Identifying urban hotspots of dengue, chikungunya, and Zika transmission in Mexico to support risk stratification efforts: a spatial analysis

Abstract

Background: Effective Aedes aegypti control is limited, in part, by the difficulty in achieving sufficient intervention coverage. To maximise the effect of vector control, areas with persistently high numbers of Aedes-borne disease cases could be identified and prioritised for preventive interventions. We aimed to identify persistent Aedes-borne disease hotspots in cities across southern Mexico.

Methods: In this spatial analysis, geocoded cases of dengue, chikungunya, and Zika from nine endemic Mexican cities were aggregated at the census-tract level. We included cities that were located in southern Mexico (the arbovirus endemic region of Mexico), with a high burden of dengue cases (ie, more than 5000 cases reported during a 10-year period), and listed as high priority for the Mexican dengue control and prevention programme. The Getis-Ord Gi*(d) statistic was applied to yearly slices of the dataset to identify spatial hotspots of each disease in each city. We used Kendall's W coefficient to quantify the agreement in the distribution of each virus.

Findings: 128 507 dengue, 4752 chikungunya and 25 755 Zika clinical cases were reported between Jan 1, 2008, and Dec 31, 2016. All cities showed evidence of transmission heterogeneity, with a mean of 17·6% (SD 4·7) of their total area identified as persistent disease hotspots. Hotspots accounted for 25·6% (SD 9·7; range 12·8-43·0) of the population and 32·1% (10·5; 19·6-50·5) of all Aedes-borne disease cases reported. We found an overlap between hotspots of 61·7% for dengue and Zika and 53·3% for dengue and chikungunya. Dengue hotspots in 2008-16 were significantly associated with dengue hotspots detected during 2017-20 in five of the nine cities. Heads of vector control confirmed hotspot areas as problem zones for arbovirus transmission.

Interpretation: This study provides evidence of the overlap of Aedes-borne diseases within geographical hotspots and a methodological framework for the stratification of arbovirus transmission risk within urban areas, which can guide the implementation of surveillance and vector control.

Funding: USAID, the US Centers for Disease Control and Prevention, the Canadian Institutes of Health Research, International Development Research Centre, Fondo Mixto CONACyT (Mexico)-Gobierno del Estado de Yucatan, and the US National Institutes of Health.

Translation: For the Spanish translation of the abstract see Supplementary Materials section.

Figure 1

Distribution of census areas in Mexico Figure shows map of Mexico (light grey) showing the states located in southern Mexico (dark grey) and the location and distribution of census areas in the nine cities selected for this study.

Figure 2

Dengue hotspots in nine cities in Mexico, 2008–16 Colours indicate the number of years each census unit was identified as a statistically significant hotspot using the Getis-Ord Gi*(d) method. Grey areas indicate census units where no significance was detected.

Figure 3

Dengue and chikungunya transmission hotspots in nine endemic cities in Mexico Dengue historical hotspots for 2008–16 are shown in dark grey and red polygons indicate hotpots for chikungunya for 2015–16.

Figure 4

Dengue and Zika transmission hotspots in nine endemic cities in Mexico Dengue historical hotspots for 2008–16 are shown in dark grey and red polygons indicate hotpots for Zika for 2016.

Figure 5

Location of dengue hotspots during 2017–20 compared with the historical hotspot distribution of 2008–16 Results are from a logistic generalised linear mixed model. Values show odds ratios and 95% CIs for each city. Neither Campeche nor Coatzacoalcos had converging models because of their low number of cases (shown as 0).