasive Pneumococcal Disease (IPD) or its complications. IPD has been associated with 100,000-135,000 hospitalizations for pneumonia, 57,000 cases of bacteremia, and 300 cases of meningitis in the United States every year. Little is known about IPD epidemic patterns beyond annual seasonality, and this lack of understanding has limited the ability to detect early changes in IPD epidemiology that may lead to large outbreaks. To mitigate this gap in understanding, a retrospective cohort design study was conducted using a population-based cohort from the National Hospital Discharge Survey for the period 1979-2006. This study set out to determine whether invasive infection by S. pneumoniae in the United States occurs in an epidemic pattern of a predictable recurrent nature and definable frequency. The theoretical basis for the study was drawn from the dynamic modeling of stochastic epidemic systems, and the analysis utilized time-series methods to examine the data. These analyses lead to the finding that IPD epidemics demonstrate a chaotic dynamic and a discrete, non-Markov process; that is, there is no predictable pattern to epidemics of IPD. The results of this study, that recurrent events consistent with periodic epidemics could not be identified, provide support for the current method of IPD surveillance and existing models of IPD dynamics. The present practice of mass vaccination by risk group, as opposed to vaccination for a predicted outbreak, is supported by the results of this study. These evidence-supported interventions will yield significant reductions in the morbidity and mortality associated with IPD, and the positive social change that results from improved health outcomes, reductions in suffering, and decreased health care costs.