Development of a probabilistic model to determine efficacy of WNV control activities

Predictive probabilistic models were examined to evaluate the success of the mosquito control efforts in the City of New York. The model developed as part of this study integrated multiple complex interrelated parameters so as to improve the accuracy of the prediction. In this study, Relational Probabilistic Modeling (RPM) or Relational Bayesian Modeling (RBM) provided the basis for assessing the impact of New York City's control activities from mosquito surveillance and control data collected from the field. RBM techniques were used to evaluate observational mosquito surveillance and control data and allowed for the integration of complex environmental and ecological parameters and combination of parameters which traditionally had been treated as separate domain models for mosquitoes, primary hosts (i.e., birds) and humans.; Study findings showed a clear positive correlation between areas treated with pesticides and a reduction in positive dead birds in the same areas and identified quantitative thresholds that correlated with increased risk to humans. These correlations were particularly strong in those spray zones that had demonstrated the most effective mosquito control in post application surveillance. The study also found a positive correlation between dead birds and standing water reports and found that positive dead songbirds were a better indicator of WNV activity in an area and more accurate predictor of WNV cases in humans than were crows, which have heretofore received the most-attention regarding host surveillance.; The findings in this study suggest that the mosquito control efforts in New York City were successful at reducing disease transmission within avian species and were likely to have disrupted the amplification cycle of the WNV in the treated areas. The findings point to ways in which current host surveillance methodologies may be improved so that effective, targeted and timelier control measures can be implemented. They support continued application of mosquito control efforts in areas where surveillance should increase risk of WNV transmission to humans.