Background Typhoid fever remains a major cause of morbidity and mortality in low- and middle-income settings. In the last 10 years, several reports have described the reemergence of typhoid fever in southern and eastern Africa, associated with multidrug-resistant H58 Salmonella Typhi. Here, we identify risk factors for pediatric typhoid fever in a large epidemic in Blantyre, Malawi. Methods A case-control study was conducted between April 2015 and November 2016. Cases were recruited at a large teaching hospital, and controls were recruited from the community, matched by residential ward. Stepwise variable selection and likelihood ratio testing were used to select candidate risk factors for a final logistic regression model. Results Use of river water for cooking and cleaning was highly associated with risk of typhoid fever (odds ratio [OR], 4.6 [95% confidence interval {CI}, 1.7–12.5]). Additional risk factors included protective effects of soap in the household (OR, 0.6 [95% CI, .4–.98]) and >1 water source used in the previous 3 weeks (OR, 3.2 [95% CI, 1.6–6.2]). Attendance at school or other daycare was also identified as a risk factor (OR, 2.7 [95% CI, 1.4–5.3]) and was associated with the highest attributable risk (51.3%). Conclusions These results highlight diverse risk factors for typhoid fever in Malawi, with implications for control in addition to the provision of safe drinking water. There is an urgent need to improve our understanding of transmission pathways of typhoid fever, both to develop tools for detecting S. Typhi in the environment and to inform water, sanitation, and hygiene interventions.

Typhoid fever is endemic across sub-Saharan Africa. However, estimates of the burden of typhoid are undermined by insufficient blood volumes and lack of sensitivity of blood culture. Here, we aimed to address this limitation by exploiting pre-enrichment culture followed by PCR, alongside routine blood culture to improve typhoid case detection. We carried out a prospective diagnostic cohort study and enrolled children (aged 0–4 years) with non-specific febrile disease admitted to a tertiary hospital in Blantyre, Malawi from August 2014 to July 2016. Blood was collected for culture (BC) and real-time PCR after a pre-enrichment culture in tryptone soy broth and ox-bile. DNA was subjected to PCR for invA (Pan- Salmonella ), staG ( S . Typhi), and fliC ( S . Typhimurium) genes. A positive PCR was defined as invA plus either staG or fliC (CT<29). IgM and IgG ELISA against four S . Typhi antigens was also performed. In total, 643 children (median age 1.3 years) with nonspecific febrile disease were enrolled; 31 (4.8%) were BC positive for Salmonella (n = 13 S . Typhi, n = 16 S . Typhimurium, and n = 2 S . Enteritidis). Pre-enrichment culture of blood followed by PCR identified a further 8 S . Typhi and 15 S . Typhimurium positive children. IgM and IgG titres to the S. Typhi antigen STY1498 (haemolysin) were significantly higher in children that were PCR positive but blood culture negative compared to febrile children with all other non-typhoid illnesses. The addition of pre-enrichment culture and PCR increased the case ascertainment of invasive Salmonella disease in children by 62–94%. These data support recent burden estimates that highlight the insensitivity of blood cultures and support the targeting of pre-school children for typhoid vaccine prevention in Africa. Blood culture with real-time PCR following pre-enrichment should be used to further refine estimates of vaccine effectiveness in typhoid vaccine trials.

Background Typhoid fever remains a major source of morbidity and mortality in low-income settings. Its most feared complication is intestinal perforation. However, due to the paucity of diagnostic facilities in typhoid-endemic settings, including microbiology, histopathology, and radiology, the etiology of intestinal perforation is frequently assumed but rarely confirmed. This poses a challenge for accurately estimating burden of disease. Methods We recruited a prospective cohort of patients with confirmed intestinal perforation in 2016 and performed enhanced microbiological investigations (blood and tissue culture, plus tissue polymerase chain reaction [PCR] for Salmonella Typhi). In addition, we used a Poisson generalized linear model to estimate excess perforations attributed to the typhoid epidemic, using temporal trends in S. Typhi bloodstream infection and perforated abdominal viscus at Queen Elizabeth Central Hospital from 2008–2017. Results We recruited 23 patients with intraoperative findings consistent with intestinal perforation. 50% (11/22) of patients recruited were culture or PCR positive for S. Typhi. Case fatality rate from typhoid-associated intestinal perforation was substantial at 18% (2/11). Our statistical model estimates that culture-confirmed cases of typhoid fever lead to an excess of 0.046 perforations per clinical typhoid fever case (95% CI, .03–.06). We therefore estimate that typhoid fever accounts for 43% of all bowel perforation during the period of enhanced surveillance. Conclusions The morbidity and mortality associated with typhoid abdominal perforations are high. By placing clinical outcome data from a cohort in the context of longitudinal surgical registers and bacteremia data, we describe a valuable approach to adjusting estimates of the burden of typhoid fever.

Background Invasive Salmonella infections cause significant morbidity and mortality in Sub-Saharan Africa. However, the routes of transmission are uncertain. We conducted a case-control study of index-case and geographically-matched control households in Blantyre, Malawi, sampling Salmonella isolates from index cases, healthy people, animals, and the household environment. Methodology Sixty index cases of human invasive Salmonella infection were recruited (March 2015-Oct 2016). Twenty-eight invasive Non-Typhoidal Salmonella (iNTS) disease and 32 typhoid patients consented to household sampling. Each index-case household was geographically matched to a control household. Extensive microbiological sampling included stool sampling from healthy household members, stool or rectal swabs from household-associated animals and boot-sock sampling of the household environment. Findings 1203 samples from 120 households, yielded 43 non-Typhoidal Salmonella (NTS) isolates from 25 households (overall sample positivity 3.6%). In the 28 iNTS patients, disease was caused by 3 STs of Salmonella Typhimurium, mainly ST313. In contrast, the isolates from households spanned 15 sequence types (STs). Two S. Typhimurium isolates from index cases closely matched isolates from their respective asymptomatic household members (2 and 3 SNP differences respectively). Despite the recovery of a diverse range of NTS, there was no overlap between the STs causing iNTS disease with any environmental or animal isolates. Conclusions The finding of NTS strains from index cases that matched household members, coupled with lack of related animal or environmental isolates, supports a hypothesis of human to human transmission of iNTS infections in the household. The breadth of NTS strains found in animals and the household environment demonstrated the robustness of NTS sampling and culture methodology, and suggests a diverse ecology of Salmonella in this setting. Healthy typhoid (S. Typhi) carrier state was not detected. The lack of S. Typhi isolates from the household environment suggests that further methodological development is needed to culture S. Typhi from the environment.

Background Diverse environmental exposures and risk factors have been implicated in the transmission of Salmonella Typhi, however, the dominant transmission pathways through the environment to susceptible humans remain unknown. Here, we utilize spatial, bacterial genomic, and hydrological data to refine our view of Typhoid transmission in an endemic setting. Methods 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 (SNV) 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 less than 15 to over 100 cases/100,000/year. Pairwise SNV distance and physical household distances were significantly correlated (p=0.001). 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=0.003). We also found spatial correlation at a smaller spatial scale, of households living &lt;192 meters 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 multi-faceted data can be used to identify high incidence areas, understand the connections between them, and inform targeted environmental surveillance, all of which will be critical to shape local and regional typhoid control strategies.

Background Invasive Salmonella infections are a major cause of morbidity and mortality in Sub-Saharan Africa (SSA), but the sources and transmission routes are uncertain. We investigated potential sources for cases of invasive disease by sampling healthy people, animals, and the environment in index-case and geographically-matched control households. Methods Sixty index cases of human invasive Salmonella infection were recruited (28 invasive Non-Typhoidal Salmonella (iNTS) disease, and 32 typhoid). Each index-case household was geographically matched to a control household. Extensive microbiological sampling included stool sampling from healthy household members, stool or rectal swabs from household-associated animals and boot-sock sampling of the household environment. Results 1203 samples were taken from 120 households, yielding 43 Salmonella isolates from 25 households (overall sample positivity 3.6%). Isolates from households were all NTS and spanned 15 STs. iNTS disease was caused by 3 STs of Salmonella Typhimurium, mainly ST313. Two S. Typhimurium isolates from index cases closely matched isolates from their respective asymptomatic household members (2 and 3 SNPs different respectively). There was no overlap of STs causing iNTS disease with environmental or animal isolates, despite recovery of diverse NTS. Conclusions The finding of NTS strains from index cases matching household members, coupled with lack of overlap with either animal or environmental isolates, supports a hypothesis that healthy humans are the source of iNTS infections in the household. The breadth of NTS strains found in the household environment across all sites demonstrated the robustness of sampling and methodology to detect NTS, and suggests a diverse ecology of Salmonella in this setting. The lack of S. Typhi isolated from the household environment may suggest a need for further methodological development to culture sources of typhoid.