Estimating Zika virus attack rates and risk of Zika virus-associated neurological complications in Colombian capital cities with a Bayesian model

Abstract

Zika virus (ZIKV) is a mosquito-borne pathogen that caused a major epidemic in the Americas in 2015–2017. Although the majority of ZIKV infections are asymptomatic, the virus has been associated with congenital birth defects and neurological complications (NC) in adults. We combined multiple data sources to improve estimates of ZIKV infection attack rates (IARs), reporting rates of Zika virus disease (ZVD) and the risk of ZIKV-associated NC for 28 capital cities in Colombia. ZVD surveillance data were combined with post-epidemic seroprevalence data and a dataset on ZIKV-associated NC in a Bayesian hierarchical model. We found substantial heterogeneity in ZIKV IARs across cities. The overall estimated ZIKV IAR across the 28 cities was 0.38 (95% CrI: 0.17–0.92). The estimated ZVD reporting rate was 0.013 (95% CrI: 0.004–0.024), and 0.51 (95% CrI: 0.17–0.92) cases of ZIKV-associated NC were estimated to be reported per 10 000 ZIKV infections. When we assumed the same ZIKV IAR across sex or age group, we found important spatial heterogeneities in ZVD reporting rates and the risk of being reported as a ZVD case with NC. Our results highlight how additional data sources can be used to overcome biases in surveillance data and estimate key epidemiological parameters.

Keywords

• arboviruses
• emerging infectious diseases
• reporting rates
• outbreak analysis
• Guillain–Barré syndrome
• mathematical modelling