Seasonality of drinking water sources and the impact of drinking water source on enteric infections among children in Limpopo, South Africa

Enteric infections and water-related illnesses are more frequent during times of relative water abundance, especially in regions that experience bimodal rainfall patterns. . However, it is unclear how seasonal changes in water availability and drinking water source types affect enteric infections in young children. This study investigated seasonal shifts in primary drinking water source type and the effect of water source type on enteric pathogen prevalence in stool samples from 404 children below age 5 in rural communities in Limpopo Province, South Africa. From wet to dry season, 4.6% (n=16) of households switched from a source with a higher risk of contamination to a source with lower risk, with the majority switching to municipal water during the dry season. In contrast, 2.6% (n=9) of households switched from a source with a lower risk of contamination to a source with higher risk. 74.5% (n=301) of the total households experienced interruptions in their water supply, regardless of source type. There were no significant differences in enteric pathogen prevalence between drinking water sources. Intermittent municipal water distribution and household water use and storage practices may have a larger impact on enteric infections than water source type. The limited differences in enteric pathogen prevalence in children by water source could also be due to other exposure pathways in addition to drinking water, for example through direct contact and food-borne transmission.


Introduction
Water quality varies greatly across South Africa, with rural areas being the most vulnerable to poor water access and/or quality (Meissner et al., 2018;Mellor et al., 2012). Health outcomes of poor drinking water quality include enteric infections, diarrhea, childhood stunting, anxiety and depression, and death (Abia et  Households may change their water source due to seasonal variations in rainfall and water availability or maintenance and governance issues (Bisung et al., 2015;Dos Santos et al., 2017;Heleba, 2011;Majuru et al., 2012). Therefore, access to water can vary across seasons and throughout the year. Differences in water source types used in the wet versus dry seasons may have different health implications and contribute to the seasonality of enteric infections observed in many contexts (Lal et al., 2012), such as the seasonality of Shigella in Limpopo, South Africa (Rogawski McQuade et al., 2020). For example, there may be an increase in infections in times of relative water abundance or the wet season (Ashbolt, 2015). This can be due to reductions in water quality, such as increased chemical or microbial contamination from runoff (Ngoye & Machiwa, 2004;Ouyang et al., 2006). In contrast, during times of water scarcity or the dry season, lack of water may negatively influence health due to unsanitary conditions, inadequate water availability for consumption and hygiene, and the reliance on alternative sources of water. Alternative water sources can increase the risk of water-related diseases if they are of lower water quality (Bradley & Bartram, 2013). Low-income households are the most vulnerable during periods of low water abundance (Majuru, et al., 2016). Additionally, seasonality can affect resource mobilization and external support availability in communities (Holmatov et al., 2017;Kelly et al., 2018).
The risk of fecal contamination varies by water source type (Kostyla et al., 2015). Improved water sources include piped water, public tap or standpipe, borehole, protected dug well or spring, and rainwater. In contrast, unimproved water sources include unprotected dug well or spring, tanker truck, and surface water from rivers, dams, lakes, ponds, streams, canals, or irrigation channels (WHO & UNICEF, 2016). Unimproved water sources have significantly higher risk of being contaminated compared to improved water sources (Shields et al., 2015;Kumpel & Nelson, 2016). The use of water sources with higher contamination during times of lower water abundance is one reason why water-related disease prevalence may differ across seasons (Pearson et al., 2016).
We previously assessed total coliform bacteria and E. coli in local water sources (Edokpayi et al., 2018) and samples of drinking water from households (manuscript under review) within a community in Limpopo, South Africa. We documented high levels of contamination of water sources in this area, which changed seasonally. Here, we further analyze data collected during this study to describe changes in primary water sources of households across wet and dry seasons and investigate how water source type affects the risk of enteric infections in children under the age of 5 in this community.

Procedures
We analyzed data from a randomized controlled trial of the impact of low-cost point-of-use water treatment technologies on enteric infections and linear growth among children in Limpopo, South Africa. The study design and primary results have been previously reported (Hill et al., 2020). Briefly, the study enrolled one child under the age of 3 from 404 households in rural villages from the Dzimauli community in Limpopo, South Africa. Participant households were recruited from 18 villages and completed a baseline questionnaire between June-November 2016. Smaller villages were combined into two larger groups (n=40 in Group 1 and n=64 in Group 2) for data analysis. The remainder of villages were analyzed by village ( Figure 1).
The baseline questionnaire included questions relating to demographics, primary drinking water sources, and water use practices of the child and their household. Primary drinking water sources were classified into 5 categories: municipal, surface water from tap/pipe, directly from surface, groundwater, and other/unknown. The five water source types are defined in Table 1. Mutale Water Works is the municipal water treatment plant for the communities in this study, as shown in Figure 1. However, not all households necessarily receive and use municipal water due to inadequate infrastructure and intermittent supply. Demographics recorded for each child included age, sex, and socioeconomic status, defined by the WAMI index, a score based on water and sanitation access, asset ownership, maternal education, and average monthly household income (Psaki et al., 2014). Additionally, the primary caregivers' demographics were collected, including their mother's age and education level. Water use practices included drinking water storage type and treatment. Drinking water storage types included metal and plastic buckets, jerry cans, and plastic bottles, while typical drinking water treatment included letting the water stand and settle, adding chlorine, boiling, and no treatment.
Field researchers conducted follow-up questionnaires and collected a stool sample from the youngest child under 3 years of age every 3 months. During each visit, a questionnaire was given to the child's caregiver to collect information on primary water sources, water use practices, and recent health information about the child, including diarrhea in the past 7 days. Stool samples were stored at −70°C before testing, which was conducted in batches every 6 months. DNA was extracted from stool using the Qiagen QIAamp Fast DNA Stool MiniKit (Qiagen, Valencia, VA, USA) and tested for enteropathogens using multiplex realtime PCR, as previously described (Hill et al., 2020). Targeted genes for amplification identified enteroaggregative E. coli (EAEC), enterohemorrhagic E. coli/enteropathogenic E. coli (EHEC/EPEC), Giardia, Campylobacter, Cryptosporidium, enterotoxigenic E. coli (ETEC), Shigella, and adenovirus (Liu et al., 2014). Detections at a PCR cycle threshold of <35 were considered positive as previously (Hill et al., 2020).

Data Analysis
All analyses were conducted using SAS software (version 9.4). We categorized the wet season from November to March, and the dry season from April to October. We categorized drinking water source for each household by season and assessed changes in primary water source between wet and dry seasons. In a minority of cases, households used more than 2 types of water sources during one wet/dry season, in which case, the water source for that season was categorized as "mixed." If a change in water source occurred between seasons, the change was noted if it was to a water source with a higher risk or lower risk of contamination. "Municipal" and "groundwater" were classified as low risk, improved water sources while "surface water from tap/pipe," directly from surface water," and "Other/ unknown" was classified as high risk, unimproved water sources (WHO & UNICEF, 2016).
We estimated prevalence ratios for the effects of primary water source on the prevalence of enteric infections using log-binomial regression. Poisson regression was used for a combined outcome of the total count of positive pathogen detections. Generalized estimating equations were used in both models to account for correlation between stool samples from the same child, and models were adjusted for age, socioeconomic status, season (wet/dry), and randomization group for the water treatment intervention.

Results
Study participants were geographically distributed within the Dzimauli community in Limpopo, South Africa. 30.7% (n=124) of study participants were from Village A, 18.8% (n=76) from Village B, 18.3% (n=74) from Village D, 15.8% (n=64) from Group 2, 9.9% (n=40) from Group 1, and 6.4% (n=26) from Village C ( Figure 1). Surface water from tap/ pipe was the most common source of drinking water in 4 out of the 6 village groups. In the remaining 2 village groupings, the most common source of drinking water was municipal. Village A had the largest number of participants (n=124) with 89% (n=110) relying on municipal water and most of the rest (8%, n=10) obtaining surface water from tap/pipe as their primary source of drinking water. Approximately a third (n=21, 32.8%) of participants in Group 2 obtained their drinking water directly from surface water ( Figure 1).
The study population consisted of 404 children under the age of 3, with the highest percentage of children (38.4%) being under the age of 1 year old. 48.3% of the study children were male and 51.7% were female. Across both study years, 41.6% (n=168) households used municipal water as their primary source of drinking water, 36.1% (n=146) used surface water from tap/pipe, 6.7% (n=27) used direct collection from surface water, 11.9% (n=48) used groundwater, and 3.7% (n=15) used another or unknown source ( Table  2). No form of drinking water treatment was used in most households across drinking water source groups (n=342, 84.7%). While water treatment was rare prior to consumption, boiling water was the most used form of treatment (n=33, 8.2%). In terms of drinking water storage, the majority of participants used covered storage vessels (n=328, 81.2%), with jerrycans (n=200, 49.5%) being the most common, followed by plastic buckets (n=162, 40.1%), regardless of their water source. Water supply was not continuously available (was interrupted) for 14.8% (n=4) of those who relied on direct collection from surface water, but interruptions were much more common for those who relied on municipal (n=165, 98.2%), surface water from tap/pipe (n=110, 75.3%), and groundwater (n=21, 43.8%; Table 2).
In terms of childhood growth, the highest prevalence of stunting at baseline was found among children who used other/unknown water sources (n=7, 53.9%), while the highest prevalence of underweight was found among children who used direct collection from surface water (n=4, 14.8%). Lastly, wasting was most common among children who used groundwater (n=4, 8.5%). Across all groups, the average age of the child's mother was 28 years old, with the most common school grade completed by the mother being secondary school. The average socioeconomic status score was the lowest for those who used directly from surface water (mean WAMI score: 0.63) compared to the other groups (mean WAMI score: 0.80).

Seasonality of Water Sources
Changes in primary drinking water sources across wet and dry seasons over the 2 years of the study were rare. A majority of the households (n=275, 78.3%) used the same water source year-round (Table 3). Of those that changed water sources, 2.6% (n=9) switched to a higher risk source. The majority of those who changed to a higher risk source (n=6, 66.7%) went from using groundwater during the wet season to surface water during the dry season. Contrastingly, 4.6% (n=16) of households switched to a lower risk source, with the majority switching to municipal water during the dry season (Table 3).

Enteric infections
Overall, EAEC, EHEC/EPEC, and Giardia infections were the most prevalent across the different water source groups ( Figure 2). While the prevalence ratios for some enteric infections suggested small differences in infection prevalence across drinking water source types, the estimates were imprecise given the small numbers of infections for some water sources (Table 4). There were no significant differences in enteric infections between drinking water sources. Additionally, there were no differences in the total number of pathogens detected among children drinking surface water from tap/pipe (adjusted prevalence ratio (aPR): 0.99, 95% CI: 0.89, 1.10), directly from surface (aPR: 0.92, 95% CI: 0.70, 1.20), groundwater (aPR: 0.98, 95% CI: 0.87, 1.11), and other/unknown source (aPR: 0.95, 95% CI: 0.71, 1.27) compared to municipal water. The prevalence of Giardia and ETEC infections among children who drank groundwater were 1.21 (95% CI: 0.94, 1.55) and 1.19 (95% CI: 0.84, 1.69) times the prevalence of those infections among those who drank municipal water over the 2-year follow-up period (Table 4). EHEC/EPEC infections among children who relied directly from surface water was 1.19 (95% CI: 0.87, 1.62) times as likely as those infections among those who relied on municipal water.

Discussion
We investigated seasonal shifts in primary drinking water source type and the effect of water source type on pathogen prevalence in children in rural communities in Limpopo Province, South Africa. Overall, changes in primary drinking water sources across seasons were small. Higher risk sources were more common during the wet season. Both of these findings were similarly found in a study by Pearson et al. (2016) in Uganda and Tanzania. Our study area and those in Pearson et al. (2016) in rural communities exhibit a unimodal dry/wet seasonality with an extended dry season in a semi-arid climate. Pearson et al. (2016) found that more households used lower risk sources during the dry season, compared to the wet season, as we did in South Africa. Lower risk source usage in the dry season would suggest lowest risk for enteric infections in the dry season according to Kostyla et al. (2015). This is additionally supported from earlier microbiological analyses of water samples from our study area that found that bacterial levels of surface water sources were higher in the wet season than in the dry season (Edokpayi et al., 2018). The increase in bacteria during the wet season could be caused by increased runoff carrying bacteria from contaminated sources to these bodies of water.
There were no significant differences in enteric pathogen prevalence between drinking water sources. Earlier microbiological analysis from this study (Edokpayi et al, 2018) found that while municipal water never tested positive for fecal contamination from E. coli from 2016-2017, intermittent water distribution and prolonged storage made municipal water an inconsistent water source with high risk of recontamination during storage.. There was a high prevalence of total coliform bacteria in household-stored water in those who used municipal water, even though the municipal water had no fecal contaminants when sampled directly from the tap (Edokpayi et al., 2018). Intermittent availability could therefore result in similar levels of contamination between "high risk" and "low risk" water sources and could explain why households might switch from lower risk municipal water to a higher risk source regardless of season. Prior research suggests that water contamination often occurs during storage through the use of water transfer devices, using a hand or bowl to transfer from the storage container for consumption (Heitzinger et  . These factors may have a larger effect on enteric infection than water source type, since a majority of households experienced interrupted water supply and practiced no form of drinking water treatment regardless of water source type. Water quantity and consistency of availability for consumption and hygiene may ultimately be more important for preventing transmission of certain enteric pathogens than contamination of the water source. The limited effects of water source on enteric pathogen prevalence in children could also be due to the fact that children were exposed through other pathways in addition to drinking water. All of the pathogens observed are transmitted via the fecal-oral route, where direct person-to-person contact and indirect contact through contaminated food are examples of other potential transmission pathways through which children could be exposed (Anin- ). Additionally, this study was limited by only analyzing primary drinking water source data. We did not consider that households may use more than one drinking water source or use separate sources for different household purposes, like cooking and bathing. Children could be exposed to enteric pathogens through water or contact with infected persons from outside sources, for example neighbors, non-household family members, or daycare centers, or through eating contaminated foods.
Access to safe and clean drinking water is a critical determinant of health, with social, economic, and environmental consequences. Especially among young children who are at the greatest risk for enteric infections, contaminated drinking water can contribute to environmental enteropathy, a syndrome marked by increased intestinal permeability, impaired immune function in the gut, and malabsorption, which may lead to child growth stunting (Korpe and Petri, 2012). Because households' primary drinking water source can vary across seasons, water sources and the potential for seasonal differences are key components that need to be evaluated to ensure equitable access to safe drinking water yearround. Future research could involve qualitatively investigating the decision-making process behind drinking water source choice and reasons for changes in water sources. Understanding these dynamics as well as how water is used and stored are important to reduce enteric pathogen exposure from high risk sources and could be useful in community planning, distribution, and management of clean water.
Increasing access to, and the consistency of, the municipal water supply are clear opportunities to improve water quality in Limpopo. Access to municipal water varied by village in the same community, but was not related to distance from the municipal water treatment plant. This may reflect varied ability of the traditional leadership across villages to negotiate with the municipal leadership to improve water access based on the requests of community members. Previous research in Dzimauli has documented strong citizen engagement through traditional authority structures to advocate for improved water services (Bulled, 2017). However, these engagement efforts may be more successful among individuals with higher socioeconomic or political status, ultimately resulting in inequitable access. The South African constitution documents that water is a human right (Bill of Rights, 1996); continued advocacy to ensure continuous supply equitably throughout communities is a key step towards achieving this right for all. Increasing access to and consistency of municipal water is one of the many opportunities to improve health outcomes in this community. effort was supported by the University of Virginia (UVA) Jefferson Public Citizens Program, the UVA Center for Global Health, and the National Institutes of Health, National Institute on Minority Health and Health Disparities (grant T37MD008659 to RAD). The funders had no involvement in the collection, analysis, or interpretation of data; in writing the report; or in the decision to submit the article for publication. The opinions, findings, conclusions or recommendations are those of the authors alone and do not necessarily reflect the views of NAS, USAID, the National Research Foundation of South Africa, the Thrasher Research Fund, NSF or the NIH.

•
We report water source seasonality and effects on enteric infections in children

•
Changes in primary drinking water sources between wet and dry seasons were rare • There were no differences in enteric infection prevalence by drinking water source      Int J Hyg Environ Health. Author manuscript; available in PMC 2022 January 01.