Spatial model for risk prediction and sub-national prioritization to aid poliovirus eradication in Pakistan

Abstract

Background: Pakistan is one of only three countries where poliovirus circulation remains endemic. For the Pakistan Polio Eradication Program, identifying high risk districts is essential to target interventions and allocate limited resources.

Methods: Using a hierarchical Bayesian framework we developed a spatial Poisson hurdle model to jointly model the probability of one or more paralytic polio cases, and the number of cases that would be detected in the event of an outbreak. Rates of underimmunization, routine immunization, and population immunity, as well as seasonality and a history of cases were used to project future risk of cases.

Results: The expected number of cases in each district in a 6-month period was predicted using indicators from the previous 6-months and the estimated coefficients from the model. The model achieves an average of 90% predictive accuracy as measured by area under the receiver operating characteristic (ROC) curve, for the past 3 years of cases.

Conclusions: The risk of poliovirus has decreased dramatically in many of the key reservoir areas in Pakistan. The results of this model have been used to prioritize sub-national areas in Pakistan to receive additional immunization activities, additional monitoring, or other special interventions.

Keywords: Disease mapping; Hurdle models; Pakistan; Polio eradication; Risk mapping; Risk prioritization; Spatial epidemiology; Supplementary immunization activities; Vaccination campaigns.

Fig. 1

A map of smoothed estimates of zero routine immunization (RI) doses for January through June, 2016 (left) and an example of the smoothing models for observed zero RI rates for Khyber district in the Federally Administered Tribal Areas (FATA) of Pakistan from 2003 to 2016 (right)

Fig. 2

A map of smoothed estimates of underimmunized fraction for January through June, 2016 (left) and observed underimmunized fraction with smoothed estimates for Khyber district from 2003 to 2016 (right)

Fig. 3

A map of the vaccine-derived population immunity for type 1 poliovirus estimated by the dynamic immunity model as of June 30, 2016 (left), and the dynamic immunity traces and 95% credible interval for immunity in Khyber with supplementary immunization activity (SIA) calendar from 2003 to 2016 (right). The dashed marks along the horizontal axis show the timing of the SIAs and the color represents the vaccine used in the SIA, where mOPV1 represents monovalent oral polio vaccine (OPV) for serotype 1, bOPV represents bivalent OPV for serotypes 1 and 3, tOPV represents trivalent OPV (serotypes 1-3), and IPV is the inactivated poliovirus vaccine

Fig. 4

Area under the curve, the probability that a randomly selected district with a case will have a higher risk score than a randomly selected district without a case, for prediction of WPV1 cases by district as predicted based on model data from 2003 until 6 months prior to observed data (left) and sensitivity, or true positive rate, of a list containing the top 50 high risk districts for each time point (right)

Fig. 5

The probability of at least one case (left), expected number of cases given at least one case (center), and the overall risk score (expected number of cases) (right), for July through December, 2016

Fig. 6

The final risk tier classification to be included in the National Emergency Action Plan for polio eradication in Pakistan for 2016–2017 with Karachi in the inset. This final list incorporates modeling output, genetic sequencing results, and local knowledge of access and security