Epidemiology of the 2014-2017 zika and chikungunya epidemics in Colombia

The introduction of Zika virus (ZIKV) and chikungunya virus (CHIKV) into the Americas in the mid-2010s triggered large epidemics across the region fueled by high population-level susceptibility. Colombia was one of the most affected countries, reporting approximately 413,000 cases of chikungunya fever (CF) and 106,000 cases of Zika virus disease (ZVD) in only three years. In this thesis, mathematical models were used to estimate key epidemiological parameters from cases of CF and ZVD that were reported to Colombia’s national population-based surveillance system on a weekly basis from 2014-2017. Both epidemics spread widely throughout the country. Out of 32 departments, all reported ZVD cases, while 31 departments reported CF cases. Females of child-bearing age comprised a large proportion of reported ZVD cases, most likely due to increased reporting linked to the risk of congenital birth defects associated with ZIKV infection during pregnancy. Of the 418 reported cases of ZIKV-associated neurological complications in the country, most were diagnosed with Guillain-Barré syndrome. The estimated reporting rate of CHIKV was higher than that of ZIKV according to models based on the renewal equation. This result was expected due to the higher rate of asymptomatic ZIKV infection compared to that of CHIKV. Basic reproduction number estimates at the department level were similar for CHIKV compared to ZIKV. Reassuringly, estimates of the time-varying reproduction number from the parametric model were in good agreement with those obtained from the software EpiEstim. ZIKV infection attack rates, reporting rates of ZVD, and the risk of ZIKV-associated neurological complications were estimated for 28 Colombian capital cities by incorporating multiple data types into a Bayesian hierarchical model. ZIKV infection attack rates varied considerably across cities. The overall estimated reporting rate for ZVD was similar to that estimated previously. Results also showed a low estimated risk of ZIKV-associated neurological complications. Important differences in estimated ZVD reporting rates and the risk of ZIKV-associated neurological complications between sex and age group were found for some cities. Finally, gravity models, Stouffer’s rank models, and radiation models were used to investigate the spatial and temporal invasion dynamics of CHIKV and ZIKV. Both geographic distance and travel time between cities were evaluated. Invasion risk was best captured by a gravity model which accounted for geographic distance and intermediate levels of density dependence. Results also showed that Stouffer’s rank model with geographic distance performed well. Short-distance transmission played an important role in spatial spread, and a few long-distance transmission events were identified. Jointly fitted models highlighted similarities between the epidemics. However, ZIKV spread faster than CHIKV. With new interventions on the horizon, including vaccines and novel methods of vector control, as well as the emergence of new arboviral diseases, having robust estimates of epidemiological parameters will be important for informing surveillance and preparedness for future epidemics.Open Acces