Measuring Environmental Exposure to Enteric Pathogens in Low-Income Settings: Review and Recommendations of an Interdisciplinary Working Group

Infections with enteric pathogens impose a heavy disease burden, especially among young children in low-income countries. Recent findings from randomized controlled trials of water, sanitation, and hygiene interventions have raised questions about current methods for assessing environmental exposure to enteric pathogens. Approaches for estimating sources and doses of exposure suffer from a number of shortcomings, including reliance on imperfect indicators of fecal contamination instead of actual pathogens and estimating exposure indirectly from imprecise measurements of pathogens in the environment and human interaction therewith. These shortcomings limit the potential for effective surveillance of exposures, identification of important sources and modes of transmission, and evaluation of the effectiveness of interventions. In this review, we summarize current and emerging approaches used to characterize enteric pathogen hazards in different environmental media as well as human interaction with those media (external measures of exposure), and review methods that measure human infection with enteric pathogens as a proxy for past exposure (internal measures of exposure). We draw from lessons learned in other areas of environmental health to highlight how external and internal measures of exposure can be used to more comprehensively assess exposure. We conclude by recommending strategies for advancing enteric pathogen exposure assessments.


■ BACKGROUND
Exposure to enteric pathogens is associated with a heavy disease burden, largely born by young children living in lowincome settings. 1 Globally, enteric infections represent the third leading cause of death among children under five, accounting for approximately 589 000 deaths in 2017. 2 Methods to characterize human exposure to enteric pathogens have not been advanced in the same way as they have in other areas of environmental health, such as exposure to air pollution or industrial chemicals, perhaps in part because the resources are not available in the settings where enteric diseases carry a disproportionally high burden of disease.
Exposure science was born out of a need to measure exposures to industrial and occupational air pollution and chemical toxicants. 3 This need arose largely in high-income countries following the industrial revolution, which ushered in a period of rapid urbanization and proliferation of factories in urban centers exploiting processes involving fuel combustion. Perhaps one of the most notable examples of this sudden increase of population exposure to high levels of air pollution is the London smog incident, which caused an estimated 12 000 casualties in 1952. 4 Thereafter, air pollution exposure scientists developed methods that advanced exposure assessments, ranging from stationary equipment to measure aggregate population-level exposures and pollutant trends, to portable air pollution monitors that provide personal data in real time. 5 Chemical exposure science has since evolved to include both external and internal biological measures of several hundred toxicants in large, diverse populations.
There are a number of factors that contribute to the difficulty of characterizing exposure to enteric pathogens; some of these factors are associated with the microbial agent, others with the environment and yet others with the host response. First there are technological difficulties of detecting and quantifying a wide range of microbes whose pathogenicity, transmissibility, and fate and transport can vary under different environmental conditions. 6,7 Even identifying the agents is challenging, as there is still considerable uncertainty about whether and under what conditions detectable microbes are pathogenic, specific pathogens can be present at low concentrations in the environment, and due to frequent discoveries of new and emerging pathogens or known organisms that have acquired disease causing genes. While the source of enteric pathogens is mainly from fecal contamination, there are mixed correlations between the presence of indicators of fecal contamination and enteric pathogens in the environment. 8,9 Second, environmental enteric pathogen exposure assessments need to consider multiple potential fecal−oral transmission pathways, commonly represented by the F-diagram: 10 fingers, flies, food, fluids (water sources), fields (soil), and fomites. 11−13 Of increasing interest is the role of zoonotic transmission of enteric pathogens via animal excreta. 14 Third, the biological relevance of enteric exposures is complicated by responses to exposure, which is affected by various host susceptibility factors that mediate dose−response relationships, including those engendered by vaccines or previous exposures. 15 Response to exposure may also be influenced by gastrointestinal health, such as the diversity of the gut microbiome 16 or increased risk due to enteropathies such as environmental enteric dysfunction. 17 External versus Internal Measures of Exposure to Enteric Pathogens. Exposure to an environmental contaminant is conditional on a complex source-to-host pathway that includes (1) the release of a contaminant from its source; (2) the fate and transport of the contaminant in the environment leading to a specific concentration in different environmental media; (3) the host interaction with those media, (4) a portal of entry into the human body (i.e., an exposure route), and (5) uptake of the contaminant into the body. Traditional exposure science uses numerous approaches for measuring human exposure along this pathway continuum. A recent National Research Council report, Exposure Science in the 21 st Century, presented a summary of these approaches, ranging from those that measure environmental concentrations of contaminants to predict exposures before the contaminant reaches the human boundary, to those that estimate a dose after the contaminant has been taken up into the body. 18 Here, we propose a similar approach to defining exposure to enteric pathogens, by differentiating between external and internal exposure assessments. We define enteric pathogens as microorganisms transmitted via the fecal−oral route that can cause gastrointestinal infections, leading to acute (e.g., diarrheal disease, gastroenteritis, enteric fevers) and chronic infectious disease outcomes (e.g., environmental enteric dysfunction, growth faltering, impaired cognitive ability). 19 Enteric pathogens include bacterial, viral and protozoan pathogens, with fungi and helminths increasingly receiving more attention as causes of neglected tropical infectious disease. 20,21 A common approach for external exposure assessments is to detect indicators of fecal contamination or specific pathogens in a known size of environmental sample. Environmental measures such as these can act as proxies for external exposure as they assess specific fecal−oral transmission pathways. Measuring host-specific microbial source tracking markers in environmental samples can also provide information on the source of fecal contamination (i.e., animal versus human).
However, environmental measures on their own do not provide precise measures of the magnitude of exposure (i.e., how much contaminated water one actually ingests) which must be estimated or imputed, so can only be considered as an indirect proxy for actual ingestion of pathogens. Augmenting environmental data with data on human interactions with their environment, such as those published by the United States Environmental Protection Agency in the Exposure Factors Handbook, can help provide actual estimates of exposure (i.e., pathogen ingestion). 22 While the handbook is focused on chemical exposures in the United States, it demonstrates the types of frameworks that can be employed to inform assumptions for external exposure assessments, such as ingestion rates (drinking water, soil, and food) and object mouthing. 22 In contrast, internal enteric pathogen exposure assessments using human biological specimens may estimate the actual exposure to enteric pathogens after crossing the human body envelope, 23 typically via oral ingestion. In this respect, they address the main shortcoming of the external exposure assessment, i.e., whether a pathogen has actually been ingested. On the other hand, they provide little information about the source or transmission pathway that external exposure assessments offer. They can, however, provide some information on the presence, types, and frequency of past exposure to enteric pathogens as well as indications of potential health impacts. While the ingested dose of enteric pathogens is not typically measured, past exposure events can be inferred from serology, detection of pathogens in feces, and other biomarkers as proxies of internal exposure.
The Need for Improved Exposure Methods: The WaSH Case. Systematic reviews of water, sanitation, and hygiene (WaSH) evaluations, conducted to identify the health effects of interventions designed to reduce enteric pathogen exposure, have generally found improved WaSH to be protective against diarrhea, 24 soil-transmitted helminthiasis, 25 and malnutrition. 26,27 However, much of the evidence is from observational studies or short, smaller-scale trials. Recent experimental field evaluations of some of these interventions found either no evidence of health benefits, 28−30 a reduction in diarrhea but no improvement in child growth, 31 or improved growth but no impact on diarrhea. 32,33 These mixed results have focused greater attention on the need for more rigorous exposure assessment, in part to explain why effects of WaSH interventions are realized in some trials and not others. A recent consensus statement recommends that interventions need to radically reduce fecal contamination in the environment to achieve more consistent child health benefits, 34 though which pathogens and pathways are most important and the necessary reductions needed to achieve health impactsmay be highly context-specific. Filling these gaps requires greater attention on more rigorous exposure assessment, as few studies have attempted to directly measure exposure along the various transmission pathways, and those that do normally rely on quantifying indicators of fecal contamination in the environment as a proxy for pathogens. Moreover, these approaches are largely confined to assessments of external exposure, sometimes combined with observations or modeling to estimate ingested dose of enteric pathogens.
The theory of change underlying the impact of a WaSH intervention on enteric health outcomes is that an intervention will prevent disease if the intervention (a) is capable of  35 but only a few have assessed the impact of an intervention on exposure to fecal contamination along the transmission pathways targeted by the interventions, in part because there is no consensus methodology for how to measure exposure. Some of the null findings from recent WaSH interventions are consistent with the WaSH theory of change, reporting null effects from potentially effective interventions delivered to a vulnerable population when coverage and uptake were low. 28,30,36,37 However, other studies have reported null effects on diarrhea 29,38 and/or stunting 29,31,38 even with higher levels of coverage and use; others have reported protective effects on stunting (but not diarrhea) with high levels of coverage and use, especially from reductions in open defecation. 32,33,39 Some WaSH evaluations, identified from a systematic review investigating the relationship between indicators of fecal contamination and child health outcomes, included exposure assessments that raised questions about the manner in which they characterized exposure (Figure 1). In a recent trial of individual and combined WaSH interventions, Luby and colleagues reported reduced diarrhea from WaSH interventions, except for water treatment using chlorine, even though reductions in fecal indicator bacteria were found in both stored drinking water and food in households that received the water treatment intervention compared to control households. 40 This suggests perhaps that the chlorine-susceptible indicator bacteria used to estimate fecal contamination were not representative of chlorine-resistant pathogens, such as protozoa that may have been contributing to diarrhea symptoms. 41 This is consistent with the trial's findings of reductions in Giardia infection in all WaSH arms except for the water treatment arm. 42 The same evaluation also reported reduced diarrhea and prevalence of infections with protozoa and soil-transmitted helminths in the sanitation-only arm 42,43 even though there was no evidence of a change in fecal indicator bacteria in water, food, soil, on hands, or a change in fly density 44 in this arm compared to controls, suggesting reductions in disease transmission not captured by the fecal indicator bacteria measurements. Reese and colleagues reported no impact on diarrhea but reduced stunting from a water supply and sanitation intervention despite no evidence of a reduction in fecal contamination of drinking water or hands. 45 Pickering and colleagues also found no effect of a community-led sanitation intervention on diarrhea but an improvement in child growth, despite no reduction in fecal contamination in drinking water; however, latrine fly presence and observed human and animal feces did significantly decrease in the treatment group. 32 Collectively, these studies illustrate the need for improved characterization of external exposures to enteric pathogens. Similarly, limited tools currently exist for assessing internal exposures. As described below, some studies combine external assessment with behavioral observations to estimate actual ingestion (e.g., measuring pathogens in soil and frequency of geophagia, or measuring fecal indicators deposited by flies when alighting on food and the number of fly landings). However, these methods rely heavily on assumptions about conditions and behaviors that vary significantly within and between individuals, and over time. Others have begun using proxies of internal exposure as an indicator of past exposure, such as measuring enteric pathogen shedding in child stool or serology from dried blood spots as indicators of past exposure. 46,47 While investigators have sought to reconcile the inconsistent relationship between exposure measurements and health with accepted theories of change, 48 the lack of a clear and consistent progression between these proxies of exposure and health outcomes raises fundamental questions about current exposure assessment methods. In addition, they raise questions concerning "how clean is clean enough" for realizing reductions in rates of infection and disease. For program  implementers to optimize interventions that improve health, rapidly identifying the leading sources of exposure in a community can help develop and test context-specific interventions that address them. Epidemiologists are seeking reliable data on which to build models to ascertain determinants of disease risk and prioritize disease control strategies. Public health officials are seeking data on exposure to implement cost-effective population-based strategies to mitigate health burdens. All of these demands rest on our ability to generate better, cheaper, faster, field-deployable approaches that can be used in low-resource settings with minimal training. With this as background, an interdisciplinary group of environmental health researchers convened at Emory University in Atlanta, Georgia (U.S.A.) in September 2019 for a workshop aimed at identifying priorities for improved approaches to measuring enteric pathogen exposure. While much of the discussion focused on exposure assessment in household-level WaSH studies, the group agreed that the challenges extended to exposure to enteric pathogens in lowand middle-income countries more generally that are transmitted via the fecal−oral pathway, including exposures from agricultural, commercial and recreational activities. The objectives of the workshop were to (a) identify applications and criteria under which to compare exposure assessment approaches, (b) explore potential lessons in exposure science from other areas of environmental health, (c) critically review current and emerging enteric pathogen exposure assessment practices across a range of applications, and (d) define research priorities to help move the field of enteric pathogen exposure science forward. Participants in the workshop were U.S.-based experts purposely selected from a variety of disciplines who were available and willing to meet and explore the issues presented. Such expert gatherings present the potential for selection bias in the opinions expressed.

■ APPLICATIONS AND CRITERIA
We identified four primary applications that could benefit from further resources dedicated to improving enteric pathogen exposure methods.
1. Identifying primary sources and modes of transmission: Improved measures to identify the leading sources and modes of transmission of enteric pathogens in a given setting could inform programs and investments designed to improve health by reducing fecal exposure. Specific pathogens are differentially mediated by various fecal−oral transmission pathways, and the importance of these pathways may differ between settings (e.g., urban vs rural), subpopulations (e.g., children in different age groups) or time points (e.g., season). Improved external exposure measures would contribute to a better understanding of what exposure pathways need to be interrupted in a given setting to achieve consistent child health benefits. 2. Evaluating interventions: Better tools to estimate enteric pathogen exposure could be leveraged to evaluate the extent to which interventions are able to reduce exposure, and could have implications for the sample sizes needed to detect differences between groups. These tools could include measuring pathogens or proxies along specific pathways that interventions are designed to interrupt, and measuring pathogens in human biological samples to characterize past exposure. Using exposure assessments to evaluate interventions would potentially enable a faster and more objective evaluation of interventions designed to improve health by reducing exposure compared to studies that focus on health end points. 3. Surveillance: Novel approaches to characterizing exposure to enteric pathogenssuch as monitoring emerging agents in wastewatercould be applied for surveillance purposes. These purposes could range from monitoring human infection prevalence, rapid assessments of environmental threats, such as potential infectious disease outbreaks or bioterrorism, to monitoring the progress of regional or national efforts toward reducing infections. 4. Exploratory: There are broader research applications that could benefit from improved measures of enteric pathogen exposure. Examples include previously understudied links between enteric pathogen exposure and its effects on the gut microbiome and malnutrition, as well as its relationship with comorbidities such as respiratory and vector-borne diseases. Improved exposure assessments can also be linked to important subclinical conditions, as well as to a better understanding of inflammatory responses that are increasingly associated with a wide range of health effects. There is also a need to better understand the underlying mechanisms of transmission and pathogenesis associated with environmental exposures to enteric pathogens. When considering different approaches to measuring exposure to enteric pathogens for each of these applications, there are a number of criteria to consider. These include the following: • External vs internal: Is exposure characterized in the environment as a proxy for external exposure and does it provide data on the source of exposure, or is it measured after the contaminant has crossed the human boundary (i.e., internal exposure)? For external exposure assessments, is exposure characterized proximal to the human boundary or is it a more distal measure that requires modeling to estimate more proximal exposure? For internal exposure assessments, are the measures mediated by host susceptibility to infection? • Pathway-specific: Can the exposure assessment quantify the relative contribution to total exposure by different transmission pathways? deployment in the field in low-resource or emergency settings? Does the assay require cold-chain transport or a consistent energy source? Are the required materials bulky or dangerous to handle? How much training and material is required to collect samples, conduct analyses, and interpret results? Is it fast enough to provide actionable feedback to reduce exposures in a population of interest? • Ethics: Are exposure assessment methods potentially burdensome on the communities and individuals where they are conducted? Do they require respondents to provide a substantial amount of resources (e.g., large volume water samples) and time? Do they involve an invasion of the respondents' privacy? Do they provide interpretable information for end users and how do users respond to that information?
■ LESSONS FROM OTHER AREAS OF ENVIRONMENTAL HEALTH Air Quality. In other areas of environmental health, such as air and chemical pollution epidemiology, moving from ecologic or population-level measures of exposure to measures that better reflect exposure at the individual level has been a focus of research for the past two decades. In 1999, for example, the National Research Council Committee on Research Priorities for Airborne Particulate Matter outlined the quantification of the difference between proxy and personal measures of exposure as a key research priority for better understanding differences in observed health risk estimates across individuals and locations. 49 Since then, new approaches have been designed to measure an individual's inhalation exposure to particulate matter in air, using personal breathing zone samplers. For characterizing exposures to cookstove emissions, in particular, new devices, such as the Exposure Child Monitor (ECM) 50 and the Ultrasonic Personal Air Sampler (UPAS), 51 provide more precise measurements for use in public health research applications. Personal samplers for enteric exposure would face significant additional barriers given the diverse agents, the fate and transport of those agents in the environment, their infectious doses, and the myriad of exposure pathways.
Measuring long-term exposures to air pollution, a known driver for a range of chronic adverse health effects, necessitates alternative approaches for characterizing exposure, and often employ hybrid methods that combine both modeling and personal monitoring. A promising, and increasingly common approach for estimating spatiotemporally resolved long-term exposures to particulate matter and several gaseous pollutants comes from satellite remote sensing. 52 These methods have the ability to use satellite optical instrumentation, calibrated with ground-level ambient monitoring data, to create long-term global exposure surfaces. 53 Some water quality parameters can be measured via remote sensing (e.g., chlorophyll-a), but these methods apply more to large water bodies than specific volumes of water consumed by humans. 54 Other methods integrate human activity patterns, questionnaires related to sources of exposure, and measurements conducted within defined settings (i.e., microenvironments) to predict individual level air pollution exposures over short-and long-term periods. The Air Pollution Exposure Model (APEX) is an example of this class of air pollution exposure model, 55 which was developed in response to prior limitations related to air pollution exposure and which may offer insights for novel, combined approaches for characterizing exposures to enteric pathogens as well. Similarly, there are analogous hydrological water quality models that have been developed for estimating exposures to water pollution for surface water bodies. 56 Chemical Toxicants. Chemical toxicant exposure assessments, the measurement of a chemical, or its metabolite, degradate, reaction product or surrogate, include external and internal exposure assessments. External chemical exposure assessments can be pathway-specific (e.g., water, air) and route-specific (e.g., ingestion, inhalation) while internal assessments integrate all pathways and routes of exposure through which a chemical has entered the body. For example, assessment of dermal chemical exposures includes the use of hand wipes, patches, and body suits as dosimeters of exposure. The dosimeters are removed after exposure and chemical concentrations are measured in them, estimating dermal exposure. 57 Similarly, personal air space pumps or patches are used to estimate inhalational exposures and duplicate diet or water measurements may be used to estimate ingestion exposures. Emerging techniques such as the use of silicon wristbands to absorb airborne contaminants have also been used as efficient means to capture exposure to up to 150 air contaminants including polychlorinated biphenyls, pesticides, flame retardants, polycyclic aromatic hydrocarbons and volatile organic chemicals. 58 Internal chemical exposure assessments (i.e., biomonitoring) have burgeoned over the last two decades and are often the "gold standard" methods for assessing exposures in relation to adverse health outcomes. Biomonitoring of chemical contaminants in urine and serum with mass spectrometry-based methods have resulted in many population-based data reports on human internal exposure in the United States, Canada, and Korea. 59−61 Advances in exposomics using high-resolution metabolomics allow for the detection of numerous endogenous and exogenous chemical metabolites simultaneously. 62 Biomonitoring is also seeing application in air pollution exposure assessments in low-income settings, for example with the detection of polycyclic aromatic hydrocarbons in urine to quantify household air pollution exposure. 63 In comparison, internal enteric pathogen exposure assessments face the difficulties of pathogens as dynamic biological organisms that can amplify and die-off inside the host as well as acquired immunity and other modifiers of dose−response relationships.
Measurement Error. Analytical frameworks have been introduced to assess the impact of exposure measurement error, i.e., the difference between the measured and the true exposure, in health effects models. They have shown that some forms of error may lead to biases and greater uncertainties estimating exposure-outcome relationships. 64 Although the nature of measurement error in enteric pathogen exposure assessments may be complex and multifactorial, quantifying sources of error can enable prioritization of how to eliminate the greatest sources of error and better capture biologically relevant exposures.
Prior research on the effects of measurement error on waterborne disease epidemiology has shown that enumeration methods for indicators of fecal contamination can attenuate the relationship between indicator levels and diarrhea by up to 57%. 65 Another source of error occurs when true exposure variability is not sufficiently reflected when using proxy indicators. For example, spatiotemporal rainfall variability can attenuate associations between heavy rainfall and diarrhea by  66 Single measures of water quality can attenuate the relationship between fecal indicator concentrations in water and child linear growth by up to 56% compared to repeated measures that account for temporal variability in water quality. 67 The same study also reported that neglecting exposures to water sources outside of the household can attenuate the water quality-diarrhea associations by up to 21%. 67 Other examples of sources of measurement error include sampling protocol design introducing data collector error (i.e., through nonspecific sampling instructions), assigning household-or community-level exposure measures to individuals, and environmental sample processing errors (i.e., during sample collection, transport, or analyses) that can lead to under-or overestimates of concentrations of enteric pathogens or fecal indicator organisms in environmental media or to false negative or false positive results.

■ METHODS
As part of this workshop, participants collectively identified a set of current and emerging methods, some of which have been widely used in other environmental health sectors, that may contribute to improved measures of enteric pathogen exposure. We describe those methods here and summarize them against the criteria listed above in Supporting Information Tables S1 and S2.
Measuring Enteric Pathogens in the Environment. Studies investigating the association between enteric pathogen exposure and disease, or evaluations assessing the effectiveness of WaSH interventions that include an exposure assessment, have mainly attempted to characterize fecal contamination in different environmental media as a proxy for enteric pathogen exposure, neglecting to characterize human interaction with those media. A systematic review of the effects of sanitation interventions on fecal−oral transmission pathways identified the following approaches used: enteric pathogens or indicator bacteria in environmental samples (drinking water, hands, sentinel toys, food, household and latrine surfaces, and soil); the presence or abundance of flies; and observations of human and animal feces. 68 There are a number of factors to consider when measuring enteric pathogen prevalence in the environment, including environmental sampling strategies, the use of indicators as proxies for enteric pathogens, differentiating between human and animal sources of contamination, detection limits, and selecting which specific pathogens to target.
Sampling Strategies. The vast majority of studies that have measured environmental fecal contamination coupled with child health outcome data have focused on drinking water and, to a smaller extent, fly density, hands, and fomites, with limited data on food and soil contamination. 69 Drinking water samples can be collected directly from the source, from the household storage container or from the utensil used to drink water, each progressively capturing additional pathways of contamination between the source and the point of consumption. Often a choice has to be made between multiple sources of water used by the household, only some of which are dedicated for drinking. Hand contamination has been assessed through rinsing hands in sterile water and analyzing the rinsewater. 70 Food samples can include items prepared and stored at home or bought outside the home (e.g., produce typically eaten raw or streetfood). 71,72 Surfaces can be sampled by swabbing, and soil by scraping topsoil from a designated area. 73 The site of collection of surface swabs or soil is a key decision in sampling protocols. A tool for capturing overall contamination in the domestic environment is sentinel toys, a nonporous object (e.g., plastic ball) that is left in the household for a prespecified amount of time for household members to interact with and then rinsed. 11 Fecal contamination in all of these domains is highly variable temporally, 74−76 seasonally, 77,78 and spatially, 79,80 and additional variability can be introduced by the methods used to analyze samples. 81 Some strategies to address this variability include selecting sampling locations relevant for exposure for the target population (e.g., soil in areas where children spend time), focusing on key times of exposure (e.g., measuring hand cleanliness before eating) and collecting longitudinal samples. Additionally, when the goal of sampling is to assess the relationship between pathogen exposure and health outcomes, environmental samples can be collected prior to ascertaining health end points, allowing an appropriate incubation period before onset of health outcomes. 82−84 The common practice of collecting samples and health data during the same household visit, typically for logistical convenience, risks capturing contamination levels at a time that is not relevant for disease transmission. As a result, this practice is vulnerable to introducing reverse causation (e.g., water boiled in response to illness, fecal contamination in the environment due to diarrhea), 85 obscuring the true relationship between exposure and disease.
Indicators of Fecal Contamination. Fecal contamination in the environment is commonly estimated by using indicators of fecal contamination. These indicators have the advantage that they are easier and less expensive to measure compared to multiple specific pathogens and they can be indicative of a range of enteric pathogens. 86 They provide an indication of the presence of fecal matter in a sample, but do not confirm the presence or absence of pathogens, nor do they provide any indication on the infectivity or diversity of enteric pathogens in a sample. 87 There are a number of indicators of fecal contamination, including chemical indicators (e.g., fecal sterols, caffeine, estrogen hormones) 88,89 as well as microbial indicators.
Fecal indicator bacteria are often grouped into the coliform and streptococcal bacterial groups. Total coliforms include a broad spectrum of bacteria occurring in feces, but can also be found in nonfecal matter. Fecal coliforms or thermotolerant coliforms are a group of bacteria that are more specific to fecal contamination, with the exception of Klebsiella. 90 Escherichia coli (E. coli) is the most commonly found fecal coliform bacteria and is more specific to human and animal fecal matter, though genome comparisons suggest that some clades of E. coli primarily originate from nonhost natural environments, 91 and there are reports of E. coli detection in pristine areas of tropical and even temperate environments. 92−96 In recent years, several laboratory-grade products have been validated in the field as predictive of E. coli. 97,98 However, these instruments require frequent cleaning, are not intended for long-term autonomous operation, are subject to the drawbacks of methods measuring indicators of fecal contamination and are expensive. Recent efforts are focused on reducing these product costs and enabling continuous in situ water quality monitoring. 99 Fecal streptococci were identified as an alternative to total coliform in the 1950s when it became clear that total coliform was a nonspecific indicator for fecal contamination, but the use of this indicator diminished once more specific methods for E. coli culturing were established. 86 Enterococci are a subset of Environmental Science & Technology pubs.acs.org/est Critical Review species of the fecal streptococci group, more specific to fecal contamination than fecal streptococci and more persistent in the environment. 100 Coliphages and crAssphages, types of bacteriophages (viruses that infect bacteria), are also used as indicators of fecal contamination because of their similar morphological characteristics and ability to mimic the persistence of viral pathogens in the environment. 101,102 The use of fecal indicators was historically established to measure fecal contamination in drinking water and recreational waters 86 and to monitor the performance of water treatment processes. In the United States, total coliform is still used to monitor treatment efficacy and post-treatment contamination of drinking water supplies in accordance with the Total Coliform Rule, which requires the monitoring for the presence of total coliform in public water systems at a frequency proportional to the number of people those systems serve. 103 E. coli and fecal coliform are presently the most commonly used fecal indicator bacteria for monitoring fecal contamination in water. The World Health Organization (WHO) uses levels of E. coli and fecal coliform to define microbial quality of drinking water. 104 Meta-analyses of the association of drinking water quality measured using fecal indicator bacteria and diarrheal disease in low-income settings have presented mixed results. Gundry et al. found no significant association between E. coli or fecal coliform in household water and diarrhea, 105 Gruber et al. found a significant association between levels of E. coli in water and diarrhea but not for fecal coliform, 106 whereas Hodge et al., using an individual participant data (IPD) meta-analysis, showed a higher risk of diarrhea with increasing levels of fecal coliform in water. 107 A recent IPD meta-analysis found evidence of association between both E. coli and fecal coliform levels in household drinking water and diarrhea and impaired linear growth. 69 Fecal indicator bacteria are increasingly used to characterize fecal contamination along other fecal−oral transmission pathways. 69 The shortcomings related to indicators of fecal contamination, when used for purposes other than their original intended use of monitoring water supply systems, have been well documented, 108 including that E. coli has been found in environments where fecal contamination is improbable. 109 These shortcomings are amplified by the limited association between indicators and the presence of enteric pathogens. 8,9 However, measurement of fecal indicator bacteria may be valuable as an indication of "total fecal load" in environmental samples and to provide a common metric for comparison with previous studies.
Source Tracking. FIB used in solitude and some other indicators of fecal contamination do not distinguish between different sources of contamination. Fecal source tracking aims to provide data on the source of fecal contamination by detecting signatures of specific sources (e.g., particular animals or particular geographies). In addition to humans as a source of fecal loading in the environment, animal fecal contamination is of particular interest in low-income settings where households often cohabitate with animals in confined spaces. A recent systematic review outlined the importance of animals as a source of fecal contamination to human health, suggesting an association between animal feces exposure and diarrhea, soil-transmitted helminth infection, environmental enteric dysfunction, growth faltering, and trachoma. 14 A variety of source tracking methods have been described in the literature, ranging from fecal coliform/fecal streptococcus ratios to host-specific molecular markers. 110 Host-specific Bacteroidales are commonly used in high-income countries to complement the use of fecal indicators, 111 and are seeing increased use in low-income settings to distinguish between human and animal sources of fecal contamination. 112−114 Bacteroidales can distinguish between sources of fecal contamination, because they adapt to their host differentially, allowing for the identification of host-specific fecal contamination. 115 However, fecal source tracking assays developed in one geographic location must be validated to be used in additional locations, due to geographic variation in human and animal fecal microbiomes, and the molecular methods used to detect Bacteroidales provide no viability or infectivity data. Host-specific Bacteroidales have been measured in Tanzania to characterize human-specific fecal contamination on hands and in drinking water, 116 and test the relationship between humanspecific fecal contamination and diarrhea. 117 A nested study within a cluster-randomized controlled sanitation trial in Odisha, India utilized fecal source tracking to discern the effectiveness of a sanitation intervention on the reduction of human-specific Bacteroidales compared to animal-specific Bacteroidales. 118 Phage-based methods are a rapid and low-cost approach for distinguishing between human and animal feces. GB-124 phages that infect human Bacteroides fragilis have been used in a number of countries to identify human fecal contamination. 119−121 Like the molecular Bacteroidales method, these phage-based methods also need to be validated in each geographic location. Other methods use next generation sequencing data to distinguish between different host microbial communities within a given household or community, 122,123 and use that information to attribute sources of contamination. 124,125 However, these library-dependent methods are still under development, can be expensive and difficult to standardize, and require shipping samples to sequencing facilities. Currently, reference databases used to ascribe taxonomy in these studies are biased toward bacterial communities associated with persons living in high-income communities, but there are continuing efforts to expand these reference databases.
Enteric Pathogen Detection with Traditional Methods. Specific pathogen occurrence or concentration, instead of the use of indicators as proxies, are less commonly measured in the environment although they may be more representative of the actual health risk associated with enteric pathogen exposure. Exposure assessments that measure specific pathogens need to consider not only the diversity of potential pathogens occurring in the environment but relevance of each included pathogen for health outcomes of interest, which is highly context specific. 80 The possibility for improved specificity from measuring specific pathogens instead of indicators of fecal contamination may come at a loss of sensitivity, since selected pathogens may not be representative of all possible pathogens in the environment and are typically present in lower concentrations in the environment.
Monitoring water supplies for pathogens can be prohibitively resource intensive due to the array of pathogens and the low concentrations of specific pathogens in water, requiring sensitive assays. The WHO estimates that concentrations of pathogens in water corresponding to 10 −6 Disability-Adjusted Life Years (DALYs) per person per year are typically less than 1 organism per 10 4 −10 5 liters. 104 Testing for pathogens in water is achieved by concentrating the water sample, such as filtering large quantities of water, on the order of 1−1000 L, Environmental Science & Technology pubs.acs.org/est Critical Review followed by using culture-dependent or culture-independent methods to enumerate the density of pathogens. 6 However, ultrafiltration using kidney dialysis filters is a low-cost method to efficiently and simultaneously concentrate viruses, bacteria, and protozoa from large volumes of water and wastewater. 126−128 Culture-dependent methods are also limited by their low sensitivity and their resource intensity. Furthermore, some enteric pathogens (Salmonella Typhi, Vibrio cholerae, Campylobacter spp., and others) can enter a viable but nonculturable state in the environment that may require special resuscitation steps or molecular methods to detect. While resuscitation or enrichment steps increase the probability of pathogen detection from an environmental sample, they only confirm the presence of the pathogen and do not provide quantitative data. Enteric Pathogen Detection with Molecular Methods. Culture-independent molecular methods have been developed for many enteric pathogens, but these have traditionally required extensive laboratory equipment and highly skilled technical staff. 129 As molecular methods, such as as polymerase chain reaction (PCR) based assays have become lower-cost, easier to multiplex, and more robust to inhibitors in environmental samples, a number of studies have successfully detected bacterial, viral, helminth, and protozoan pathogens in drinking water, on hands, and in soil in low-resource settings. 116,118,130−132 Liu et al. developed the customizable Taqman Array Card (TAC), an emerging method for quantitative detection of multiple enteric pathogens encompassing viral, bacterial, protozoal, and helminth targets for enteric infections. 133 TAC is a Taqman probe based real-time PCR assay which can test up to eight samples with 48 individual singleplex (could be expanded to duplex) reaction wells per sample in a 384-well format. These TAC assays were validated and field tested in many low-and middle-income country settings. Examples of recent environmental applications of the TAC method include the simultaneous detection of a number of enteric pathogens in surface water, soil, and infant weaning food in Kenya. 134,135 Pholwat et al. developed a custom environmental surveillance TAC to identify antimicrobial resistance genes, poliovirus, and enteric pathogens in environmental samples such as fecal sludge, food, sewage, soil, and water in low resource settings. 136 Advantages of TAC are the ability to test 48+ targets/pathogens simultaneously, to obtain quantitative results relatively quickly, and to utilize the same method for both environmental and human samples. An important step in detecting a broad spectrum of targets is to extract high quality total nucleic acid (both DNA and RNA) from the sample. Unlike culture-based methods where pathogen growth is dependent on many factors such as antibiotic history, culture media, and temperature, high throughput molecular diagnostics such as TAC can detect many pathogens quantitatively and be used across wideranging sample types. However, molecular methods cannot be used to establish viability of the organism and can be prohibitively costly. Resource-intensive field collection and lab processes mean these methods are challenging to deploy in low-income settings where resources are limited.
Environmental Metagenomics. Metagenomics, the sequencing and analysis of all DNA in environmental samples, circumvents the problem that many enteric pathogens cannot be easily cultured 137 and metagenomic data can provide information on the abundance and diversity of microorganisms in environmental samples. Unlike many molecular methods, it does not require prespecification of targets, allowing the user to probe for all potential enteric pathogens present in a sample. While it has been historically difficult to identify specific enteric pathogens with amplicon sequencing or low coverage metagenomic approaches (i.e., capturing 16S rRNA genes), shotgun metagenomic approaches (i.e., capturing all metagenomic DNA) can be used to distinguish between specific enteric pathogen strains. Reduced sequencing costs over time as well as recent advances in sequencing technologies and bioinformatics pipelines will continue to open up opportunities for enteric pathogen detection. Long read sequencing technologies, such as Oxford Nanopore Technologies (ONT), enable pathogen identification through assembly or direct alignment of long DNA reads to published pathogen genomes. 138 Metagenomics can be used to characterize both pathogens in environmental samples and in human stool samples. Environmental metagenomics has recently been used to profile viral pathogen diversity in environmental waters, 139 examine antibiotic resistance diversity in wastewater, 140 and to demonstrate exchange of antibiotic resistance genes between soil bacteria and clinical pathogens. 141 Limitations of metagenomics include poor sensitivity if enteric pathogens are at low prevalence in the microbial community or when sequencing depth is low, it does not confirm viability or infectivity of the pathogens, high cost, required bioinformatics expertise, and the need for improved analysis pipelines for identifying pathogens.
Wastewater Surveillance. Wastewater monitoring is an emerging approach to surveillance and has the potential to deepen our understanding of community health. Wastewater has been shown to provide useful community-level information regarding illicit drug use, 142 antimicrobial resistance, 143,144 signals of chronic disease, 145 and advanced warning of viral outbreaks, 146,147 including recent efforts to detect SARS-COV-2 in wastewater. 148 Most prominently, wastewater is used for poliovirus detection in global eradication efforts. 149 Identification of enteric pathogens in wastewater can indicate downstream exposure potential, as well as suggest past exposure to that pathogen among those contributing waste. 23 However, there are challenges with wastewater surveillance: (1) large dilution from other households in the catchment area of pathogens shed by a small number of infected people; (2) need for concentration methods and sensitive detection methods compared to stool samples; (3) need to understand the wastewater collection network, flow, and catchment area in order to identify strategic sample collection points, and (4) differential pathogen transport, die-off or regrowth by wastewater network need to be considered. Dynamic models can be valuable for optimizing the development of a sensitive environmental surveillance system. 150,151 Wastewater surveillance may not capture feces from the poorest populations who are most likely to have the highest burden of infection and exposure because these populations may practice open defecation or use on-site sanitation, and as a population-level approach limits individual-level scientific inference.
Biosensors. Instead of using laboratory intensive approaches to estimating pathogen prevalence in the environment, there is room for innovation to detect enteric pathogens on-site in environmental media by using biosensing technologies. 152 Biosensors isolate DNA encoded biological responses to an electrical signal that can be logged digitally. 153 Such methods have the advantage that they have the potential to be deployed in resource-constrained and remote settings, because they Environmental Science & Technology pubs.acs.org/est Critical Review remove the need for sample collection, transport, and intensive laboratory processing. Bioreceptors can include tissues, microorganisms, enzymes, antibodies, and other biologically derived elements, although most current applications are limited to indicators of fecal contamination rather than specific pathogens. 152 A recent application of biosensors to detect fecal contamination in drinking water used an odorant-binding protein derived from mosquitos to test for the presence of coliform bacteria. 154 Bacteriophage with a broad host range specificity for E. coli have been combined with magnetic beads to capture and then separate E. coli in drinking water. 155 Measuring Enteric Pathogens In Humans. As a proxy measure of internal exposure, infection with enteric pathogens can provide confirmation of actual exposure following ingestion, thus potentially providing important data on past exposure for the evaluation of interventions and for surveillance. Findings from these measures are complicated by host immunity, and absence of infection or serological indication of prior exposure cannot confirm that exposure did not occur. More specifically, these data can be employed to evaluate how well interventions reduce exposure to specific pathogens and which pathogens humans are exposed to in their community. Data generated from surveillance and epidemiological studies has benefited from advances in surveillance capacity and detection methodologies for estimating the burden of disease, disease severity, attributing health outcomes to pathogens, and pathogens to exposure pathways. 156 These data facilitate hazard characterization, the first step of risk assessment.
Dose response data are highly variable by pathogen and are limited because human challenge studies are difficult, expensive, usually single-pathogen focused, primarily performed among adults in high-income countries, cannot control for previous exposures and differences in immune responses to infection, and may be at odds with acceptable ethical standards. 157 Furthermore, little data is available on how much of an ingested dose passes through specific and nonspecific host defense barriers and reaches the target cells and is capable of inducing infection. Animal models often do not exist for enteric pathogens or do not cause the same health outcomes as in humans. Exposure assessments can take into account the hazard characterization information derived from epidemiologic and dose−response studies to focus on pathogens that cause the greatest disease burden in the region of interest, taking a more narrowed approach to exposure assessments by focusing on specific pathogen-source pairs.
Pathogen Shedding in Stool. Methods to detect enteric pathogens in stool samples range from using microscopy 158 or enzyme-linked immunosorbent assays 159 to detect single pathogens, to using molecular or metagenomics methods to characterize multiple pathogens in a sample. Multiplex PCR is a technique that has been widely employed in enteric disease surveillance (sporadic and outbreak) and epidemiological studies such as in multicountry case-control 160 and longitudinal birth cohort studies, 161 as well as in recent studies measuring health impacts of WaSH interventions. 46,162 Shotgun metagenomic approaches have also been employed, for example to distinguish between foodborne disease outbreak strains of Salmonella, 163 and to identify the likely causes of diarrheagenic E. coli in Ecuador. 164 The limitation of these techniques is not only the intensive resources they require, but it can also be difficult to attribute a specific pathogen to a disease outcome when multiple enteric pathogens are detected in stool simultaneously or when asymptomatic infections are common. Additionally, the sensitivity of these methods can vary. These methods identify infections by detecting pathogens in human stool samples, and pathogens may only be shed by an infected person for a short period of time (days, weeks) or shed intermittently, so infections between sampling events may be missed. Furthermore, the duration of shedding after infection is highly pathogen-specific, so these methods can be biased toward persistent pathogens that are shed for a longer period of time compared to more transient pathogens.
Pathogen-Specific Immunoassays. Another way of estimating past exposure to enteric pathogens is through immunological assays detecting pathogen-specific antibodies in serum or saliva, which can be multiplexed to detect exposure to multiple enteric pathogens. 165 For these immunological methods, the timeline of exposure can be difficult to ascribe as low levels of pathogen-specific antibodies (Immunoglobulin (Ig)A, and particularly IgG) can be present in saliva and serum for weeks to years after infection. 166 This can also be an advantage, as the methods can be used to integrate prior exposure over longer periods of time, rather than relying on pathogen shedding in stool. Exposure data without regard to history of infection can be useful for some applications, such as to determine if a population has been exposed to a rare or emerging pathogen or particular microbial strain, which could be important for focusing exposure assessment approaches. Sero-epidemiology is a promising approach to measurement of force of infection of enteropathogens across entire populations. 167−169 However, one's immune response depends on a number of host-specific factors including history of previous exposure (acquired immunity 170 ), nutritional status, 171 genetics, 172 composition of the gut microbiome, 16 underlying disease such as HIV infection, 173 and age (antibodies can appear in low concentrations in young children, particularly in saliva). 174 Measuring Interaction with the Environment. Exposure to enteric pathogens is not only conditional on pathogen presence in the environment, but also on host interaction with that environment. Collecting these data can complement exposure assessments and the analysis of exposure data, as has been done in estimating human exposure to other environmental pollutants. 55 Survey data on self-reported behaviors or observational data on practices can enable the targeting of environmental media and locations where the study population is predominantly exposed. Quantitative observational data can be combined with environmental measurements of enteric pathogens to estimate pathogen ingestion rates. The SaniPath tool, an example of such an approach implemented in nine countries to date, combines environmental sample collection and analyses for E. coli with surveys of behavior to estimate exposure. 72,175 Here we describe current methods that have been used to characterize host interaction with the environment and summarize those methods in Supporting Information Table  S2. When using these methods, it is important to consider the collection of behavior data disaggregated by gender and age group because of the differences in behavior between men and women, boys and girls, and adults and children.
Surveys and Self-Reports. Surveys have been used as a rapid and cost-effective tool to collect information on a range of self-reported behaviors that serve as proxies for exposure patterns. While surveys carry the risk of various types of bias, such as recall bias, courtesy bias, and reporting bias associated with self-report of socially desirable behaviors, 176,177 surveys are nonetheless a useful tool to obtain information on neutral behaviors that do not trigger these biases. There are multiple approaches for collecting self-reported data. Many studies have used household surveys with the head of household or primary caregiver for young children. One major limitation of this approach is that the reported behavior for the respondent is often incorrectly seen as a proxy for behaviors for the entire household. Community participatory surveys and surveys in school classrooms, that combine some discussion of the behavior with a method for the participants in the group to confidentially report their own behavior, such as pocket voting, have also been used to identify high-risk behaviors. 72 Surveys and self-reports can also be used to inform sampling locations (e.g., where a household obtains their drinking water, prepare their food etc.).
Observations. Spot-check observations can capture WaSH infrastructure and behaviors that result in high risk of exposure (e.g., latrine cleanliness, presence of a handwashing station, handwashing at key moments, washing raw produce before consumption) that can be difficult to elicit by self-report due to biased reporting. Structured observations, 178,179 including the use of videography, 180,181 offer an opportunity to gather information on complex behaviors, including recording the frequency, duration, and type of interaction with the environment, which could subsequently be used to estimate ingestion rates. These tools are resource intensive at scale due to high heterogeneity of behavior within-host, i.e., observations of individuals, and between-hosts, i.e., observations of public domains. 182 Observations can also cause reactivity in participants where the presence of an outside observer leads individuals to alter their behaviors while observed. 183 Sensors have been used to compare observed behaviors to reported behaviors with some studies indicating that reported behaviors are inconsistent with sensor measured use. 184−187 Sanitary surveyssurvey-based inspections of water systems or sanitation facilitieshave been designed and promoted, often in combination with periodic water quality testing, as screening and risk assessment tools for fecal contamination exposure. However, recent studies have shown a poor correlation between sanitary inspection scores and fecal contamination in drinking water supplies. 188−190 Location Tracking. Location tracking, for example using personal global positioning system (GPS) tracking, can inform where a host is spending time and can thus provide data on where to collect environmental samples, for example by identifying potential hot spots in communities where community members may be experiencing frequent exposures. The use of GPS devices has broadly been validated for exposure assessments, 191 and has been used to inform air pollution 192 and chemical exposure assessments. 193 A study in Brazil found that GPS tracking was an effective tool to quantify personal movements of urban slum residents and evaluate exposure sources of environmental leptospirosis transmission. 194 Tracers. Tracers, substances introduced into the environment so that their distribution can be detected from their distinctive properties, can provide data on where and how hosts are interacting with their environment. They have been used in air pollution epidemiology to differentiate between indoor and outdoor exposures 195 and a study in China estimated child soil ingestion by measuring concentrations of tracer elements in soil. 196 Challenges with using tracers include that seeding several common fecal−oral transmission pathways simultaneously to quantify relative exposure contributions from different pathways may be impractical beyond certain microenvironments, and tracers are needed that do not degrade in the environment and pose no risk to human or environmental health.

■ MODELING TRANSMISSION PATHWAYS
Modeling tools can complement emerging pathogen diagnostics to provide a more nuanced understanding of both pathogen dynamics in the environment as well as infection transmission dynamics. These tools are used to interpret data, are an inherent part of scientific inquiry and can take on many forms. Statistical models describe the data in the context of sampling variation and to examine associations that inform environmental determinants of disease risk. For example, they are used to estimate pathogen prevalence, persistence and how that may vary in time and space. 197 Conceptual models, sometimes implicit and sometimes explicit, are distinct from statistical models because they represent a hypothesis of the causal relationship between exposure and disease outcome, and are an important driver of how we conduct studies and interpret data. These conceptual models can be codified using mathematics, allowing simulation of different intervention scenarios to conduct thought experiments. Such models are complementary to epidemiological analysis. Their benefits include lower cost than conducting epidemiology studies that involve recruitment of human subjects. They are also mechanistic, reflecting hypothesized or measured quantitative relationships between key variables that describe complex transport (in the environment) and transmission dynamics (from and between individuals, including via animals and the environment), and have clearly articulated assumptions about these relationships that can be evaluated with real-world data.
To study enteric pathogen exposures and risk, there are various types of mathematical models described in the literature. These models allow for in silico experiments and we mention two here. First, quantitative microbial risk assessment (QMRA) models estimate the infection or disease risk from a single pathway as a function of pathogen exposure measurements in the environment. 198 For example, by measuring the contamination in surface water, QMRA can be used to estimate the risk associated with the drinking water pathway and examine the potential for disease reduction given different mediation efforts. 199−203 Importantly, the uncertainty in the exposure measurements, or any other data measurement used in the model, can be propagated and reflected in the risk estimate. Whether the variance of the risk estimate is too high depends on how this estimate is used; in any case, the variance can be lowered through collection of more precise data. Second, infection transmission models dynamically estimate the infection risk associated with multiple pathways of transmission. These models can test scenarios, for example, to understand causes of disease outbreaks from contaminated drinking water, 204 or to explain the interdependency of water, sanitation, and hygiene pathways. 205,206 In general, these pathways are often interdependent; i.e., an increase in soil contamination due to poor sanitation can result in increases in contamination of drinking water sources due to runoff events, or poor hygiene can increase the person-to-person transmission rates that result in more people contaminating water sources used for bathing and washing clothes. Dynamic infection transmission models can explicitly account for these Environmental Science & Technology pubs.acs.org/est Critical Review interdependencies to gain an understanding of how mediation in one or more pathways may affect overall infection rates. The power of these tools is that they can generate hypotheses without the need for collection of empirical data, and can then be tested through empirical studies. Thus, empirical study design can be driven by model simulation results, and models can be further refined by better parametrization with empirical data in an iterative process.
A number of challenges remain in effectively applying these modeling approaches to the study of enteric pathogen exposures. First, modeling results are only as good as the data used to inform parameter estimates. For example, enteric pathogen transmission dynamics are complex and influenced by a range of variables not always measured in field studies. Second, modeling results assume that the model structure is correct. For example, dose−response models are highly uncertain and variable, given the limited data available for deriving these relationships, the artificial setting in which dosing trials are conducted and the many variables that may influence the dose−response relationship. More broadly, a transmission system is complex and assumptions are implicit. For example, our models often assume that humans are the most important source of enteric pathogens, that we have accurately identified all relevant pathways of potential exposure, or that the data we have available for use are representative over space and time.
Many of these limitations apply to science in general and reflect the limitations of the scientific process. Modeling approaches bring a number of advantages to understanding enteric exposures. Models yield data that may be compared with empirical observationincluding epidemiological data and can therefore help elucidate important mechanistic processes at play, potentially explaining the "why" questions that contextualize epidemiological findings. Modeling approaches allow us to interrogate theories of change and identify which interventions have the potential to reduce risks of exposure, and under what conditions. Such methods may allow for more rapid development of technologies and approaches that have the greatest potential for improving public health.

■ RECOMMENDATIONS
We have presented the need for new approaches to improve measures of enteric pathogen exposure, identified principal applications that would benefit from better measures, and have considered the merit of current and emerging methods to be employed for these applications as well as lessons learned from exposure assessments for other environmental contaminants. Based on this review, we have identified a suite of recommendations that we believe are critical to moving enteric pathogen exposure method development forward: 1. Outline the benefits and potential health impacts of improved enteric pathogen exposure assessments. Identify the entities that may harness improved exposure assessments to positively impact health (e.g., governments, international organizations, donors, and funders). Emphasize the role of exposure monitoring with respect to other threats to public health, including climate change and antimicrobial resistance. Catalog opportunities to link improved exposure assessments with other strategies and campaigns, including Sustainable Develop-ment Goal (SDG) six targets on safely managed drinking water and sanitation services. 2. Create a path forward for exposure assessment methods across the principal applications. Develop a vision and the innovations that meet a broad range of applications and criteria necessary to radically improve measures of enteric exposure. Consider novel technologies for use in the field and laboratory. Collaborate across sectors to comprehensively measure exposure by combining modeling, observational, microbiological, epidemiological, and statistical tools. Explore opportunities for integration with One Health, Planetary Health, climate action, nutrition programs, and other initiatives. 3. Dedicate research to defining biologically relevant exposure assessments. Consider biological relevance in exposure assessments as it pertains to health outcomes. Use archived specimens from previous studies to test new methods, where possible, and implement longitudinal studies to provide rigorous evidence on relative contributions from specific transmission pathways and pathogens to adverse health outcomes. Leverage external and internal measures of exposure as complementary approaches for more holistic assessments of exposure. Consider how varying approaches reduce differences between measured and true exposure and their subsequent cost-effectiveness in reducing exposure measurement error to design more biologically relevant exposure assessments. 4. Compile exposure data and fill evidence gaps. Define and standardize best practices for microbiological and observational methods used for exposure assessments and develop a database framework with uniform data reporting standards to enable better comparisons across different studies and settings. Promote the archiving of data and banking of samples for further analysis. Identify and fill evidence gaps, including generating data on the temporal and spatial variability of enteric pathogens in environmental matrices, and neglected fecal−oral transmission pathways such as food and soil, to provide a more complete picture of exposure. 5. Demonstrate the potential market and economic drivers of improved exposure methods. Model the size of the market across various sectors, and describe how exposure monitoring products could be commercialized. Describe the value of these improved methods in higher income settings where there is greater potential to attract investment into research and development while delineating the benefits for improving health in resource-challenged environments. Explore the potential use of exposure data in results-based financing, i.e., payfor-performance by defining performance as reducing exposure.
In this review, we outlined a number of approaches used to characterize exposure to enteric pathogens. Exposure to enteric pathogens is highly context-specific, the density and diversity of pathogens in the environment as well as human interaction with that environment is highly variable over space and time. We presented emerging methods and through lessons learnt from other areas of environmental health distinguished between external and internal measures of exposure. We provided a set of recommendations to narrow the gap between Environmental Science & Technology pubs.acs.org/est Critical Review the measured and the true exposure, including more holistic exposure assessments that consider approaches across sectors to provide more complete measures of exposure.
■ ASSOCIATED CONTENT
Summary of methods to measure enteric pathogens and indicators of fecal contamination (Table S1); summary of methods to characterize interaction with the environment (Table S2) (PDF)