Study design and methods of the Wells and Enteric disease Transmission (WET) Trial: a randomised controlled trial

Abstract

Introduction

The burden of disease attributed to drinking water from private wells is not well characterised. The Wells and Enteric disease Transmission trial is the first randomised controlled trial to estimate the burden of disease that can be attributed to the consumption of untreated private well water. To estimate the attributable incidence of gastrointestinal illness (GI) associated with private well water, we will test if the household treatment of well water by ultraviolet light (active UV device) versus sham (inactive UV device) decreases the incidence of GI in children under 5 years of age.

Methods and analysis

The trial will enrol (on a rolling basis) 908 families in Pennsylvania, USA, that rely on private wells and have a child 3 years old or younger. Participating families are randomised to either an active whole-house UV device or a sham device. During follow-up, families will respond to weekly text messages to report the presence of signs and symptoms of gastrointestinal or respiratory illness and will be directed to an illness questionnaire when signs/symptoms are present. These data will be used to compare the incidence of waterborne illness between the two study groups. A randomly selected subcohort submits untreated well water samples and biological specimens (stool and saliva) from the participating child in both the presence and absence of signs/symptoms. Samples are analysed for the presence of common waterborne pathogens (stool and water) or immunoconversion to these pathogens (saliva).

Ethics

Approval has been obtained from Temple University’s Institutional Review Board (Protocol 25665). The results of the trial will be published in peer-reviewed journals.

Trial registration number

NCT04826991.

Correspondence to Dr Heather M Murphy; [email protected]

Strengths and limitations of this study

• The Wells and Enteric disease Transmission trial is the first randomised controlled trial globally to investigate the attributable impact of untreated private well water consumption on waterborne illness.

• The intervention is a whole-house ultraviolet water treatment system that inactivates waterborne pathogens throughout the residence (including bath water) instead of a point-of-use system placed, for example, at the kitchen tap.

• To reduce recall bias, participants report symptoms through a weekly short message service and illness questionnaire system, which requires prompt response and includes a unique calendar feature to aid symptom and exposure recall.

• Because of sampling feasibility and financial constraints, only a subcohort of participants provide water, stool and saliva samples.

• The scale and spatial heterogeneity of groundwater water samples (270) is a strength of this study.

Introduction

The Safe Drinking Water Act (SDWA) of 1974 was enacted to ensure that public water systems in the USA supply high quality drinking water to the public. However, approximately 40 million people (12%) in the USA rely on water from private wells, which do not fall under the purview of the SDWA.¹ Households with private wells are responsible for monitoring and maintaining their wells. This issue is not unique to the USA as most countries do not regulate private well water, and homeowners bear the responsibility of testing and treating their own water. Research suggests that testing and treatment are infrequent. For example, fewer than half of the private well owners tested their water annually when offered free testing services.² Other USA well water stewardship studies have observed that less than half of the surveyed households reported treating their water even when they are aware that their wells might contain nitrates and arsenic.^{3 4}

A common misconception is that groundwater is free of contaminants, but a growing body of evidence demonstrates that well water can contain chemical and microbial contaminants.^5–9 Well water can be adulterated by agricultural activity, septic systems and other sources.^10–13 Contaminants can enter well water after precipitation and flooding, and intrusion may be facilitated by local hydrogeology.^{11 14 15}

Epidemiological data regarding the effect of microbial content of groundwater on disease in households are scant.¹⁶ The US Environmental Protection Agency estimates that over 16 million annual cases of acute gastrointestinal illness (GI) occur from regulated public water systems, but there are no reliable corresponding estimates for those relying on private wells.¹⁷ It is likely that the burden of waterborne disease is greater for those consuming unregulated private well water. Better estimates of this burden would promote awareness around the risks of private well water consumption and the benefits of treatment. In addition, it would provide evidence for policymakers to recommend interventions, which may include more stringent guidelines/regulations for well construction and septic system management and opt-in, fee-based household treatment services run by local water services that have been proposed outside of the USA (eg, Aquarevo in Australia).¹⁸ Non-urban households relying on private wells are excluded from the health benefits of consuming water from treated public water systems.

The Wells and Enteric disease Transmission (WET) trial attempts to fill this gap in our understanding of the burden of waterborne disease faced by households that rely on private wells. The WET trial is the first randomised controlled trial (RCT) globally to estimate the burden of childhood infectious disease attributed to the consumption of untreated private well water. The intervention involves installation of a commercially available, whole-house ultraviolet (UV) water treatment system, aimed to decrease exposure to waterborne pathogens in drinking water. Disease outcomes will be evaluated in children under 5 years of age, who are at the highest risk for enteric and respiratory infections.

Study aims

Aim 1

• Quantify the incidence rate of endemic childhood GI associated with consuming untreated private well water and compare that to the incidence rate of consuming well water treated by UV.

Aim 1a

• Construct a Quantitative Microbial Risk Assessment (QMRA) using water quality data we collect to estimate the risk of childhood GI associated with consuming untreated private well water and compare the incidence from the risk model to the incidence we calculate in Aim 1.

Aim 2

• Identify, quantify and compare the presence of viral, bacterial and protozoan pathogens in stool of children consuming UV treated or untreated (sham) private well water (including both asymptomatic and symptomatic cases).

Aim 3

• Explore the presence of pathogens in untreated well water and stools of children consuming untreated private well water (sham group only). Analyses for Aim 3 will take place after Aims 1 and 2 are completed so that the biostatistician can be unblinded.

Methods and analysis

The present study is a triple-blinded RCT of the impact of whole-house UV well water treatment on the incidence of gastrointestinal and respiratory illness. The study collects data on the incidence of gastrointestinal and respiratory illness because, while many waterborne pathogens result in GI, infections with some waterborne pathogens (eg, adenovirus, enteroviruses) manifest as respiratory infections. An overview of study activities can be found in figure 1.

Figure 1. Overview of study activities. UV, ultraviolet.

Study region

Participants are being recruited from 11 contiguous Pennsylvania (USA) counties (Berks, Bucks, Carbon, Chester, Delaware, Lancaster, Lebanon, Lehigh, Montgomery, Northampton and Schuylkill) (figure 2). We crudely estimate that more than 1 million people in these counties are served by private wells. Briefly, this estimate was derived using ArcGIS and publicly available maps of public water systems¹⁹ to identify the proportion of each study county outside the boundaries of public water systems and likely relying on private wells. These counties were selected for their density of private wells (average ranges from 15 to 74 wells/square mile), density of on-site septic systems (average ranges from 28 to 70 systems/square mile), diversity of land use (average number of fertilised acres per square mile ranges from 1 to 217 acres/square mile) and geology which can impact well water contamination.^{20 21} The study region could be expanded to neighbouring Pennsylvania (Monroe and York) and New Jersey (Burlington, Camden, Gloucester, Hunterdon, Mercer and Salem) counties if recruitment targets from the primary counties fall short.

Figure 2. Map of study counties in Pennsylvania, USA. Enrolment will begin in the primary counties and, if enrolment numbers in the primary counties are low, will expand into secondary counties. WET, Wells and Enteric disease Transmission.

Inclusion criteria and rationale

• Household must have a private groundwater supply (well or spring) to examine the risks of consuming untreated groundwater from a private well.

• Child residing in household must be 3 years old or younger (maximum age: 3 years and 364 days) at the time of enrolment so that they will be under 5 years of age for the duration of the study period. This is an age group that is particularly vulnerable to gastrointestinal and respiratory infections, including those due to waterborne pathogen exposure.

• Child must be a full-time resident of the home, defined as living in the home for 4 or more days every week. Because the intervention is administered by installation of the device at the home, children must reside in the home regularly to ensure exposure to intervention/sham.

• Child must consume at least one cup of untreated well water daily to ensure exposure to intervention/sham. This includes untreated well water used to make juice or formula.

Exclusion criteria and rationale

• Child has a chronic, gastrointestinal condition (eg, Crohn’s disease), which may complicate differentiation between signs and symptoms of waterborne illness and those of the pre-existing condition.

• Child is immunocompromised, which may result in increased susceptibility to waterborne illness. Children taking daily oral steroids are also excluded due to their compromised immunity.

• Parents/guardians do not have access to short message service (SMS) and are unable to contribute data through text messages (no need for a smartphone, only SMS capabilities).

• Household is currently treating their water for microbiological contamination, which would bias the results of our intervention.

When multiple eligible children are present in the home, only the youngest, eligible child in each household is enrolled. Records will be maintained of eligible participants who decline to participate in the study.

Intervention

The intervention is a commercial, whole-house UV water treatment system (figure 3) operated at a fluence of 50 mJ/cm² (VIQUA VH410, Trojan Technologies, Ontario, Canada). UV disinfection is an established and effective treatment that inactivates >99.9% of all bacteria, protozoa and most viruses in water supplies.^22–24 Similar prior trials that focused on public water systems have implemented UV systems^25–27 given their efficacy and blinding potential—these systems minimise the change in taste and odour of the water supply following treatment, so that households remain blinded to the intervention. The sham device appears identical to the intervention device and differs only in the lamp, which does not emit germicidal UV. Lamps are not visible to participants. Lamps are labelled by the manufacturer with a code for their identity (intervention/sham). At study conclusion, all participants will receive a new UV lamp and quartz sleeve so that all households have an active device.

Figure 3. Picture of device installation at participant’s home. Photograph provided and permission granted by Debbie Lee (author).

Intervention and sham selection

The device was selected by a two-stage process. First, the research team conducted market research to identify and characterise commercially available devices, and then conducted a series of interviews with leading whole-house and point-of-use water treatment suppliers in the USA and installers in Pennsylvania. Following the interviews, we narrowed down our selection to three UV device suppliers based on reported quality of devices and ease of installation. We obtained point-of-use and point-of-entry devices from all three suppliers and subsequently performed microbial challenge tests on two of the three point-of-use devices using MS2 bacteriophage (materials and methods can be found in online supplemental text S1, tables S1 and S2).^28–30 The third device was eliminated from the selection process due to its large size, which could limit the feasibility of installation in many households. All point-of-use devices were rated as capable of achieving fluences of up to 40 mJ/cm² at flow rates of 2–3 gpm but were tested at lower flow rates to achieve fluences of 50 mJ/cm².

We selected a UV dose of 50 mJ/cm² to ensure at least one log (90%) reduction of adenoviruses, organisms in groundwater most resistant to UV.³¹ Tests confirmed that the device provided by VIQUA achieved higher removal of bacteriophage MS2 at 50 mJ/cm² than the other device tested (results can be found in online supplemental text S1, tables S1 and S2). The device is designed to achieve a fluence of 50 mJ/cm² at the end of the lamp life (1 year), therefore the device will be delivering a fluence of higher than 50 mJ/cm² during the trial period.

Participation allocation and blinding

Participants are assigned to either the intervention or the sham device through simple randomisation using a randomisation list generated by the lead biostatistician. Three unblinded individuals who are not involved in any study activities or analyses verify the manufacturer’s codes on the lamps, label the devices by group code and maintain the data required to link group code, manufacturer code and study group. The randomisation code will be safeguarded until the study end by these three individuals.

Recruitment and enrolment

Families are recruited on a rolling basis with the help of community partners: paediatricians; family physicians; daycares; churches; federal, state, local government agencies; and local non-governmental environmental groups. Recruitment strategies include: referrals from partners; flyers and brochures at partner offices and clinics; tabling at community events (eg, county fairs); personalised postcards; and radio, newspaper, search engine and social media advertisements.

Individuals interested in participation are directed to an online eligibility survey or take the eligibility questionnaire over the phone. Individuals are given the opportunity to review the study description and consent form (online supplemental text S2) and to ask study personnel questions before electronically consenting to study participation. Households have the option to submit photos of their well water pressure tank and location of nearest electrical outlet so that the study team can verify that (a) the household does not currently have an ineligible treatment system installed, and (b) installation will be feasible.

Water treatment installers, who are blinded to the intervention, visit the homes of potentially eligible participants to assess the feasibility of installation and instal UV and sham systems. For most participants, the UV devices are installed to treat water for the entire house. In rare instances, only the primary drinking water source (eg, kitchen tap) is treated because of installation necessity and extant premise plumbing infrastructure. During the visit, installers conduct a site survey, survey participants and review well construction and septic records (when available) to collect information regarding the well (construction, depth, age), wastewater management (eg, on-site septic system) and agricultural activities proximal to the home. At study end, participants have the option to replace their own lamp/sleeve or request a replacement service from the installer. For participants who opt to replace their own lamp/sleeve, written consent is obtained. Written consent for submission of biological specimens from children is also obtained for a subcohort of households randomly selected to submit water, stool and saliva samples.

Data collection

Parents and guardians are instructed to collect data for the enrolled child. Following the installation of the device, participants complete an online intake baseline questionnaire, which covers demographics, household composition, domestic animals in and around the home, diaper use, water consumption and medications. At least 7 days after the device is installed, participants begin to receive weekly text messages asking if their child exhibited signs/symptoms of GI or acute respiratory infection (ARI) in the previous week. Prior studies of drinking water interventions have relied on paper-based symptom diaries often returned biweekly or monthly, which can be prone to recall bias.^{27 32} More frequent contact and more prompt response through weekly SMS is likely to reduce such bias. Participants respond ‘Yes’ or ‘No’ to these text messages. When symptoms are reported, participants are texted (or emailed) a link to an online illness questionnaire that requests information on symptom type, severity, onset and duration. The illness questionnaire also includes questions on potential other sources of infection, including travel, food and daycare during the 7-day pre-illness exposure period. The illness questionnaire uses an interactive calendar feature (figure 4) to aid recall. Text message responses and illness questionnaires are compiled in a secure, online dashboard so that select members of the research team (eg, trial manager) can conduct additional follow-up via phone, text or email for participants who have not responded within 48 hours of receiving their initial text or questionnaire link.

Figure 4. Screenshots of the calendar feature in illness questionnaires. Depending on the (A) first day of signs and symptoms reported in the illness questionnaire, participants will be shown a calendar with (B) an exposure period highlighted to promote recall of the 7-day period prior to the onset of signs and symptoms.

Online midpoint and exit questionnaires, similar to the baseline instrument, are administered to assess for changes in child or household characteristics since the baseline period. Additionally, the exit questionnaire requests participants to guess their intervention group so that we can assess the effectiveness of blinding by the James Blinding Index.^{33 34} Midpoint and exit questionnaires are also compiled in the dashboard. All survey instruments can be found in online supplemental text S3–S9.

Sample collection and analysis

A subcohort of 180 households is randomly selected to submit stools and saliva from the enrolled child and one or more untreated groundwater samples (figure 1). This subcohort provides additional consent for the submission of stool and saliva samples.

Water

For all subcohort homes (n=180), one untreated well water sample (ie, water upstream of the device and not subject to treatment) is collected at the beginning of the participant’s study year in either spring or fall, which are periods of groundwater recharge when contamination is more likely. These samples are collected regardless of GI or ARI symptoms. In addition, untreated well water samples (n=90) are collected following the report of a GI or an ARI. We estimate that approximately half of the subcohort (90 of 180) will report symptoms of GI or ARI, so a total of 270 untreated well water samples will be analysed (180 regardless of symptoms and 90 in the presence of symptoms).

Untreated well water is collected from an outdoor tap or a tap on the well water pressure tank. The tap is sterilised and flushed for 5 min before an 800-L sample is collected by ultrafiltration using REXEED 25S Dialyzer (Asahi Kasei Medical, Tokyo, Japan)³⁵ (figure 5). Ultrafilters are shipped on ice overnight to the United States Department of Agriculture (USDA)/United States Geological Survey (USGS) laboratory in Marshfield, Wisconsin, USA, and are backflushed³⁵ and concentrated by polyethylene glycol flocculation.^{36 37} Concentrated samples are stored at −80°C until analysis. Nucleic acids are extracted from concentrated samples, and RNA viruses are reverse-transcribed using random hexamers and SuperScript III. The quantitative PCR (qPCR)/reverse transcription-qPCR is performed for 45 cycles using a LightCycler 480 instrument (Roche Diagnostics) and hydrolysis probes as previously described.¹³ Samples are tested for noroviruses GI and GII, human adenovirus (groups A–F), enterovirus, hepatitis A virus, rotavirus (A and C), diarrheagenic Escherichia coli, Salmonella, Shigella, Campylobacter, Giardia and Cryptosporidium, Shiga toxin-producing bacteria (stx1 and stx2), as well as microbial indicators of human wastewater, human Bacteroides (HF183, HumM2). Primer and probe sequences are reported in online supplemental text S10. Sample results are quantified and reported in genomic copies/litre.

Figure 5. Picture of ultrafiltration set-up for untreated water sample collection. Ultrafilters are connected to either an external hose bibb or pressure tank tap to collect an untreated water sample. Photograph provided and permission granted by Debbie Lee (author).

Negative controls are included for secondary concentration, nucleic acid extraction, reverse transcription and qPCR; these are analysed for all microbial targets and must exhibit no fluorescence above the baseline (ie, no Cq value). Positive controls (bovine herpes virus and bovine respiratory syncytial virus vaccines) are included for nucleic acid extraction and reverse transcription, and gBlocks or Ultramers (IDT) of each target are used as qPCR positive controls and to produce standard curves. All samples are assessed for qPCR and reverse transcription inhibition as described by Borchardt et al.³⁸ All qPCR procedures and controls comply with the EMMI (Environmental Microbiology Minimum Information) guidelines for qPCR.³⁸

In addition, one 1 L and two 250 mL untreated well water samples are collected, transported on ice and analysed at Temple University within 24 hours of receipt. Samples are analysed for total coliforms, E. coli, enterococci and physicochemical parameters.

Stool

Families of the subcohort submit stools from their enrolled child to accompany the water samples, both at baseline and immediately following report of illness. We also ask families to submit one additional stool sample following another report of illness. This sample is not accompanied with a water sample due to budget limitations.

Stool sampling kits are provided and contain instructions, sterile specimen container with storage medium (Zymo DNA/RNA Shield; Zymo Research, Irvine, California, USA), sterile specimen container for samples without storage medium, collection ‘hat’ for toilet-trained children, insulated envelope, prepaid shipping label, gloves, biohazard bags and ice packs. Samples are mailed overnight to researchers at Temple University in Philadelphia. Subsections of neat samples are stored at −80°C. Aliquots of samples in storage medium are shipped on ice to the USDA/USGS laboratory in Marshfield, Wisconsin, USA, and are stored at −80°C until analysis. Nucleic acid extraction, reverse transcription and qPCR analysis are completed as described for water samples,¹³ and pathogens are reported as present/absent. Samples are tested for noroviruses GI and GII, human adenovirus (groups A–F), enterovirus, hepatitis A virus, rotavirus (A and C), SARS-CoV-2, diarrheagenic E. coli, Salmonella, Shigella, Campylobacter, Giardia, and Cryptosporidium and Shiga toxin-producing bacteria (stx1 and stx2); online supplemental text S10 lists assay information.

Saliva

The subcohort also collects monthly saliva swab samples over the 1-year period. Families are directed to rub an Oracol sponge swab (Malvern Medical Developments, Worcester, UK) along the gum of their child. Swabs are stored in participants’ freezers (−20°C) until they are ready to be shipped with stools. Swabs are shipped to Temple University where staff recover saliva from swabs using centrifugation.^{39 40} Saliva samples are stored at −80°C until analysis using multiplex fluorescent microsphere immunoassays using the Luminex xMAP platform (Luminex, Austin, Texas, USA) to quantify IgG responses to specific antigens. Microsphere sets with distinct fluorescence are coupled with target proteins to allow for simultaneous detection of antibodies to multiple targets in samples. Saliva samples are analysed for IgG responses to common waterborne pathogens, including noroviruses GI and GII, Campylobacter jejuni and Cryptosporidium.^39–41

Compliance

Compliance is assessed through regular surveillance of the device. Participating households are directed to check that their device is functioning (ie, the device is powered on) quarterly via email reminders requiring a response, with photos of their device and in particular, the controller screen indicating the status of the device.

Study incentives

Participants are compensated for their participation through US$20 gift cards in the middle and at the end of their study year. Participants in the subcohort receive US$10 gift cards for each stool sample and a report of their untreated water quality at study end. For timely response to text messages and surveys, participants will be entered to win a weekend trip to a local theme park (value US$2000). Three draws will be conducted over the study period. All participants, regardless of the study group, will have a fully functional device at the end of the study that they can keep. The value of the device with installation is estimated at ~US$1000.

Outcome measure (primary)

The weekly text messages and illness questionnaires will be used to quantify the attributable risk of GI by subtracting the incidence rate of incident GI of those consuming well water treated by UV from the incidence rate associated with consuming untreated private well water (primary outcome).

GI

Incident GI is defined by the reporting of a minimum of three episodes of diarrhoea or vomiting over 1 day. Each illness will be considered distinct when separated by ≥6 symptom-free days. Additionally, potential secondary analyses will focus on the incidence of modified highly credible GI, defined as having any of the following symptoms over 1 day: two instances of loose stools, two instances of vomiting or one instance of loose stools with one instance of vomiting. We will also examine illness severity using Modified Vesikari scores (gastroenteritis severity scales), calculated using diarrhoea duration, number of loose stools, vomiting duration, vomiting episodes, fever, healthcare visits and treatment as described by Schnadower et al.⁴²

ARI

Because infections with some waterborne viral pathogens (adenovirus, enteroviruses) manifest as ARIs, samples will also be collected when parents/guardians observe and report children’s symptoms of such infections, defined as: nasal congestion, nasal discharge, sore throat, mouth sores or cough lasting at least 1 day, in the absence of another explanation such as isolated cough with a known history of asthma. Each incident respiratory infection will be considered an event when preceded by ≥6 symptom-free days.

Outcome measure (exploratory)

The presence of waterborne pathogens (adenovirus, enterovirus, hepatitis A virus, norovirus, rotavirus, diarrheagenic E. coli, Campylobacter, Salmonella, Shigella, Cryptosporidium, Giardia) in untreated well water samples and stools from children in the subcohort will be used to carry out Aims 1a (water), 2 (stool) and 3 (water and stool).

Sample size and power analysis

Sample size estimations for the study were based on the primary outcome. These calculations were conducted using simulated estimates that the GI incidence rate in children may vary from 1.2 to 1.6 episodes/child-year^{25–27 43 44} and a detectable reduction of 16–20% in GI incidence rate can be observed in children <5 years of age consuming treated groundwater. For a two-sided test using the Variance Stabilised test statistic, samples of 399 subjects in each group with exposure time of 1.0 year will achieve 80% power to detect an incidence rate ratio of 0.84 when the incidence rate in consuming untreated groundwater group is 1.45 and the significance level (two-sided alpha) is 0.05. The attrition rate for sample size calculations is estimated at 14%, which is greater than the rate from our pilot study (28 participants; rate=0%). Using this conservative estimate for attrition, we will enrol 454 individuals in each group to ensure 399 subjects per group for final analyses to attain our desired level of power. A total of 908 participants will be enrolled in the study.

Data analysis

Descriptive statistics will be computed for all variables to ensure data quality and evaluate assumptions of statistical tests. We will compare the baseline characteristics between the sham and intervention groups in the trial to assess the success of randomisation in producing two comparable groups. Although we do not expect imbalances between the intervention and sham groups, we will adjust for any differences in analyses to be sure any intervention effect is not attributed to potential confounders. All subsequent testing of intervention effects will be based on an intent-to-treat analysis, including participants who do not adhere to study protocols.

Aim 1

Quantify the incidence rate of GI associated with consuming untreated private well water and compare that to the incidence rate of consuming UV treated well water. The primary outcome of this work is to compare the incidence rate (cases/child-year) of GI (attributable risk) over a 1-year interval in the two treatment groups. The intervention effect will be tested using a Poisson regression model with offsets to model the natural logarithm of GI cases per child-days at-risk follow-up for 1 year. Both crude (covariate unadjusted) and covariates-adjusted incidence rates will be estimated and compared between the two groups. Possible covariates include any baseline characteristics (including potential drivers of well contamination) that differ between the intervention and sham groups after random assignment at a liberal p value of 0.10. All models will include an overdispersion component. An alternative negative-binomial regression model to account for overdispersion will be used if the variance significantly exceeds the mean.

Aim 1a

Use QMRA to quantify the risk of GI associated with consuming untreated private well water and compare to Aim 1. Using approaches similar to other QMRA models of risk from well water consumption,^45–47 a multi-pathogen QMRA model will be developed using the untreated well water quality data collected from the subcohort of 180 households (270 samples). The following pathogens will be included in the QMRA: adenovirus, enterovirus, norovirus, diarrheagenic E. coli, Campylobacter, Salmonella, Shigella, Cryptosporidium, Giardia. The model will be developed before completing the final analysis of Aim 1 to allow for true unblinded comparison of the results. Individual models will be developed for each pathogen detected using established pathogen-specific dose-response curves. The concentration of pathogens found in the well water in our study will be fit to probability distributions and are the key inputs into the mathematical models. The model endpoint will be the annual number of cases/child-year for each pathogen. The results of the different pathogen-specific models will be combined to produce a total number of cases/child-year of GI expected in the 908 children studied in the trial. The final estimates will be reported as means with 90% probability intervals (PI) around the mean generated using Monte Carlo simulation (10 000 iterations using @Risk (Palisade, USA)). The PIs from the QMRA will be compared with the 90% CIs of intervention effect from Aim 1 to assess whether these intervals intersect.

Aim 2

Identify, quantify and compare the presence of viral, bacterial and protozoan pathogens in stools of those consuming treated and untreated private groundwater supplies. The exploratory outcome is to compare the presence of pathogens in the stools of children consuming treated and untreated water. This analysis will be performed using the stools collected from the subcohort of children (90 from UV and 90 from sham groups). The intervention effect will be tested using mixed effects logistic regression models. In each model (presence of one specific pathogen in stool), child participants will be included as random effects while the intervention effect will be evaluated as a fixed effect. The covariance structure within participants will be specified using an unstructured model to account for in-participant correlation over time.⁴⁸

The comparison of stool and water samples from children drinking untreated groundwater is an exploratory aim (Aim 3). The analysis for Aim 3 will be performed after Aims 1 and 2 have been completed. At this stage, the statistician will be unblinded and use the water quality and stool data from the 90 sham households only. Cohen’s Kappa statistics⁴⁹ will be calculated to quantify the degree of agreement of each pathogen’s presence between collected water and stools. The same approach will be used to compare the presence of pathogens in the well water with the pathogen-specific antibodies in the saliva samples.

Participant and public involvement

Study participants and the public were not involved in the design, conduct, reporting or dissemination plans of this research. Study participants do not assess the burden of the intervention themselves.

Strengths

This innovative study is the first RCT to estimate the burden of waterborne disease attributed to drinking untreated private well water. One strength of this study is the provision of a whole-house water treatment system (vs a point-of-use system) to cover all potential within-house water exposures (eg, bath water) for the children. Another strength is the implementation of weekly text message-based illness reporting with an interactive calendar feature to aid families with recall (figure 4).

The analysis of groundwater, stool and saliva samples from the subcohort will represent a novel contribution to the understanding of the influence of private well water quality on child health. The concurrent collection of untreated groundwater and stools will make possible a previously unexplored examination of the relationship between the presence of pathogens in well water and in children’s stool. The many viral, bacterial and protozoan pathogens (n=11) included for groundwater analysis is also novel, as studies of pathogen occurrence in groundwater typically include only two or three pathogens, with viruses tested more often than bacterial or protozoan pathogens.^{8 16} Furthermore, this study will implement an innovative multiplex method of using antibody response to waterborne pathogens in saliva samples as a biomarker of infection in children.

Additionally, the water samples collected through the study will allow for a QMRA alongside the large-scale epidemiological study. Few studies have directly compared epidemiological data to a QMRA.⁵⁰ If the burden estimates from the QMRA and RCT are similar, the results of this study would provide support for the use of QMRA for estimating the burden of disease attributed to private well water contamination.⁴⁵ Moreover, most groundwater QMRAs use assumed pathogen values rather than measurements,^{46 51} and few studies that use pathogen measurements consider more than one pathogen.^{45 47 52} Our QMRA is based on 270 measurements of each pathogen, so risk estimated by QMRA represents many of the viral, bacterial and protozoan pathogens that can cause the GI recorded by our epidemiological observations, allowing a robust comparison.

Another trial strength is the diversity of drivers of groundwater contamination in the study counties. The study region is localised to the counties surrounding Philadelphia, Pennsylvania, USA, but comprises counties with a wide range of geology, land use, contaminant sources and septic system density. The subcohort of participants is randomly selected from these counties, and thus, the samples from these households will be accordingly diverse.

Limitations

One limitation of this trial is the reliability of self-reported disease data. This study relies on a shorter recall period (7 days) for parents/guardians than the recall period employed in many prior RCTs (weekly vs biweekly or monthly). However, parents/guardians may still find it difficult to recall specific details regarding illness even in the previous week. Capturing detailed illness histories may be further complicated by the presence of multiple caretakers (family members and daycare). Furthermore, parents/guardians are asked to report symptoms and potential exposures that occurred the week prior to illness onset. Hence, they may be asked about details extending back as far as 2 weeks, which may make recall challenging. Even so, compared with paper-based diaries, the use of weekly text message-based illness reporting with prompt follow-up potentiates the collection of illness data less impacted by recall bias.

While the WET trial is the largest RCT of its kind globally, only a randomly selected subcohort of 180 participating families will submit stool, saliva and groundwater samples because of budget restrictions. The size of this subcohort and the geographical limitation (area surrounding Philadelphia) presents a challenge to achieving spatial and temporal heterogeneity in groundwater quality. Even so, the groundwater sample size of the WET trial with 180 wells (270 samples) is similar to, if not greater than, many well water studies globally that have examined pathogen presence.^{8 13}

Ethics and dissemination

Ethics review and approval was provided by Temple University’s Institutional Review Board (Protocol 25665). Study protocols will be made publicly available through Open Science Framework. The results of this study will be published in peer-reviewed journals.

The authors would like to thank and acknowledge the support of the advisory panel during the development and ongoing execution of the trial. We would like to thank the following advisory panel members: Drs Jack Colford, Tim Wade, Tom Clasen, Lori Holtz, Greg Storch and Karl Linden. We would also like to thank Dr Joe Brown, Joseph Wallace, Lisa Reynolds Fogarty and Meredith Nevers for their review of the protocol. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the US Government.

Ethics statements

Patient consent for publication

Not applicable.

Footnote

Contributors DL contributed to the design of the work, the development of protocols for data and sample (water, stool, saliva) collection, management and analysis, and drafting the manuscript. DD contributed to the design of the work and the development of survey instruments and protocols for stool sample collection and management. PT contributed to the design of the work and the development of protocols for stool sample collection and management. JW contributed to the design of the work and the development of survey instruments and protocols for randomisation and survey data analysis. JPS contributed to the design of the work, the development of protocols for water and stool sample collection, management and analysis. MB contributed to the design of the work and the development of protocols for water and stool sample collection, management and analysis. HMM conceived of this work and substantially contributed to study design, the development of protocols for data and sample collection, management and analysis, and drafting this manuscript. All authors made substantial contributions to the design of the work and the development of protocols pertaining to the acquisition and analysis of data for this work. All authors contributed to critically revising this manuscript, approved the final manuscript and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

Funding This work was supported by National Institutes of Allergy and Infectious Diseases (NIAID) of the National Institutes of Health (NIH) under award numbers: R34AI127305, R01AI153376. The work was also supported by the Pennsylvania Department of Health’s Pennsylvania Commonwealth Universal Research Enhancement Program (PA CURE) Formula Funds 2019 (Grant # 4100083099 to Temple University). The ultraviolet water treatment systems in the study were donated by London, Ontario, Canada based company, Trojan Technologies. None of the sponsors or funders have had any role or authority over the study design, data collection and management, nor will they have any role or authority in the analysis, interpretation of the data or publication of the trial results. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Map disclaimer The inclusion of any map (including the depiction of any boundaries therein), or of any geographical or locational reference, does not imply the expression of any opinion whatsoever on the part of BMJ concerning the legal status of any country, territory, jurisdiction or area or of its authorities. Any such expression remains solely that of the relevant source and is not endorsed by BMJ. Maps are provided without any warranty of any kind, either express or implied.

Competing interests None declared.

Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Provenance and peer review Not commissioned; peer reviewed for ethical and funding approval prior to submission.

Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

References

¹ Johnson TD, Belitz K, Lombard MA. Estimating domestic well locations and populations served in the contiguous U.S. for years 2000 and 2010. Sci Total Environ 2019; 687: 1261–73. doi:10.1016/j.scitotenv.2019.06.036

² Jones AQ, Dewey CE, Doré K, et al. Public perceptions of drinking water: a postal survey of residents with private water supplies. BMC Public Health 2006; 6: 94. doi:10.1186/1471-2458-6-94

³ Flanagan SV, Marvinney RG, Johnston RA, et al. Dissemination of well water arsenic results to homeowners in central maine: influences on mitigation behavior and continued risks for exposure. Sci Total Environ 2015; 505: 1282–90. doi:10.1016/j.scitotenv.2014.03.079

⁴ Lewandowski AM, Montgomery BR, Rosen CJ, et al. Groundwater nitrate contamination costs: a survey of private well owners. J Soil Water Conserv 2008; 63: 153–61. doi:10.2489/jswc.63.3.153

⁵ Abbaszadegan M, Lechevallier M, Gerba C. Occurrence of viruses in US groundwaters. J Am WATER Work Assoc 2003; 95: 107–20. doi:10.1002/j.1551-8833.2003.tb10458.x

⁶ Fout GS, Borchardt MA, Kieke BA, et al. Human virus and microbial indicator occurrence in public-supply groundwater systems: meta-analysis of 12 international studies. Hydrogeol J 2017; 25: 903–19. doi:10.1007/s10040-017-1581-5

⁷ Lee D, Murphy HM. Private wells and rural health: groundwater contaminants of emerging concern. Curr Environ Health Rep 2020; 7: 129–39. doi:10.1007/s40572-020-00267-4

⁸ Hynds PD, Thomas MK, Pintar KDM. Contamination of groundwater systems in the US and Canada by enteric pathogens, 1990-2013: a review and pooled-analysis. PLoS One 2014; 9: e93301. doi:10.1371/journal.pone.0093301

⁹ Schaider LA, Rudel RA, Ackerman JM, et al. Pharmaceuticals, perfluorosurfactants, and other organic wastewater compounds in public drinking water wells in a shallow sand and gravel aquifer. Sci Total Environ 2014; 468–469: 384–93. doi:10.1016/j.scitotenv.2013.08.067

¹⁰ Schaider LA, Ackerman JM, Rudel RA. Septic systems as sources of organic wastewater compounds in domestic drinking water wells in a shallow sand and gravel aquifer. Sci Total Environ 2016; 547: 470–81. doi:10.1016/j.scitotenv.2015.12.081

¹¹ Invik J, Barkema HW, Massolo A, et al. Escherichia coli contamination of rural well water in alberta, canada is associated with soil properties, density of livestock and precipitation. Can Water Resour J / Rev Can Des Ressources Hydriques 2019; 44: 248–62. doi:10.1080/07011784.2019.1595157

¹² Borchardt MA, Bradbury KR, Alexander EC, et al. Norovirus outbreak caused by a new septic system in a dolomite aquifer. Ground Water 2011; 49: 85–97. doi:10.1111/j.1745-6584.2010.00686.x

¹³ Borchardt MA, Stokdyk JP, Kieke BA, et al. Sources and risk factors for nitrate and microbial contamination of private household wells in the fractured dolomite aquifer of northeastern wisconsin. Environ Health Perspect 2021; 129: 67004. doi:10.1289/EHP7813

¹⁴ Uejio CK, Christenson M, Moran C, et al. Drinking-water treatment, climate change, and childhood gastrointestinal illness projections for northern Wisconsin (USA) communities drinking untreated groundwater. Hydrogeol J 2017; 25: 969–79. doi:10.1007/s10040-016-1521-9

¹⁵ Murphy HM, McGinnis S, Blunt R, et al. Septic systems and rainfall influence human fecal marker and indicator organism occurrence in private wells in southeastern Pennsylvania. Environ Sci Technol 2020; 54: 3159–68. doi:10.1021/acs.est.9b05405

¹⁶ Murphy HM, Prioleau MD, Borchardt MA, et al. Review: epidemiological evidence of groundwater contribution to global enteric disease, 1948–2015. Hydrogeol J 2017; 25: 981–1001. doi:10.1007/s10040-017-1543-y

¹⁷ Messner M, Shaw S, Regli S, et al. An approach for developing a national estimate of waterborne disease due to drinking water and a national estimate model application. J Water Health 2006; 4 Suppl 2: 201–40. doi:10.2166/wh.2006.024

¹⁸ South East Water Aquarevo. Available: https://southeastwater.com.au/residential/upgrades-and-projects/projects/aquarevo [Accessed 8 Sep 2022 ].

¹⁹ Pennsylvania geospatial data access public water supplier’s (PWS) service areas. n.d. Available: https://www.pasda.psu.edu/uci/DataSummary.aspx?dataset=1090

²⁰ Fleeger GM. The geology of Pennsylvania’s groundwater. In: Pennsylvania Geological Survey. 3rd ed. Harrisburg, Pennsylvania: 4th ser., Educational Series 3, 1999.

²¹ Pennsylvania Water Science Center. Major aquifer material types in Pennsylvania. 2007. Available: https://www.usgs.gov/media/images/major-aquifer-material-types-pennsylvania

²² Linden KG, Shin G-A, Faubert G, et al. UV disinfection of Giardia lamblia cysts in water. Environ Sci Technol 2002; 36: 2519–22. doi:10.1021/es0113403

²³ Hijnen WAM, Beerendonk EF, Medema GJ. Inactivation credit of UV radiation for viruses, bacteria and protozoan (oo) cysts in water: a review. Water Res 2006; 40: 3–22. doi:10.1016/j.watres.2005.10.030

²⁴ Chang JC, Ossoff SF, Lobe DC, et al. UV inactivation of pathogenic and indicator microorganisms. Appl Environ Microbiol 1985; 49: 1361–5. doi:10.1128/aem.49.6.1361-1365.1985

²⁵ Borchardt MA, Spencer SK, Kieke BA, et al. Viruses in nondisinfected drinking water from municipal wells and community incidence of acute gastrointestinal illness. Environ Health Perspect 2012; 120: 1272–9. doi:10.1289/ehp.1104499

²⁶ Colford JM, Wade TJ, Sandhu SK, et al. A randomized, controlled trial of in-home drinking water intervention to reduce gastrointestinal illness. Am J Epidemiol 2005; 161: 472–82. doi:10.1093/aje/kwi067

²⁷ Hellard ME, Sinclair MI, Forbes AB, et al. A randomized, blinded, controlled trial investigating the gastrointestinal health effects of drinking water quality. Environ Health Perspect 2001; 109: 773–8. doi:10.1289/ehp.01109773

²⁸ MaCkey ED, Hargy TM, Wright HB, et al. Comparing Cryptosporidium and MS2 bioassays-implications for UV reactor validation. J Am WATER Work Assoc 2002; 94: 62–9. doi:10.1002/j.1551-8833.2002.tb09407.x

²⁹ Beck SE, Rodriguez RA, Hawkins MA, et al. Comparison of UV-induced inactivation and RNA damage in MS2 phage across the germicidal UV spectrum. Appl Environ Microbiol 2015; 82: 1468–74. doi:10.1128/AEM.02773-15

³⁰ Park GW, Linden KG, Sobsey MD. Inactivation of murine norovirus, feline calicivirus and echovirus 12 as surrogates for human norovirus (NoV) and coliphage (F+) MS2 by ultraviolet light (254 nm) and the effect of cell association on UV inactivation. Lett Appl Microbiol 2011; 52: 162–7. doi:10.1111/j.1472-765X.2010.02982.x

³¹ Yates MV, Malley J, Rochelle P, et al. Effect of adenovirus resistance on UV disinfection requirements: a report on the state of adenovirus science. J Am WATER Work Assoc 2006; 98: 93–106. doi:10.1002/j.1551-8833.2006.tb07686.x

³² Colford J, Wade T, Sandhu S, et al. A randomized, blinded, crossover trial of an in-home drinking water intervention to reduce gastrointestinal illness. Epidemiology 2003; 14: S137–8. doi:10.1097/00001648-200309001-00342

³³ Bang H, Ni L, Davis CE. Assessment of blinding in clinical trials. Control Clin Trials 2004; 25: 143–56. doi:10.1016/j.cct.2003.10.016

³⁴ James KE, Bloch DA, Lee KK, et al. An index for assessing blindness in a multi-centre clinical trial: disulfiram for alcohol cessation -- a VA Cooperative study. Stat Med 1996; 15: 1421–34. doi:10.1002/(SICI)1097-0258(19960715)15:13<1421::AID-SIM266>3.0.CO;2-H

³⁵ Smith CM, Hill VR. Dead-end hollow-fiber ultrafiltration for recovery of diverse microbes from water. Appl Environ Microbiol 2009; 75: 5284–9. doi:10.1128/AEM.00456-09

³⁶ Lambertini E, Spencer SK, Bertz PD, et al. Concentration of enteroviruses, adenoviruses, and noroviruses from drinking water by use of glass wool filters. Appl Environ Microbiol 2008; 74: 2990–6. doi:10.1128/AEM.02246-07

³⁷ Mull B, Hill VR. Recovery of diverse microbes in high turbidity surface water samples using dead-end ultrafiltration. J Microbiol Methods 2012; 91: 429–33. doi:10.1016/j.mimet.2012.10.001

³⁸ Borchardt MA, Boehm AB, Salit M, et al. The environmental microbiology minimum information (EMMI) guidelines: qPCR and dPCR quality and reporting for environmental microbiology. Environ Sci Technol 2021; 55: 10210–23. doi:10.1021/acs.est.1c01767

³⁹ Egorov AI, Griffin SM, Ward HD, et al. Application of a salivary immunoassay in a prospective community study of waterborne infections. Water Res 2018; 142: 289–300. doi:10.1016/j.watres.2018.05.030

⁴⁰ Wade TJ, Griffin SM, Egorov AI, et al. Application of a multiplex salivary immunoassay to detect sporadic incident norovirus infections. Sci Rep 2019; 9: 19576. doi:10.1038/s41598-019-56040-7

⁴¹ Griffin SM, Chen IM, Fout GS, et al. Development of a multiplex microsphere immunoassay for the quantitation of salivary antibody responses to selected waterborne pathogens. J Immunol Methods 2011; 364: 83–93. doi:10.1016/j.jim.2010.11.005

⁴² Schnadower D, Tarr PI, Gorelick MH, et al. Validation of the modified vesikari score in children with gastroenteritis in 5 US emergency departments. J Pediatr Gastroenterol Nutr 2013; 57: 514–9. doi:10.1097/MPG.0b013e31829ae5a3

⁴³ Payment P, Richardson L, Siemiatycki J, et al. A randomized trial to evaluate the risk of gastrointestinal disease due to consumption of drinking water meeting current microbiological standards. Am J Public Health 1991; 81: 703–8. doi:10.2105/ajph.81.6.703

⁴⁴ Payment P. Epidemiology of endemic gastrointestinal and respiratory diseases: incidence, fraction attributable to tap water and costs to Society. Water Sci Technol 1997; 35: 7–10. doi:10.2166/wst.1997.0701

⁴⁵ Burch TR, Stokdyk JP, Spencer SK, et al. Quantitative microbial risk assessment for contaminated private wells in the fractured dolomite aquifer of Kewaunee County, Wisconsin. Environ Health Perspect 2021; 129: 67003. doi:10.1289/EHP7815

⁴⁶ Murphy HM, Thomas MK, Schmidt PJ, et al. Estimating the burden of acute gastrointestinal illness due to Giardia, Cryptosporidium, Campylobacter, E. coli O157 and norovirus associated with private wells and small water systems in Canada. Epidemiol Infect 2016; 144: 1355–70. doi:10.1017/S0950268815002071

⁴⁷ Burch TR, Stokdyk JP, Rice N, et al. Statewide quantitative microbial risk assessment for waterborne viruses, bacteria, and protozoa in public water supply wells in Minnesota. Environ Sci Technol 2022; 56: 6315–24. doi:10.1021/acs.est.1c06472

⁴⁸ Fitzmaurice GM, Laird NM, Ware JH. Applied longitudinal analysis. In: Applied Longitudinal Analysis. 2nd ed. Hoboken, NJ, USA: John Wiley & Sons, 2011. doi:10.1002/9781119513469

⁴⁹ Cohen J. A coefficient of agreement for nominal scales. Educ Psychol Meas 1960; 20: 37–46. doi:10.1177/001316446002000104

⁵⁰ Burch T. Validation of quantitative microbial risk assessment using epidemiological data from outbreaks of waterborne gastrointestinal disease. Risk Anal 2019; 39: 599–615. doi:10.1111/risa.13189

⁵¹ United States Environmental Protection Agency. Economic analysis for the final ground water rule EPA-815-R-06-014. Washington, D.C: Office of Water, 2006.

⁵² Rodriguez-Alvarez MS, Weir MH, Pope JM, et al. Development of a relative risk model for drinking water regulation and design recommendations for a peri urban region of Argentina. Int J Hyg Environ Health 2015; 218: 627–38. doi:10.1016/j.ijheh.2015.06.007