Fig 1.
Illustration of the two components of the polio surveillance system - AFP surveillance and environmental surveillance - and the sequence of events which must occur for a single infection to result in a reported detection of poliovirus.
Table 1.
Specification of event probabilities within each surveillance component. Each event must occur for a given infected individual to result in a reported detection of poliovirus. For all parameters except catchment of ES, a Beta distribution is fit to the given value and uncertainty interval from which random draws are used in the subsequent analysis. Uncertainty in the catchment population is explored through a sensitivity analysis of the assumed radius.
Fig 2.
Detections of wild poliovirus serotype 1 (WPV1) among cases of(A) AFP and (B) environmental surveillance (ES) samples. The total number of environmental samples collected is indicated by unfilled blue squares. (C) The recorded locations of four WPV1-positive cases of AFP detected in Borno state in July 2016, after 23 months of surveillance without detection (AFP cases are plotted randomly within LGA boundaries). Administrative boundaries are sourced from the Office for the Surveyor General of the Federation of Nigeria (OSGOF), Ehealth, United Nations Cartographic Section (UNCS) via the Humanitarian Data Exchange (https://data.humdata.org/dataset/cod-ab-nga), made available under the CC-BY-IGO license.
Fig 3.
Summary of observed AFP surveillance performance per LGA.
(A,B) The 12-month rolling rate of reported AFP per LGA for the period of Aug 2013-July 2016. LGAs are classified by the proportion of time in which the reported rate exceeds the target threshold of 2 cases per 100,000 under-15s. (C,D) The 12-month rolling percentage of adequate stool sample collection among AFP cases reported by each LGA, in this case classified with respect to the target of 85%. In all panels, performance is classified into three levels according to how consistently the WHO target threshold is met; completely (blue), mostly (green) and partially (red). LGAs are mapped by this classification in the two right-hand figures. Administrative boundaries are sourced from the Office for the Surveyor General of the Federation of Nigeria (OSGOF), Ehealth, United Nations Cartographic Section (UNCS) via the Humanitarian Data Exchange (https://data.humdata.org/dataset/cod-ab-nga), made available under the CC-BY-IGO license.
Fig 4.
Summary of observed ENV surveillance performance by LGA.
(A, B) The coverage of ES by LGA and 12-month-rolling coverage over time. (C, D) The 12-month rolling detection rate of enterovirus (EV) among samples collected per LGA, classified with respect to the target of 50%. As in Fig 3, performance in panel C is classified according to how consistently the WHO target threshold is met, here into four levels; completely (blue), mostly (green), partially (yellow) and not at all (red). LGAs are mapped by this classification in panel D. Administrative boundaries are sourced from the Office for the Surveyor General of the Federation of Nigeria (OSGOF), Ehealth, United Nations Cartographic Section (UNCS) via the Humanitarian Data Exchange (https://data.humdata.org/dataset/cod-ab-nga), made available under the CC-BY-IGO license.
Fig 5.
(A) Estimated sensitivity of AFP (orange) and ENV (blue) surveillance per month, for detecting infection at the specified design prevalence of 1 per 100,000 in any LGA. The box plots illustrate uncertainty across 1,000 draws for each probability in the scenario tree. (B) Inferred probability of freedom from infection for each accumulating month without detection of WPV1 through either AFP or ENV surveillance. Uncertainty is represented by 95% quantile intervals across 1,000 draws. Thresholds of 95% and 99% probability are illustrated with dashed lines. Results from the combined surveillance system are compared to those obtained by considering each component alone.
Fig 6.
Equivalent illustrations of estimated sensitivity and probability of freedom from infection as in Figure 5, for the subsequent period from 2016-2020.
The estimated timeline from Eichner and Dietz [1] for the risk of continued “silent” infections to fall below 5% after detection of the last clinical case is marked in (B), along with the official declaration of WPV elimination from Nigeria in August 2020.