A spatial scan statistic for group testing data

Group testing protocols physically combine biomaterial from multiple individuals into a single 'pooled' specimen, which is then tested for markers of infection. Group testing offers a cost-effective approach for surveillance of low-prevalence infections because it greatly reduces the number of tests required relative to individual testing protocols. However, group testing poses challenges for detecting spatial clusters of disease because only pooled test results are observed. This article introduces a spatial scan statistic tailored to group testing data with variable pool sizes, using a likelihood ratio test to compare homogeneous and heterogeneous infection probability models across candidate zones. A simulation study is performed to evaluate the power and type I error rate of the spatial scan statistic test across multiple sample sizes, pooling strategies, and infection prevalences. The results demonstrate that geographically homogeneous pooling improves power relative to heterogeneous pooling. After validating the method, we apply the spatial scan statistic test to pooled testing data from ticks collected across South Carolina as part of spotted fever group Rickettsia surveillance. We detect a cluster of Rickettsia infection in Amblyomma americanum ticks along the southeast coast of South Carolina.