key: cord-0282712-ocpib3mj authors: Sikorski, M. J.; Ma, J.; Hazen, T. H.; Desai, S. N.; Tupua, S.; Nimarota-Brown, S.; Sialeipata, M.; Rambocus, S.; Ballard, S. A.; Valcanis, M.; Thomsen, R.; Robins-Browne, R. M.; Howden, B.; Naseri, T. K.; Levine, M. M.; Rasko, D. A. title: Spatial-temporal and phylogenetic analyses of epidemiologic data to help understand the modes of transmission of endemic typhoid fever in Samoa date: 2022-04-01 journal: nan DOI: 10.1101/2022.03.28.22272797 sha: 17746842a433fafa1f4f5cb79f6f8dd62ee2d4f9 doc_id: 282712 cord_uid: ocpib3mj Salmonella enterica serovar Typhi ( S. Typhi) is either widely distributed or proximally transmitted via fecally-contaminated food or water to cause typhoid fever. In Samoa, where endemic typhoid fever has persisted over decades despite water quality and sanitation improvements, the local patterns of S. Typhi circulation remain undistinguished. From April 2018-June 2020, epidemiologic data and GPS coordinates were collected during household investigations of 260 acute cases of typhoid fever, and 27 asymptomatic shedders of S. Typhi were detected among household contacts. Spatial and temporal distributions of cases were examined using Average Nearest Neighbor and space-time hotspot analyses. In rural regions, infections occurred in sporadic, focal clusters contrasting with persistent, less clustered cases in the Apia Urban Area. Restrictions to population movement during nationwide lockdowns in 2019-2020 were associated with marked reductions of cases. Phylogenetic analyses of isolates with whole genome sequences (n=186) revealed one dominant genotype 3.5.4 (n=181/186) that contains three Samoa-exclusive sub-lineages: 3.5.4.1, 3.5.4.2, and 3.5.4.3. Variables of patient sex, age, and geographic region were examined by phylogenetic groupings, and significant differences (p<0.05) associated genetically-similar isolates in urban areas with working ages (20-49 year olds), and in rural areas with age groups typically at home (<5, 50+). Isolates from asymptomatic shedders were among all three sub-lineages. Whole genome sequencing also corroborated bacterial genetic similarity in 10/12 putative epidemiologic linkages among cases and asymptomatic shedders as well as 3/3 repeat positives (presumed relapses), with a median of one single nucleotide polymorphism difference. These findings highlight various patterns of typhoid transmission in Samoa that differ between urban and rural regions as well as genomic subtypes. Asymptomatic shedders, detectable only through household investigations, are likely an important reservoir and mobile agent of infection. This study advances a "Samoan S. Typhi framework" that supports current and future typhoid surveillance and control efforts in Samoa. Typhoid fever is a potentially fatal human host-restricted infection caused by ingestion 75 of Salmonella enterica serovar Typhi (S. Typhi) via contaminated food or water. Typhoid 76 incidence, reported as cases per 100,000 people per year, is stratified into low (<10), 77 moderate (10-100), high (>100-<500), and very high (≥500), with different transmission 78 mechanisms capable of maintaining typhoid fever at different incidence levels [1] . 79 Widespread distribution of fecally-contaminated indirect sources (called "vehicles"), 80 such as water supplies [2,3], crops irrigated with raw sewage [4, 5] , or processed foods 81 [6, 7] , results in amplified, high and very high incidence transmission of S. Typhi. 82 Sporadic, focal outbreaks of typhoid fever are typically associated with asymptomatic 83 shedding of S. Typhi from a temporary (weeks to months) [8] or chronic (lifelong) 84 carrier, such as "Typhoid Mary" [9] . These asymptomatic shedders inadvertently 85 contaminate food that is consumed by contacts in close proximity, such as household 86 members or attendees at gatherings (e.g. picnics and celebrations) [10] . 87 Samoa, an island nation in the Polynesian sub-region of Oceania (the continental 88 grouping of thousands of islands comprising 14 countries and 9 dependencies in the 89 Central and South Pacific Ocean), has endured persistent endemic typhoid fever for 90 decades despite improvements in drinking water quality, sanitation infrastructure, and 91 economic development [11] [12] [13] [14] . We have previously described endemic typhoid in 92 Samoa by person, place, and time using blood culture surveillance data from 2008-2019 93 to what extent endemic typhoid in Samoa is sustained by widespread, continuous 96 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. 126 This study took place from 2018-2020 in Samoa, a Pacific Island nation of area 2,842 127 km 2 (1,097 sq. mi) and ~200,000 population. Samoa comprises two populated islands, 128 Upolu and Savaii, each with a central clinical laboratory and several peripheral health 129 facilities (Fig 1) . Typhoid fever is a notifiable infectious disease in Samoa that triggers 130 epidemiologic investigation of any and all culture-confirmed cases [19, 20] . Since 2018, 131 the STFCP has strengthened epidemiologic surveillance for typhoid fever, improved 132 clinical and environmental microbiology laboratory capacity, and initiated WGS at a 133 collaborating regional reference laboratory in Melbourne, Australia. Typhoid Fever 134 Epidemiologic "SWAT" Teams (equipped with specialized epidemiologic tools and 135 tactics), one on each island, were trained to perform epidemiologic investigations of the 136 household (and/or workplace or school) of every bacteriologically confirmed case of 137 typhoid fever occurring anywhere in Samoa. 138 139 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. nearest neighbor ratio (R n ) ranged from zero (significant clustering) to 2.15 (uniform 191 equidistant dispersion) and was supported by critical values (Z-score and p-value); an 192 R n equal to 1 indicated randomness or no detectable spatial pattern (S1 and S2 Figs). 193 The ANN analysis was performed on household coordinates in aggregate and then by 194 each of the four census regions (Fig 1, S3A Fig) is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 1, 2022. ; CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted April 1, 2022. Characteristics of the dataset 244 From April 27, 2018 to June 9, 2020, the STFCP identified 260 blood culture-confirmed 245 acute typhoid fever cases through a strengthened surveillance system and detected 27 246 asymptomatic S. Typhi fecal shedders through household investigations. We examined 247 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. ; case household coordinates for spatial and spatial-temporal clustering, and then 248 incorporated phylogenetic analyses of the sequenced subset of 186 Samoan S. Typhi 249 isolates -172 from acute cases and 14 from asymptomatic shedders -among 250 household contacts (S1 Table) . Among the 186 whole genome sequenced isolates, 251 three S. Typhi genotypes were identified: genotype 3.5.4 (181/186, 97.3%), genotype 252 4.1 (4/186, 2.2%), and genotype 3.5.3 (1/186; 0.5%). All but one genotype 3.5.4 isolates 253 could be further categorized into three distinct sub-lineages denoted 3.5.4.1 (n=49), 254 3.5.4.2 (n=51), and 3.5.4.3 (n=80) and differentiated from all other global S. Typhi 255 genomes by single canonical SNPs (S1 Table) , as previously described [15] . The 256 sequenced subset of acute cases showed no significant difference in patient 257 characteristics (patient sex, mean age, or age group distribution) when compared to all 258 reported cases (Table 1) . 259 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. ; Spatial clustering and spatial-temporal hotspot analyses of 261 acute cases 262 When acute case households (n=260) were mapped, areas with many typhoid cases 263 (higher burden) -Apia, northwestern Upolu, southeastern Savaii, and Upolu's southern 264 coast -were prominent among the four census regions (arrows in Fig 2A) . We used the 265 Average Nearest Neighbor (ANN) and Space-Time Hotspot analyses in ArcGIS Pro 266 v2.9 to assess spatial and spatial-temporal clustering, respectively. As expected for a 267 human host-restricted communicable disease, case households (n=260) demonstrated 268 significant clustering by ANN analysis (R n <1, p<0.005). However, when grouped by 269 census region (S3A Fig, Table 2 ), local differences emerged in the degree of clustering, 270 i.e., R n closer to 0, indicating more tightly grouped clusters. Case households in the 271 AUA (n=42), where population density is greatest and where residents receive 272 putatively treated drinking water through a reticulated system, were the least clustered 273 of the census regions (R n =0.62). In contrast, case households in the rural ROU (n=55), 274 where districts each have independent (often untreated) water schemes, were the most 275 clustered (R n =0.25) ( Table 2) . Notably, SAV (n=53), which is more rural than ROU, 276 demonstrated a similar degree of clustering (R n =0.60) to AUA (Table 2) (1 km 2 x 2-months) (A) Typhoid fever burden and genotype distribution in Samoa, April 2018 through June 2020 (C) Method and legend . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. p<0.005 D ANN is a statistical comparison of the observed mean (Ō) and expected mean (Ē) distances between points yields a nearest neighbor ratio (R n ) further described by a probability (p-value) that the observed spatial pattern was created by a random process and standard deviations (z-score). The R n ranges from 0 (indicating discrete clustering of points) to ~2.15 (indicating uniform dispersion of points equidistant from one another). An R n value of 1 indicates complete spatial randomness. a Only household coordinates of acute cases were included to assess circulating S. Typhi detected through routine passive surveillance and to avoid biasing the point-pattern analyses to households where asymptomatic shedders were localized by active investigation. b All z-scores are small (<-1.96) corresponding to a confidence level above 95%. c All p-values are small (<0.005) which indicate statistical significance. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. ; To assess for temporal clustering by census region, a Space-Time Hotspot 306 analysis was performed using a window of 1 km 2 x 2-months to bin case household 307 coordinates (n=260) and visualized in three-dimensions at two altitudes, 20,000 meters 308 and 10,000 meters, to aid in visualization of each time series (Fig 2B) . Temporal 309 persistence of typhoid burden was detectable in AUA as sequential spatial-temporal 310 hotspots, visible as contiguous red columns, (p<0.01) associated with that geographic 311 region (Fig 2B-C) . This was in contrast with sporadic (short-lived) hotspots of infection 312 evident in the rest of the country as identified by the discrete red bands (p<0.01) 313 between grey bars in those geographic regions (Fig 2B-C) . (Table 3 , Fig 3A) . Sub-327 lineage 3.5.4.2 was significantly (p<0.05) associated with the patient's sex, being 328 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. ; https://doi.org/10.1101/2022.03.28.22272797 doi: medRxiv preprint isolated from males more frequently than females (Table 3 , Fig 3A) . Sub-lineage 3.5.4.3 329 was significantly (p<0.05) associated with 20-49 year-olds and LTEF in the 5-19 and 330 50+ year old age groups, as well as census region (common to the four census regions 331 but GTEF in the AUA and ROU) (Table 3, Fig 3A) . 332 333 . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. whole genome sequences of Samoan S. Typhi genotype 3.5.4 isolates from April 27, 2018 through June 9, 2020 (N=186, 337 S1 Table) . Core genome single nucleotide polymorphisms (SNPs) were identified relative to the 2012 Samoa Reference 338 genome H12ESR00394-001A (GCA_001118185.2). Genotype/sub-lineage is indicated by isolate label shading. The 339 labels of the three genotype 3.5.4 sub-lineages 1-3 are identified with orange, yellow, and blue label shading, respectively. 340 . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. . It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. The copyright holder for this preprint this version posted April 1, 2022. Geospatially, case households associated with each genotype/sub-lineage were 350 found in the four census regions (Table 3 ) and were overlapping (Fig 2A) , indicating 351 widespread circulation among the islands. Spatial clustering of each 3.5.4 sub-lineage 352 was significant (p<0.005) and intermediate (0.4