Imported cases and minimum temperature drive dengue transmission in Guangzhou, China: evidence from ARIMAX model

• Published in: Epidemiology and infection Vol. 146; no. 10; pp. 1226 - 1235
• Main Authors: Jing, Q L, Cheng, Q, Marshall, J M, Hu, W B, Yang, Z C, Lu, J H
• Format: Journal Article
• Language: English
• Published: England Cambridge University Press 01-07-2018
• Subjects: Aedes; Animals; Aquatic insects; Arboviruses; Autoregressive models; China - epidemiology; Cities; Climate change; Communicable Diseases, Imported - epidemiology; Communicable Diseases, Imported - transmission; Dengue - epidemiology; Dengue - prevention & control; Dengue - transmission; Dengue fever; Disease control; Disease prevention; Disease transmission; Epidemics; Epidemiology; Ethics; Humans; Incidence; Infections; Infectious diseases; Medicine; Model testing; Models, Statistical; Mosquito Vectors; Multivariate Analysis; Original Paper; Outbreaks; Rain; Regression analysis; Risk factors; Sensitivity and Specificity; Surveillance; Temperature; Temperature effects; Time series; Tropical diseases; Vector-borne diseases; Viral diseases; China; Thailand; Australia; United States > US; epidemiology; local transmission; ARIMAX; Dengue fever
• ISSN: 0950-2688; 1469-4409
• DOI: 10.1017/s0950268818001176

Dengue is the fastest spreading mosquito-transmitted disease in the world. In China, Guangzhou City is believed to be the most important epicenter of dengue outbreaks although the transmission patterns are still poorly understood. We developed an autoregressive integrated moving average model incorporating external regressors to examine the association between the monthly number of locally acquired dengue infections and imported cases, mosquito densities, temperature and precipitation in Guangzhou. In multivariate analysis, imported cases and minimum temperature (both at lag 0) were both associated with the number of locally acquired infections (P < 0.05). This multivariate model performed best, featuring the lowest fitting root mean squared error (RMSE) (0.7520), AIC (393.7854) and test RMSE (0.6445), as well as the best effect in model validation for testing outbreak with a sensitivity of 1.0000, a specificity of 0.7368 and a consistency rate of 0.7917. Our findings suggest that imported cases and minimum temperature are two key determinants of dengue local transmission in Guangzhou. The modelling method can be used to predict dengue transmission in non-endemic countries and to inform dengue prevention and control strategies.