Integrating environmental and clinical surveillance for typhoid: insights from a prospective wastewater surveillance study in urban India

Summary

Background

Wastewater surveillance can be used to monitor Salmonella enterica serovar Typhi (S Typhi) transmission in the community, allowing early outbreak detection and vaccine effect evaluation. This study aims to establish the baseline detection rate of S Typhi in wastewater and to evaluate its relationship with the clinical burden of typhoid before typhoid conjugate vaccine (TCV) introduction in an endemic urban context.

Methods

We conducted wastewater surveillance for S Typhi in Vellore, Tamil Nadu, India, from Dec 12, 2022, to June 30, 2023, before implementing a cluster-randomised trial of a TCV. 723 grab and 723 Moore swabs were collected from 50 sampling locations and concentrated using Nanotrap A beads for grab samples and Moore swabs as a passive sampler. S Typhi was detected using multiplex quantitative PCR targeting ttr, staG, and tviB genes. Blood culture-confirmed cases of typhoid fever were recorded through hospital-based surveillance established as a part of an independent ongoing VeVact study, an observer-blinded, cluster-randomised trial of a TCV (TyphiBEV). We calculated proportions of typhoid positivity in wastewater sampling sites and the incidence rates of blood culture-confirmed typhoid cases. Spatial analysis was performed to detect statistically significant hotspots for typhoid fever cases in the study area. Bayesian inference using the integrated nested Laplace approximation and mixed-effects spatiotemporal regression quantified associations between wastewater S Typhi detection and typhoid incidence.

Findings

S Typhi was detected in 80 (5·5%) of 1446 wastewater samples, higher in grab samples (60 [8·3%] of 723) than in Moore swabs (20 [2·8%] of 723). 207 typhoid cases were recorded, with cases peaking between March and May, 2023. Spatial clustering of typhoid cases corresponded with higher wastewater S Typhi detection. The proportion of wastewater samples that were positive over the study period correlated with typhoid incidence when analysed across the 60 vaccine trial clusters (Spearman’s ρ=0·32; p=0·014) or the non-overlapping wastewater site catchment areas (ρ=0·57; p<0·0001). Mixed-effects logistic regression adjusting for spatiotemporal random effects, levels of human faecal contamination, rainfall, and laboratory processing method, gave a 2·43 (95% CI 1·38–4·27) times higher odds of detecting S Typhi in wastewater samples for each 10-fold increase in typhoid fever incidence.

Interpretation

Wastewater surveillance data correlate with clinical disease incidence in an endemic setting and could be used to monitor targeted interventions, including TCV introduction.

Funding

The Gates Foundation.

Introduction

Typhoid fever, caused by Salmonella enterica serovar Typhi (S Typhi), is endemic in low-income and middle-income countries. The Global Burden of Disease study estimated that in 2021 there were 7·2 million new typhoid cases and between 46 300 and 158 200 typhoid-related deaths. Nearly 90% of typhoid morbidity and mortality occur in Asia, with India contributing over half the global burden.¹ Estimates from the Surveillance for Enteric Fever in India (SEFI) study published in 2023 documented typhoid fever incidence ranging from 576 to 1173 per 100 000 child-years among children younger than 15 years in urban locations.²

Research in context.

Evidence before this study

Typhoid fever remains a substantial public health burden in low-income and middle-income countries, particularly in south Asia. The Global Burden of Disease study estimates millions of typhoid fever cases annually, with India contributing over half the global burden. Conventional typhoid surveillance relies primarily on clinical case detection through blood cultures or serological assays, but these methods are constrained by health-care-seeking behaviours, availability of blood culture, and diagnostic accuracy. Wastewater-based epidemiology has emerged as a promising tool for monitoring pathogens in community settings. We searched MEDLINE (PubMed) for studies published in English from database inception up to Jan 1, 2025, on wastewater surveillance for Salmonella enterica serovar Typhi (S Typhi) using keywords: "wastewater detection of S Typhi", "wastewater-based epidemiology", "burden of typhoid fever", and "wastewater-based epidemiological monitoring", along with relevant MeSH terms. Previous studies have successfully applied wastewater surveillance for detection of poliovirus, hepatitis A, antimicrobial resistance genes, and, more recently, S Typhi. In Bangladesh, Nepal, and the Democratic Republic of the Congo, S Typhi has been detected in water sources and wastewater, often coinciding with disease outbreaks. Although S Typhi has been detected in wastewater in various countries, there is a scarcity of large-scale studies correlating it with blood culture-confirmed cases of typhoid in endemic areas of India. Additionally, incomplete understanding remains regarding how factors such as sampling type, laboratory processing methods, catchment size, and human faecal load affect the sensitivity and interpretability of wastewater detection results.

Added value of this study

This study is one of the first to assess S Typhi detection in wastewater before introducing a typhoid conjugate vaccine (TCV) in an endemic urban setting. We used a combination of grab samples with the Nanotrap microbiome particle-based concentration technique to enhance the sensitivity of detection and conventional Moore swabs, followed by enrichment in universal pre-enrichment media and membrane filtration. Over 6 months, S Typhi was detected in 80 (5·5%) of 1446 wastewater samples, and the sensitivity of the grab sample method was better than Moore swabs plus filtration.

Implications of all the available evidence

Unlike conventional clinical surveillance, which depends on symptomatic individuals seeking medical care and having access to blood culture, wastewater surveillance can capture asymptomatic carriers and infections that are treated at home or are subclinical, providing a more comprehensive understanding of disease transmission. A notable association between wastewater S Typhi detection and typhoid incidence suggests that integrating wastewater surveillance into routine disease monitoring could inform targeted interventions, such as vaccination campaigns and improvements in water, sanitation, and hygiene systems in high-burden communities. Future research should explore the long-term role of wastewater-based epidemiology in evaluating the effect of TCV introduction in endemic settings.

Typhoid transmission is mainly faecal-oral; populations lacking safe water and adequate sanitation, and economically disadvantaged groups are at higher risk.^3,4 Although most patients with typhoid recover with antibiotic treatment, 1–5% become chronic carriers, shedding S Typhi in stools or urine for over a year.⁵ Prevention strategies include ensuring safe water and adequate sanitation, proper hygiene, and typhoid vaccinations in endemic settings.⁶ Wastewater surveillance offers a promising alternative for creating a cost-effective surveillance system.⁷

Wastewater-based epidemiology operates on the premise that wastewater samples reflect community urine and faeces, allowing analysis of overall community health status.⁸ Wastewater surveillance has been applied to monitor poliovirus, hepatitis A, protozoal infections, SARS-CoV-2, and antimicrobial resistance (AMR) genes.^9,10 Wastewater-based epidemiology could complement typhoid clinical surveillance by capturing symptomatic cases and asymptomatic carriers, enabling real-time pathogen monitoring.¹¹ However, few studies have examined the association between S Typhi detection in wastewater and clinical disease incidence.¹² We present findings on S Typhi wastewater surveillance in an urban Indian setting, conducted before the implementation of a cluster-randomised trial of the typhoid conjugate vaccine (TCV). This study documents the relationship between wastewater and clinical surveillance data, thus providing baseline information for evaluating the effect of the vaccine following its introduction.

Methods

Study design and participants

This prospective, repeated wastewater surveillance for S Typhi was conducted across 33 urban wards spanning 33·8 km² in Vellore city (12·92°N, 79·13°E), in northern Tamil Nadu, India, between Dec 12, 2022, and June 30, 2023. Vellore experiences a tropical savanna climate with temperatures ranging from 13°C to 41°C, and peak temperatures occurring between April and June. Annually, Vellore receives 1034 mm of rainfall, with two monsoon seasons occurring in June and December (appendix p 1). Christian Medical College Vellore has established the Vellore Health and Demographic Surveillance System (VHDSS) across these 33 wards, covering a population of approximately 225 000. Concurrent with the wastewater surveillance, an independent, observer-blinded, cluster-randomised trial of a TCV (TyphiBEV) in urban Vellore (VeVact), was initiated in these wards to assess the relative and absolute reduction rate of symptomatic blood culture-confirmed typhoid.¹³ Vaccination of consenting individuals aged 1–30 years in half of these clusters occurred between July and September, 2023. This analysis establishes baseline detection of S Typhi in wastewater samples before TCV introduction (December, 2022, to June, 2023) in these wards and its association with blood-culture-confirmed typhoid fever cases reported from the same area.

Both the environmental surveillance and VeVact study-related research activities have been approved by the Institutional Ethics Committee at Christian Medical College Vellore (IRB number 11 170 dated Feb 28, 2018; amended as A27 dated Sept 28, 2022, for environmental surveillance and IRB number 14247, [VEVACT] dated Sept 29, 2021). Written informed consent was obtained from all study participants or from parents or guardians of all child participants. Additionally, written informed assent was obtained from children aged 11–17 years.

Procedures

Selection of wastewater sampling sites and sampling

The sewage network in the city is predominantly made of open drains. Details regarding the selection of environmental sampling sites can be found elsewhere.¹⁴ In summary, we analysed geospatially referenced datasets of wastewater networks within a geographical information system (GIS) framework to identify all confluence points where two or more channels merged. The spatial coordinates for these locations were generated and verified by the field team, leading to the final selection of sampling locations that represent the geographical area.¹⁴ Initially, 50 sampling locations were selected; later, they were expanded to 54 in August, 2023, to match the target population of the VeVact study.

Families in the study area were mapped to the sewer lines after a visual inspection of these houses by the field personnel. Catchment polygons were then mapped in ArcGIS (version 10.8) to include all the families contributing to the wastewater sampling locations. These catchments were then categorised as either non-overlapping (level 1), in which they were independent without any linkage to other catchments; some overlap (level 2), in which the catchment populations covered up to three upstream wastewater sampling locations; and large overlap (level 3), in which the catchment populations had more than three wastewater sampling locations within their boundaries (appendix p 2). Following appropriate infection control procedures, trained field personnel collected 1000 mL grab samples fortnightly from each site. Moore swabs (trap samples) were deployed 48 h before collection. Grab sample collection and Moore swab deployment and collection were consistently carried out in the early morning (0700–0900 h) to account for variation in sewage flow and other characteristics of wastewater on pathogen detection. Samples were transported to a biosafety level 2 laboratory at 2–4°C within 4 h of collection.

This study used the Nanotrap microbiome A particles (NMP-A)-based concentration method for processing grab samples with an input volume of 40 mL. The eluate was centrifuged to generate a pellet for storage at –20°C. Moore swabs were left in situ for 48 h, incubated overnight in universal pre-enrichment broth before filtration, and the filter discs were stored at –20°C for further analysis.

DNA extraction from wastewater samples and molecular testing

DNA was extracted using the QIAamp PowerFecal Pro DNA Kit (Qiagen, Germany; sourced through Shriram Enterprises), and multiplex quantitative PCR (qPCR) with primers and probes targeting three genes (ttr, staG, and tviB), and samples with all three targets present were considered positive (Ct values <38) for S Typhi).¹⁵ In addition to duplex qPCR for HF183, a human-restricted Bacteroides marker for human faecal contamination, an internal positive control and sample process control (Eurogentec RT-SPCC-Q10; sourced through Genex) were performed on all samples. Genome copy numbers (gc/mL) were estimated using standard curves generated using gBlocks DNA fragments for each target. In the laboratory, invalid qPCRs were either rerun with the eluate or the samples were re-extracted from pellets, and subsequently subjected to qPCR, ensuring that no data was missing. Primers and probes for ttr, tviB, staG, and HF183, and gBlocks DNA fragments were sourced from NuGene Technologies. The laboratory assay protocols are described in detail on protocols.io.16, 17, 18

Clinical surveillance for typhoid fever

In the VeVact trial, all the eligible participants from the study clusters were actively encouraged to contact a dedicated helpline number in the event of developing a fever. Appropriate investigations, including blood culture if the acute febrile illness lasted more than 2 calendar days, were performed if the patients visited any of the 14 health-care facilities in the VeVact study network of hospitals.¹³ Homes of blood culture-positive typhoid cases were visited and geospatially mapped by the environmental surveillance team.

Meteorological data

Common meteorological parameters, including humidity, minimum and maximum temperature, and rainfall, were collected daily from publicly available information on the website of the Regional Meteorological Centre, India Meteorological Department.¹⁹

Outcomes

The primary outcomes of the study are to establish the baseline detection rate of S Typhi in wastewater and to evaluate temporal trends and the relationship between S Typhi in wastewater and the clinical burden of typhoid before the introduction of TCV. The incidence rates of blood-culture-confirmed typhoid fever among all recruited participants in the VeVact study during the surveillance period were also estimated. The secondary outcome includes the spatial distribution and identification of typhoid fever hotspots in the study area before the introduction of TCV.

Statistical analysis

Wastewater samples positive for all three gene targets (ttr, staG, and tviB) were considered positive for S Typhi. Prevalence of detection was evaluated across all environmental surveillance sites, and longitudinal trends were analysed. For each site, the prevalence rate was defined as the proportion of wastewater samples that tested positive for all three gene targets, calculated as the number of positive samples divided by the total number of samples collected at that site. Wastewater sampling locations were mapped to the VeVact trial clusters, and positivity rates in the trial clusters were estimated.

Incidence rates of typhoid fever at the cluster or catchment level were estimated as the number of blood culture-confirmed cases reported by the central reference laboratory, divided by the corresponding population denominator from the VHDSS, and expressed as cases per 1000 person-years. Correlations between the environmental and clinical burden of S Typhi for the study period were assessed for all three catchment levels.

Spatial analysis was performed to detect statistically significant hotspots for typhoid fever using the Hot Spot Analysis (Getis-Ord Gi∗) spatial statistical tool in ArcGIS (version 10.8). Spatial autocorrelation (Global Moran's I) was assessed by overlaying 100 × 100 m grids across the entire study area.²⁰

Bayesian hierarchical modelling was performed to fit spatiotemporal models using the stochastic partial differential equation (SPDE) approach and the R-integrated nested Laplace approximation (INLA) package²¹ (R version 4.4.1), with a binomial distribution used to predict wastewater positivity and a Gaussian distribution used to predict log-transformed genome copies of the tviB gene target at the centroids of catchment areas. This technique allows for the incorporation of spatial random effects within a Bayesian framework, accounting for spatial dependence while retaining the mixed-effects structure of the model. Wastewater sampling locations and week of sampling were fitted as random effects. INLA predictions, typhoid cases, and hotspots for typhoid were spatially mapped. Locations of typhoid cases were jittered to maintain anonymity, and geospatial maps were generated. Spatial and temporal variation in tviB concentrations was modelled using a Bayesian hierarchical Gaussian model in R-INLA. The model included a random intercept for sampling site, an AR(1) term for temporal autocorrelation, and a spatially structured random effect via an SPDE approach (appendix pp 4–5). Mixed-effects logistic regression analysis was performed in Stata (version 16.1), incorporating significant predictors (p<0·05) from univariable logistic regression to study the association between typhoid disease and wastewater positivity, adjusting for human faecal contamination levels, sample processing method, and rainfall on sampling day, with site and time as random effects (appendix p 6).

Role of the funding source

The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report.

Results

Between Dec 12, 2022, and June 30, 2023, 723 grab samples and 723 Moore swabs were collected from 50 sampling locations. S Typhi was detected in 60 (8·3%) of 723 grab samples and in 20 (2·8%) of 723 Moore swabs based on the presence of all three gene targets (ttr, staG, and tviB). The overall positivity from wastewater across both methods was 80 (5·5%) of 1446. Only eight (40%) of 20 Moore swab positive samples were associated with positive grab samples collected at the same time. Across the different sites, detection varied from zero in 29 (58%) of 50 sampling locations, at least once in 21 (42%) of 50 locations, and more than 35% of the time at three of 50 sampling locations (figure 1). Wastewater detection of each of the three gene targets is provided in the appendix (p 2). Higher detections for tviB and staG gene targets were observed with grab samples processed using the NMP-A method compared with Moore swabs. The gene target tviB was detected in 165 (11·4%) of 1446 wastewater samples. The mean genome copies for this target were 69·22 (SD 195) gc/mL, whereas the median was 10·54 (IQR 4·99–42·28) gc/mL. Genome copies of tviB were detected consistently over time in a few non-overlapping and one level 3 catchment areas (figure 2).

Figure 1.

S Typhi detection in wastewater samples

(A) Nanotrap A. (B) Moore swabs. (C) Either method. S Typhi was detected in 21 (42%) of 50 sampling locations in grab samples and ten (20%) of 50 locations using Moore swabs. Higher detection rates were observed from samples collected from locations in the northeastern and central part of the study area. The dots show the locations of sample collection and are scaled according to the proportion of samples from that location that were positive by qPCR for S Typhi. The grey lines indicate the sewage network, and the black lines show the cluster allocation in the cluster-randomised trial of typhoid conjugate vaccine that began after the study period. Shapefile of the study area and vaccine trial clusters created by Christian Medical College Vellore. qPCR=quantitative PCR. S Typhi=Salmonella enterica serovar Typhi.

Figure 2.

Heatmap of tviB concentration in wastewater samples over time

Varying concentrations of the tviB gene target were detected from wastewater surveillance at each site over the study period. Detection was variable across sites, with two sites showing almost continuous detection and others where detection was intermittent or absent. Sites are grouped by the extent of overlap of their associated catchment population with that of other sites: non-overlapping catchments (A); catchments with some overlap (B); and catchments with large overlap (C). tviB=Salmonella Typhi-specific gene.

207 culture-confirmed typhoid fever cases were documented between December, 2022, and June, 2023. Most typhoid fever cases were distributed in the northeastern part of the study area, which had an outbreak of typhoid fever between March and May, 2023. The unadjusted incidence of blood culture-confirmed typhoid fever for the baseline period was 180·2 per 100 000 person-years (95% CI 155·7–204·8). Children aged 5–14 years had the highest incidence (699·1 per 100 000 person-years), followed by children younger than 5 years (685·1 per 100 000 person-years). People aged 30 years and older had the lowest incidence of typhoid fever (19·9 per 100 000 person-years) (figure 3). The incidence of typhoid fever varied from zero to a maximum of 1734 per 100 000 person-years among the 60 VeVact study clusters and from zero to 1216 per 100 000 person-years in the catchment areas of the 50 wastewater sampling locations.

Figure 3.

Unadjusted incidence of typhoid fever across different age groups

During the study period, the crude incidence of blood culture-confirmed typhoid fever was highest among children aged 5–14 years, followed by children younger than 5 years, and people aged 15–30 years. The lowest burden was noticed among people older than 30 years.

Hotspot analysis of incident cases of typhoid fever identified 15 locations in the northern part of the study area as statistically significant hotspots (figures 4 and 5). Spatial autocorrelation analysis revealed incident cases of typhoid fever to be surrounded by houses with typhoid fever cases and was significant with a Moran’s Index of 0·446 (Z score 45·195; p<0·0001).

Figure 4.

INLA model predictions for S Typhi detection in wastewater and the location of typhoid fever cases

(A) INLA SPDE model predictions for S Typhi detection in wastewater at the centroids of sampling site catchments. (B) Jittered location of incident typhoid fever cases during the study period and estimated spatial hotspots of typhoid fever cases. INLA=integrated nested Laplace approximation. SPDE=stochastic partial differential equation. S Typhi=Salmonella enterica serovar Typhi.

Figure 5.

Wastewater detection rates for S Typhi and spatial hotspots of incident typhoid cases

The proportion of S Typhi-positive wastewater samples is represented using blue circles, with increasing circle size corresponding to higher detection percentages. Locations with no S Typhi detection are marked in black. Individual typhoid cases are shown as grey dots, whereas pink areas represent statistically significant hotspots of typhoid cases. S Typhi=Salmonella enterica serovar Typhi.

S Typhi gene targets (gc/mL) in wastewater were detected for the most part of the sampling period, with higher detections (83 [12·4%] of 668) between March and May, 2023, as compared with 82 (10·5%) of 778 during other months. The incidence of typhoid fever (20·7–32·2 per 10 000 person-years) was also higher during this period. We did not observe a consistent pattern between average weekly rainfall, detection in wastewater, and typhoid burden (appendix p 3).

A positive and significant correlation was observed between the prevalence of S Typhi in wastewater and the incidence of typhoid fever across all catchments of the 50 wastewater sampling locations (Spearman’s ρ=0·44; p=0·001). These two variables were also correlated at the level of the wastewater catchment areas of the 37 non-overlapping catchments (ρ=0·57; p<0·0001) and across all catchments for the study period (figure 6).

Figure 6.

Correlation between detection of S Typhi in wastewater and typhoid disease

Correlation plots for wastewater positivity for S Typhi and the incidence of typhoid fever across the different types of catchments: non-overlapping catchments (A); catchments with some overlap (B); catchments with large overlap (C); and all catchments (D). A positive and significant correlation was observed in the upstream catchment areas of the 37 non-overlapping catchments and across all catchments. S Typhi=Salmonella enterica serovar Typhi.

Spatiotemporal modelling of S Typhi detection in wastewater samples using the SPDE approach showed heterogeneity across the study area with higher probability (>45%) in and around catchments with clinical disease hotspots and increased density of typhoid fever cases during the study period (figures 4A and 4B). The Bayesian INLA model revealed that the baseline odds of the detection of S Typhi in wastewater were low (odds ratio [OR] 0·01 [95% credible interval (CrI) 0·0008–0·069]). The random effects captured variability across different sites and over time, with high precision estimates indicating minimal residual variance for effects specific to each site and timepoint. The temporal correlation was negligible (rho ≈ 0), suggesting that the outcome had little dependence on previous timepoints. Spatial effects contributed moderately to the variation, with a standard deviation of 2·46 (95% CrI 1·30–4·35) and a large spatial range (approximately 4706). Overall, the model emphasises the low baseline odds, with spatial heterogeneity identified as the primary source of residual variation. A similar pattern was observed with tviB gene target concentrations (appendix p 4).

We assessed factors associated with the detection of S Typhi in wastewater samples using both univariable and multivariable mixed-effects logistic regression models. In the multivariable analysis, log-transformed typhoid fever incidence rate in the catchment was significantly associated with increased odds of S Typhi detection (adjusted OR [aOR] 2·43 [95% CI 1·38–4·27]; p<0·001), indicating that every 10-fold increase in incidence of typhoid fever was associated with 2·43 times higher odds of detecting S Typhi in wastewater samples. Higher levels of human faecal contamination were also significantly associated with S Typhi detection (aOR 1·32 [95% CI 1·01–1·72]; p=0·044), suggesting increased odds of detection with higher HF183 concentrations. The NMP-A method was associated with greater odds of detecting S Typhi in wastewater compared with the Moore swab (aOR 4·46 [95% CI 2·28–8·73]; p<0·001). Rainfall on the day of sampling was significantly associated with detection in the adjusted model (aOR 1·04 [95% CI 1·01–1·07]; p=0·018), suggesting a modest positive association between precipitation and wastewater detection. Although cumulative incident typhoid cases showed a significant association in univariable analysis (OR 1·32 [95% CI 1·13–1·55]), this variable was not retained in the mixed-effects model. Other catchment-level variables, such as population density and area, were not significantly associated with detection in the univariable models. Random effects for sampling location showed substantial variability in detection across sites (table).

Table.

Factors associated with S Typhi detection in wastewater

	Bivariate logistic regression		Mixed effect logistic regression
	OR (95% CI)	p value	aOR (a95% CI)
Log10 typhoid incidence per 1000 person-years	2·718 (2·118–3·487)	<0·001	2·426 (1·380–4·267)
Log10 Hf183 (gc/mL)	1·149 (1·064–1·242)	<0·001	1·317 (1·010–1·717)
NMP-A method compared with Moore swab	3·181 (1·897–5·335)	<0·001	4·463 (2·282–8·730)
Rainfall on sampling day (mm)	1·007 (1·058–1·033)	<0·001	1·038 (1·006–1·070)
Maximum temperature on sampling day (°C)	1·052 (0·990–1·117)	0·099	··
Cumulative incident typhoid cases in the catchment till sampling week	1·322 (1·131–1·546)	<0·001	··
Population density of catchment	0·928 (0·770–1·117)	0·428	··
Area of catchment (km2)	1 (1–1)	··	··
Random effects
Sampling location (variance, SE)	··	··	3·218 (1·508)
Week of sampling (variance, SE)	··	··	0·572 (0·843)

aOR=adjusted odds ratio. NMP-A=Nanotrap microbiome A. OR=odds ratio. S Typhi=Salmonella enterica serovar Typhi.

Discussion

This study aimed to establish baseline prevalence of S Typhi in wastewater and assess its correlation with typhoid fever before TCV introduction. The prevalence of S Typhi based on qPCR for three gene targets was 80 (5·5%) of 1446 across all wastewater samples; higher detection rates were observed in grab samples processed using NMP-A particles compared with Moore swabs. S Typhi was not detected in wastewater in more than half of the sampling locations, whereas higher detection rates were noticed at sites located in the northeastern and central region of the study area. A wastewater surveillance study conducted during the COVID-19 pandemic, between May, 2021, and April, 2022, used similar sample collection and laboratory processing methods in both high (Vellore) and low (Blantyre) typhoid burden settings. The study found that the detection rate of S Typhi in wastewater samples was 118 (11·4%) of 1037 in Vellore and 34 (3·0%) of 1127 in Blantyre. Additionally, the crude incidence of typhoid fever was recorded at 7·8 in Vellore and 18·4 cases per 100 000 person-years in Blantyre. In Vellore, Moore swab positivity was significantly associated with HF183 levels and upstream catchment population size, with increased detection during the monsoon.¹²

Earlier studies from Vellore have found Moore swabs performing better than grab samples processed using membrane filtration (79 [15·3%] of 517 vs 39 [7·5%] of 520). A validation study conducted by Moe and colleagues in endemic areas of Kolkata and Dhaka evaluated the effectiveness of Moore swabs and ultrafiltration of grab samples through spiking experiments. Although there was no gold-standard assay available to formally assess sensitivity and specificity, a head-to-head comparison based on the positivity of unspiked environmental samples indicated that both methods were sensitive and effective for concentrating and detecting S Typhi in wastewater.²²

NMP-A use a novel microbial concentration technology using magnetic hydrogel particles, rapidly capturing and concentrating microorganisms, and have been used for various viruses, including SARS-CoV-2, influenza, Zika, respiratory syncytial virus, dengue, and chikungunya viruses from both clinical and wastewater samples.²³ This analysis indicates that the NMP-A concentration method outperforms the Moore swab method for detecting S Typhi-related gene targets in low-volume wastewater samples. However, extended time-series data will be important to elucidate this relationship while accounting for potential seasonal trends.

Detection of pathogens in wastewater-based environmental surveillance depends on the number and characteristics of sampling locations, sampling techniques, frequency, and sensitivity of assays.²⁴ Although a simulation study conducted in Kolkata, which validated various sampling strategies under different risk scenarios for S Typhi, recommends an adaptive sampling site allocation method to enhance sensitivity,¹¹ a consistent sampling strategy might be more effective for monitoring trends and the effect of interventions over time.

The community cohort component of the SEFI study in India estimated the adjusted incidence of typhoid fever to be between 12 and 1622 per 100 000 child-years of observation among children younger than 15 years across three urban settings, including Vellore, which had the highest incidence (1173 cases per 100 000 child-years of observation).² A community serosurvey among children in Vellore in 2022 has estimated the annual seroincidence of S Typhi infection to be 10·4 per 100 person-years, 10-fold higher than the incidence of blood culture-confirmed typhoid fever.²⁵ Higher seroincidence of typhoid in this region is consistent with previous findings and can be attributed to several factors, including limitations in the validity of blood culture, prevalent antibiotic usage practices, and the ability of seroincidence surveys to capture both clinical and asymptomatic infections.

In this study, the unadjusted incidence of blood culture-confirmed typhoid fever was 180·2 per 100 000 person-years. Children between 5 and 14 years had the highest burden (699·1 per 100 000 person-years), which was lower than reported adjusted rates from the SEFI study.² The study area had a typhoid outbreak between February, 2023, and June, 2023, when 98·5% of the incident cases over the study period were documented. Nearly 85% of these cases were spread over 12 contiguous clusters located in the northeastern part of the region, forming statistically significant hotspots. Similar hotspots for typhoid cases were observed in Vellore during an outbreak between 2017 and 2020.^4,26 Higher typhoid incidence rates have been reported in populations close to other typhoid fever cases in a study in the urban slums of Dhaka city.²⁷ Higher rates of detection of S Typhi from wastewater samples have also been documented during localised outbreaks in Dhaka, Bangladesh, Kathmandu, Nepal, and the Democratic Republic of the Congo. In Dhaka, 36 (61%) of 59 water samples from households with culture-confirmed typhoid fever cases tested positive for S Typhi. Similarly, in Kathmandu, environmental surveillance has shown a high positivity rate for S Typhi in river samples based on qPCR for the staG gene only (which can also be found in non-typhoidal Salmonella), reaching up to 96·6%.28, 29, 30 In contrast, surveillance in Blantyre, Malawi, revealed lower and more variable detection rates of S Typhi in wastewater and river samples. Infrastructural differences might play a role in these contrasting findings. Dhaka and Kathmandu have informal or overwhelmed sewage systems that discharge directly into water bodies, facilitating pathogen spread. Blantyre, however, lacks centralised sewage networks and relies more on pit latrines and boreholes, which might reduce direct contamination of surface water.³¹ In the absence of vaccination, spatial clustering of typhoid cases indicates an increased risk for people residing close to these cases, and further work is warranted to evaluate the effectiveness of targeted vaccination on typhoid fever burden and S Typhi transmission.

In this study, we have reported significant positive correlations between wastewater positivity for S Typhi and the incidence of typhoid fever across densely populated urban communities in an Indian context. Although there is a paucity of information in this area, similar trends have been observed in Bangladesh and Nepal, indicating the potential of wastewater-based epidemiology to function as an early warning system for typhoid outbreaks, allowing for timely public health interventions.^27,29

Marked heterogeneity was observed in S Typhi prevalence and tviB gene concentration by sampling location. Higher detection rates were observed downstream of typhoid case hotspots. In open-channel wastewater networks with variable flow, negative detection at sampling locations need not imply the absence of people who are infected, but that infections are likely missed by wastewater surveillance. Positive detections at downstream sites with larger catchment areas where wastewater is diluted suggest high typhoid prevalence.³²

A significant association was found between S Typhi detection in wastewater and typhoid disease incidence in catchment populations. Positive associations between S Typhi detections in wastewater and typhoid seroincidence have been documented earlier in Vellore.²⁵

A significant association between wastewater positivity for S Typhi and HF183 genome copy levels across all sampling locations suggests that human faecal contamination is a key determinant influencing pathogen detection and highlights the need to consider the variability in the amount of human waste in wastewater samples.

Although the frequency of wastewater sampling might be a possible limitation, we could not factor in the water, sanitation, and hygiene practices in the catchments in our analyses. Environmental surveillance in open drains is constrained by variable flow, intermittent water inputs, dilution from rainfall or runoff, and exposure to environmental contaminants, all of which can reduce the sensitivity of pathogen detection and introduce temporal variability. In this study, rainfall data were obtained from a single weather monitoring station and generalised across all sampling sites, which might have introduced spatial misclassification and restricted precision in assessing the influence of rainfall on S Typhi detection. Additionally, the short study period of 6 months might not capture the full seasonal or annual variability in typhoid transmission and wastewater detection, restricting the generalisability of temporal trends.

In the context of TCV introduction, wastewater surveillance could serve as a cost-effective, scalable tool to assess vaccine effect and provide ongoing monitoring. A sustained reduction in S Typhi detection in wastewater could indicate decreased community transmission and serve as an early indicator of vaccine effectiveness at the population level. When combined with molecular typing and AMR profiling, wastewater surveillance can track circulating strains, detect emerging resistant strains, and provide insights into transmission dynamics. Given its ability to capture both symptomatic and asymptomatic infections, wastewater-based epidemiology has the potential to perform as a low-cost sentinel system, particularly in resource-limited urban settings, to guide targeted vaccination campaigns, optimise water, sanitation, hygiene interventions, and monitor the long-term epidemiological effect of TCV programmes and AMR trends.

Data Sharing

The anonymised data with the data dictionary for this manuscript will be provided to investigators upon request by the corresponding author after approval of the proposal by the corresponding authors, with a signed data access agreement.

Declaration of interests

We declare no competing interests.