Using spatiotemporal analysis to Identify high-risk areas of Dengue in Medellin, Colombia

Abstract

Background: Dengue is a vector-borne disease transmitted to humans by Aedes mosquitoes. Medellin is the second largest city in Colombia with more than 2.6 million inhabitants and its annual dengue incidence ranged 161-745 cases per 100,000 inhabitants during the last 10 years. Methods & Materials: In order to identify: i) districts and neighborhoods with the highest burden of the disease and ii) epidemic periods, we used routinely collected entomological and surveillance data from Medellin between 2008-2013. Using time-series analysis we determine dengue seasonality for year, month and epidemiological weeks for each district of the city. We fit Poisson and negative binomial regressions using a non-spatial approach (fixed effects model) and a spatiotemporal (Spatially-unstructured random effects model) Bayesian approach with INLA. Information was analyzed using R and maps were generated with ArcGIS software. Results: Overall, 20,477 cases were reported during the study period. From 249 neighborhoods, 44% (n = 109) reported cases every year. The lowest and highest incidences were observed in years 2008 and 2010, with 28.6 and 659.4 cases per 100,000 inhabitants, respectively. An increased number of cases was typically observed in April with a second peak occurring around September-October of every year. Two north-eastern and two western districts were identified as high dengue incidence by both spatial and non-spatial methods throughout the study period, consolidating dengue risk settings. Non-spatial approaches produced useful information but do not account for spatial autocorrelation. Conclusion: These findings provide epidemiological and spatiotemporal information of high-risk areas of dengue transmission, useful for decision-makers to address specific vector control strategies, and to help the preparedness of health services for upcoming outbreaks.

Locations

• International Journal of Infectious Diseases - View - PDF