Successive epidemic waves of cholera in South Sudan, 2014 - 2017

Background Between 2014 and 2017, successive cholera epidemics occurred in South Sudan within the context of civil war, population displacement, flooding, and drought. Understanding the determinants of cholera spread in complex settings like this can provide valuable insights for mitigating future cholera risk. Methods We analyzed cholera linelist and molecular data to describe the spatio-temporal progression of the epidemics. We explored the role of rainfall, population movement and vaccination campaigns in shaping the explaining incidence and the spatial distribution of reported cases. Findings South Sudan experienced three distinct cholera epidemic waves of cholera ranging from 6-18 months with more than 28,000 cases reported and more than 2 million cholera vaccine doses delivered to curb transmission. The 2014 and 2015 epidemics remained spatially limited while the 2016/17 epidemic exploded along the Nile river. Initial cases of each epidemic were reported in or around Juba soon after the start of the rainy season, but we found no evidence that rainfall modulated transmission during each epidemic. All isolates analyzed had similar genotypic and phenotypic characteristics, closely related to sequences from Uganda and Democratic Republic of Congo. The direction of large-scale population movements between counties with cholera outbreaks was consistent with the spatial distribution of outbreaks. As of September 2020, zero cholera cases have been confirmed within South Sudan since 2017. Interpretation The three epidemic waves were caused by V. cholerae of the same clonal origin despite the periods of no reported cases between waves. While the complex emergency likely shaped some of the observed spatial and temporal patterns of reported cases, the full scope of transmission determinants remains unclear. Timely and well targeted use of cholera vaccine can avert cases and deaths, however, most of the vaccine campaigns occurred after the epidemic peak highlighting the challenges of delivering vaccines quickly in response to an outbreak in settings like South Sudan. These analyses provide a multi-faceted template for examining cholera dynamics through epidemiological, microbiological, climatic, and behavioral lenses. Funding The Bill and Melinda Gates Foundation

We analyzed cholera linelist and molecular data to describe the spatio-temporal progression of 31 the epidemics. We explored the role of rainfall, population movement and vaccination 32 campaigns in shaping the explaining incidence and the spatial distribution of reported cases. CC-BY 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 October 13, 2020.  CC-BY 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint

Data Analysis 203
We calculated the weekly number of cases, attack rate (AR), case-fatality ratio (CFR), the 204 proportion of cases <5 years old, and proportion male. For most cases, we relied on the self-205 reported date of symptom onset for each case; when this was not available, we used the date 206 of health facility visit. Administrative vaccination coverage was calculated by dividing the total 207 number of first round doses delivered by the estimated target population size. To explore the 208 potential effect of vaccination campaigns, we compared the attack rates of outbreaks where 209 vaccine was used before the epidemic peak with those where it was used after the peak with 210 Poisson regression models using the population size as an offset. 211

212
We generated a maximum likelihood tree from the 1210 publicly available V. cholerae O1 213 whole genomes. 14-16 This tree closely resembles the phylogeny of 1203 genomes from Weill et 214 al, 14 but includes 7 genomes from Uganda published separately. 16 GenBank accession 215 numbers for all 1210 sequences are available in the supplement. 17 For each reference-based 216 assembly (reference accession:AE003852/AE003853), we masked recombinant sites manually 217 (https://figshare.com/s/d6c1c6f02eac0c9c871e) and using Gubbins v2.3.4 18 as previously 218 described. 15 We generated the tree from the `filtered_polymorphic_sites` FASTA outputted by 219 Gubbins, which contained only the 10,098 variant sites in our alignment. We used IQ-TREE 220 v1.6.10 19 with a GTR substitution model and 1000 bootstrap iterations. 20 We rooted the tree on 221 A6 by first constructing a tree containing an outgroup sequence 222 (M66,accession:CP001233/CP001234) and choosing the most ancestral non-outgroup strain. 223 We visualized phylogenetic trees using FigTree v1.4.4 224 (http://tree.bio.ed.ac.uk/software/figtree/). For MLVA analysis, we defined a clonal complex as 225 a group of isolates where each differs (in loci-specific copy number) from at least one other 226 isolate in the complex at no more than one locus. 227 228 . CC-BY 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint We used a distributed lag non-linear model 21-23 to investigate how accumulated rainfall over the 229 previous seven days (AR7D) is associated with changes in cholera transmission in Juba 230 County, as characterized by the instantaneous basic reproductive number, R 0 (t). We first jointly 231 estimated the serial interval and reproductive number for the period of epidemic growth for 232 each wave using methods described by White and Pagano. 24 R 0 (t) was calculated using 233 previously described methods 25,26 assuming a shifted gamma distributed serial interval with a 234 mean of 3.9 days, a standard deviation of 5.8 days based on the estimates of the 2015 wave, 235 which was largely confined to Juba County. We assumed that infection leads to complete 236 immunity over the study period and that 10% of infections were reported as medically attended 237 cholera cases 27 . We fit a quasipoisson model with penalized splines transforming AR7D and 238 lags up to 10 days. The results from this model provide an estimate of the relative R 0 (t) over 10 239 days following a day with a given value of AR7D compared to an AR7D of 0mm. CC-BY 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint After three years without cases, the MoH confirmed a cholera case in Juba County (Central 256 Equatoria State) on April 28, 2014 soon after the start of the rainy season. Case reports 257 increased in Juba County and in the neighboring Eastern Equatoria State (Figures 1 and 2), 258 where CFRs reached as high as 5% (Table 1)  Following a period of 34 weeks with no suspected cases, South Sudan began to experience its 281 largest reported cholera outbreak in recent history, with 20,438 suspected cases (AR=0.18%) 282 . CC-BY 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 October 13, 2020.  (Table S1) sequences. Across these fourteen sequences, there are 21 polymorphic sites (~3% of the 334 diversity observed in the T10 lineage), with no more than 10 SNPs between any two 335 sequences (Figure 3). These sequences all belonged to the T10 lineage, which has circulated 336 . CC-BY 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint in the Democratic Republic of Congo for more than twenty years ( Figure S1 (Table S2). In 2017, 363 . CC-BY 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 October 13, 2020.  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 October 13, 2020. Cholera control efforts during these outbreaks was continually hampered by conflicts and 431 limited access to areas with ongoing transmission. Of the 2,068,622 doses used reactively, 432 only 6.4% were part of campaigns initiated before the peak of the epidemic in the target county. 433 Due to the global shortage of vaccine supply, campaigns were often limited in size and rarely 434 covered the full population; only those deemed at the highest risk. Given the low coverage at 435 the county-level (though relatively high coverage in the target population), the spatial scale of 436 the most reliable epidemiologic data collected, and the relatively low testing rate among 437 suspected cases, the impact of vaccination on reducing incidence is unclear. Our simple 438 analyses exploring the relative attack rates in places that had OCV campaigns before and after 439 the peak provide hints of impact on the population-level but efforts to collect more spatially 440 resolved incidence data combined with systematic lab testing of suspected cholera cases will 441 allow for more precise quantification of vaccine impact. 442 443 . CC-BY 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 October 13, 2020. CC-BY 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 October 13, 2020. 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 October 13, 2020. CC-BY 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 October 13, 2020. 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 October 13, 2020. CC-BY 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 October 13, 2020. 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint 607 . CC-BY 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint CC-BY 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 October 13, 2020. CC-BY 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 October 13, 2020. . https://doi.org/10.1101/2020.10.09.20209262 doi: medRxiv preprint