Publications - Faculty - Recently Submitted Faculty Publications 2008
<- Back
Rogers - Annals of the Entomological Society of America
Reexamining the Pooled Sampling Approach for Estimating Prevalence of Infected Insect Vectors
T. A. Ebert, R. Brlansky and M. Rogers
Abstract.Our goal is to estimate seasonal changes in the proportion of Asian Citrus Psyllid (ACP) carrying Candidatus Liberibacter asiaticus (Las). Our approach is to test ACP using pooled samples. The initial question was about pool size and the consequences of choosing poorly. Assuming no loss in sensitivity when diluting one infected individual with many healthy individuals, then it is recommend that a combination of all the published limits be used: keep the number of pools (n) above 20, the pool size (k) below 100, and the number of infected pools less than half the total number of pools. The most conservative approach to achieving the latter is to optimize pool size given an infection rate (p) such that k=ln(0.5)/ln(1-p). The biggest risk in going outside these limits is an increased probability that all the pools will be infected. If this occurs, then that particular sample will be discarded, thereby wasting the resources utilized to collect and process that sample. Utilization of multiple pool sizes can be used to manage this risk, but this approach may not always be practical.
PooledInfRate is a good program for estimating prevalence, and it is available for free from the Center for Disease Control (CDC). The program provides corrected confidence intervals for prevalence estimates using one or multiple pool sizes. We utilized a randomization test approach as a contrasting methodology. The bias corrected CDC 95% confidence interval is an upper bound to the “true” 95% confidence interval. Within the range of our testing, the 95% confidence interval using the CDC method was at least 11% wider than necessary.
