Spatial and Genomic Data to Characterize Endemic Typhoid Transmission

Abstract

Background: Diverse environmental exposures and risk factors have been implicated in the transmission of Salmonella Typhi, but the dominant transmission pathways through the environment to susceptible humans remain unknown. Here, we use spatial, bacterial genomic, and hydrological data to refine our view of typhoid transmission in an endemic setting.

Methods: A total of 546 patients presenting to Queen Elizabeth Central Hospital in Blantyre, Malawi, with blood culture-confirmed typhoid fever between April 2015 and January 2017 were recruited to a cohort study. The households of a subset of these patients were geolocated, and 256 S. Typhi isolates were whole-genome sequenced. Pairwise single-nucleotide variant distances were incorporated into a geostatistical modeling framework using multidimensional scaling.

Results: Typhoid fever was not evenly distributed across Blantyre, with estimated minimum incidence ranging across the city from <15 to >100 cases per 100 000 population per year. Pairwise single-nucleotide variant distance and physical household distances were significantly correlated (P = .001). We evaluated the ability of river catchment to explain the spatial patterns of genomics observed, finding that it significantly improved the fit of the model (P = .003). We also found spatial correlation at a smaller spatial scale, of households living <192 m apart.

Conclusions: These findings reinforce the emerging view that hydrological systems play a key role in the transmission of typhoid fever. By combining genomic and spatial data, we show how multifaceted data can be used to identify high incidence areas, explain the connections between them, and inform targeted environmental surveillance, all of which will be critical to shape local and regional typhoid control strategies.

Keywords: Salmonella typhi; environmental transmission; genomics; spatial patterns; typhoid fever.

Figure 1.

Consort chart outlining the process of recruiting individuals to the study, reasons for exclusion, and data availability of geographic and genomic data. Abbreviation: ePAL, the electronic PArticipant Locator application.

Figure 2.

Estimated minimum annual incidence rate of typhoid fever for enumeration areas across the city of Blantyre, Malawi. The location of the recruiting hospital, Queen Elizabeth Central Hospital, is noted.

Figure 3.

A, Joint ancestral state reconstruction tree based on whole-genome single-nucleotide variant (SNV) phylogenetic analysis for the sequenced isolates of Salmonella Typhi, showing the major clades of isolates determined using a root-to-tip directional approach. B, Further resolution of variation provided by decomposition of SNV matrix into the first 2 principal coordinates (PCs) of the multidimensional scale; colors of points represent membership in major clades, corresponding to the tree. C, Empirical semivariogram (proportional to 1 − spatial correlation as a function of distance) of PC2 of the SNV decomposition.

Figure 4.

A. Major rivers of Blantyre with points indicating the approximate household locations of patients with typhoid fever. B, Approximate household locations are colored by genetic score, and river catchments are outlined using polygons. Catchments 2 and 8 are highlighted in yellow. Catchments not included in the analysis are in gray. Precise locations of households are masked by randomization.