Assessing the sensitivity of the polio environmental surveillance system

The polio environmental surveillance (ES) system has been an incredible tool for advancing polio eradication efforts because of its ability to highlight the spatial and temporal extent of poliovirus circulation. While ES often outperforms, or is more sensitive than AFP surveillance, the sensitivity of the ES system has not been well characterized. Fundamental uncertainty of ES site sensitivity makes it difficult to interpret results from ES, particularly negative results. To study ES sensitivity, we used data from Afghanistan and Pakistan to examine the probability that each ES site detected the Sabin 1, 2, or 3 components of the oral polio vaccine (OPV) as a function of virus prevalence within the same district (estimated from AFP data). Accounting for virus prevalence is essential for estimating site sensitivity because Sabin detection rates should vary with prevalence-high immediately after supplemental immunization activities (SIAs), but low in subsequent months. We found that most ES sites in Pakistan and Afghanistan are highly sensitive for detecting poliovirus relative to AFP surveillance in the same districts. For example, even when Sabin poliovirus is at low prevalence of ~0.5-3% in AFP surveillance, most ES sites have ~34-50% probability of detecting Sabin. However, there was considerable variation in ES site sensitivity and we flagged several sites for re-evaluation based on low sensitivity rankings and low wild polio virus detection rates. In these areas, adding new sites or modifying collection methods in current sites could improve sensitivity of environmental surveillance. Relating ES detections to virus prevalence significantly improved our ability to evaluate site sensitivity compared to evaluations based solely on ES detection rates. To extend our approach to new sites and regions, we provide a preliminary framework for relating ES and AFP detection rates, and descriptions of how detection rates might relate to SIAs and natural seasonality.