Study design

This study was part of a larger short-term longitudinal study in which data collection occurred over two time points, that is, time point 1 and time point 2. The data from the current study were from time point 2 (i.e., the second round of data collection) when a selection of measures (not the full set) was included.

Study sites

Between 2020 (time point 1) and 2021 (time point 2), participants were recruited from four low-income communities (Sites A, B, C and D) within the Cape Town Metropolitan area. In low-income settlements in and around Cape Town, South Africa, many households struggle with water access and storage [5]. Two of the communities (Site A and B) were classified as an urban township and were made up of mostly informal housing. Overcrowding is an issue in these communities with population densities of about 16,000 per km² and 10,000 per km². Other challenges include high rates of unemployment, food insecurity, alcohol abuse, crime and human immunodeficiency virus / acquired immune deficiency syndrome. The other two communities (Site C and D) were part of what is referred to as the Cape Flats and were made up of both formal and informal housing. The population density varies within the suburbs from which participants were recruited (between 4,000 per km² and 16,000 per km²). Gang activity and drug abuse are major challenges in these community in addition to high rates of unemployment, crime, and food insecurity [15].

Participants

This study included 192 children aged 3–5 years and their primary caregiver from low-income settings in Cape Town. Specifically, children who were not attending early childhood care and education (ECCE) services were recruited as they represent a particularly vulnerable group that remains understudied, and constitute a significant proportion (30%) of the population of 3–5-year-old children [16]. Children not attending any form of ECCE in South Africa are likely at a higher risk for poor developmental outcomes, potentially have higher levels of exposure and are consequently influenced more by the experiences within the home and community environment due to increased time spent in these environments (and not in ECCE services) [17].

Recruitment of children and their primary adult caregiver (aged >18 years) was facilitated through our partner organisation who implement a home-based stimulation programme for pre-school age children and their families living in environments characterised by poverty, unemployment, crime, and violence. Due to COVID-19 restrictions in place at the time, initial recruitment took place telephonically. Contact numbers of interested caregivers were provided by the organisation’s home visitors and passed on to the study’s research assistants who provided the potential participants with information about the study, answered any questions, invited caregivers to participate, and obtained informed consent.

The initial recruited sample at time point 1 was 243 caregiver/child dyads. Due to the hard lockdown during the COVID-19 pandemic, some of the first caregiver interviews were conducted telephonically. Of those, 11% of the child data were lost as we were not able to contact the caregiver to conduct the child testing in person at time point 1. At time point 2, 22% of the caregiver/child dyads were lost to follow-up. The reasons for loss to follow-up included difficulties contacting the caregivers (due to changing phone number or losing their phone) or caregivers moving out of the community and therefore no longer with the community organisation. Participants lost to follow-up did not differ significantly in child age, caregiver age, caregiver education, household assets, exposure to violence, or the home learning environment compared with those participants who participated in the full study. Demographic data that were missing at the time point 2 were inferred from demographic data available at the first time point. Additional missing data were very low (<2%) and therefore no further data imputation was done.

Measures: Parent/caregiver questionnaire

The parent/caregiver questionnaire collected demographic details including the child’s age and birthweight, languages spoken in the home, number of adults and children living in the home, caregiver education, and child education. It also collected data about child exposure to stressful life events, caregiver adjustment and practices, home learning environment, household income, impacts of COVID-19, and water security and safety. The caregiver questionnaire was designed to be administered in approximately 30 minutes.

Measures: Water-related household questions

The World Health Organization gathers information on water access and storage in households and reports these at country level [18]. The same variables and similar responses (tailored for South African conditions) were used for the water-related questions in this questionnaire. The questions included: main source of drinking water (tap inside dwelling, outdoor tap on site, communal tap offsite, water tanker/truck, other); storage of water for drinking or cooking in a container (no; yes), if yes, information on the type of container, size of opening, lid or no lid, means of making and keeping stored water clean (boil, bleach, other), frequency water storage container is washed, how water is removed from storage container (utensil such as scoop, cup, jug, other) and if there is one utensil that everyone uses to remove water from the storage container. We also asked, “in the last 4 weeks, how frequently did you or anyone in your household worry you would not have enough water for all your household needs?” This was asked to gauge reliability of access to water.

Procedures

All testing was conducted by three research assistants who were trained on the procedures and the assessments and could speak the home language of the participants (i.e., IsiXhosa, English or Afrikaans). Due to COVID-19 restrictions and that some of the areas that participants were recruited from posed a safety risk to the research assistants, testing took place in various ways. For two of the areas, child testing and caregiver questionnaires took place in the dwelling of the participants at a time that was suitable for the participant. For the other two areas that posed a safety risk for the research assistants, a central venue was organised, and participants (caregiver and child) were given transport money to come to the testing venue. All tests were explained to caregivers and children so that they knew what was expected of them. Child participants completed the IDELA task as well as additional assessments that were included as part of the larger study. If there was more than one child in the household that qualified to participate in the study, a separate questionnaire was administered for each child. Caregivers received grocery vouchers for completing the questionnaire and again for bringing their child to testing. Children received stickers for their participation.

All data were captured on Redcap and transferred into Stata 15.1 [21], where quality control and data cleaning were performed prior to data analysis. The procedures for this study were approved in advance by the Human Research Ethics Committee (Medical) of the University of the Witwatersrand (Reference: M200104).

Data management and statistical analyses

Data analysis was conducted using Stata 15.1 [21], and a p-value < 0.05 was considered statistically significant for all tests. Univariate regression models were used to determine unadjusted associations between source of water and IDELA scores, and water storage and IDELA scores of children from the four study sites. Multivariable regression was conducted to estimate adjusted associations between source of water and IDELA scores, and water storage and IDELA scores. Models were adjusted for covariates including age and sex of the child, number of people living in the home, and household income range.

Sample descriptives

Among the 192 child participants, the mean age was four years and the majority were female (55%) (Table 1). Most households had a monthly income below ZAR 6,000 (USD 310 in June 2023). Households comprised on average six members. The mean IDELA score by area ranged between 36% to 54% with an overall mean of 48%—the higher the score, the better the child’s development. We did not present the finding by sub-scores due to the relatively small sample size.

Water-related household findings

Around 70% of households had a tap inside their dwelling providing them with water (Table 2). Around half said that they stored water, and the largest percentage of storage containers (21%) were plastic and totally open (with no lid). Most people who stored water did not treat the stored water and those who did either boiled it or added bleach. Around 54% of participants reported that they washed the water storage container either daily, or once or twice a week. Participants storing water mainly used a utensil to decant water from the storage container. Two-thirds of participants reported worrying (mostly rarely or sometimes) about not having enough water for all household needs in the past month.

Regression results

Unadjusted univariate regression results showed that the predicted IDELA scores for children from Site B, where participants reportedly got their water from outdoor taps on the stand, were 34 points lower than those who had access to taps inside the dwelling (Table 3). This association was found to be statistically significant (p = 0.042). However, the conference interval was wide. The relationships between the main source of water and IDELA scores for children from Site A, Site C and Site D were not statistically significant.

Univariate analysis also found that storage of water was statistically significant in Site D, where the IDELA scores were calculated as being higher for children who lived in homes that reported storing water for domestic purposes (co-efficient = 10.06, p = 0.036) compared to those who did not. Once again, the confidence interval was wide likely due to the relatively small sample size.

After adjusting for potential confounding factors, that is, age and sex of the child, number of people living in the home, and household income range for each of the sites, IDELA scores among children from Site B were lower for those from households who stored water compared with those that did not store water (co-efficient = –12.13, p = 0.042, 95% CI = [–23.74] – [–0.52]) (Table 4).

IDELA scores

The IDELA scores for the children in our study were lower compared to those of Brazilian children (participant ages were 3 and 6 years) living in Sao Paulo [23]. Our mean score was 48% (range 36–54%) while it was 78% (74–89%) for the Brazilian study participants. Possible reasons for these differences are socio-economic status and schooling conditions—children in the Brazilian study may have been attending school while our child participants were not attending school.

Relations between water access/supply and IDELA scores

Properly treated water provided to indoor taps is likely to be cleaner than water from an outdoor tap (potentially shared between several people) that must be collected and brought indoors [24, 25]. The IDELA scores for children from Site B were most affected by source of water, that is, indoor tap or outdoor tap. This was the one site with a statistically significant association where children from households where water was obtained from outdoor taps on the stand had lower IDELA scores compared with children who had access to an indoor tap. Access to an indoor tap is easier than having to walk outside, perhaps in inclement weather, to fetch water for consumption. Moreover, an indoor tap is likely more “visible” and a reminder to drink water when one is thirsty compared to when a tap is outdoors. It is also likely that having an indoor tap versus having an outdoor tap is correlated with socio-economic status and income level; this site is known to have a range in household income levels.

Water storage and child development

Water storage is common in South African households due to the risk of non-continuous water supply or the need to treat the water because the supplied water is deemed unclean [26]. Several households stored water, likely because of water service interruptions, to ensure continuous access to water. In Site D, 73% of homes used water from a tap in the home and 53% stored some of that water. When making use of an indoor tap, water may be obtained and consumed immediately or stored for later use. When using an outdoor tap, water must be collected in a container and transferred into the dwelling. The container may be unclean, and the water may become contaminated between the point of collection and the place of indoor storage. Moreover, people and animals may drink water from the outdoor tap transferring contaminants from mouth to tap. An indoor tap is less likely accessible to animals and water is more likely to be poured into a container/cup for consumption keeping the tap (relatively) clean compared to a tap from which people drink water directly.

Site B had a statistically significant association in the adjusted (for age and sex) regression where IDELA scores were calculated to be lower for children from homes where water was stored. Only half the participants reported washing their water containers – an essential practice for safe water storage. Several reflections arise in relation to these water storage and child development findings. Water storage may have been perceived differently by households, for example, storing drinking water in a refrigerator versus storing water in a 10-litre drum without a lid. In this site, several households made use of outdoor taps on their stand or a communal outdoor tap. Future studies should be specific about water storage possibilities. It is also possible that water storage is an indicator of socio-economic status and this status varied widely in Site B.

The City of Cape Town regularly imposes water restrictions to keep the water system running until heavy rains occur [27]. This suggests that our study sites are vulnerable to the impacts of climate change; water storage is an important part of climate change adaptation therefore communities need to understand how to safely store water. For example, most homes that stored water did not cover the container, and stored water was not treated prior to drinking. The World Health Organization’s Household Water Treatment and Safe Storage Manual [28] advises that to keep water safe, best practices include treating water (by bringing water to a rolling boil) prior to storage, cleaning the storage container regularly, covering the container with a lid, and not contaminating (for example by drinking from) the scoop or utensil used to decant water from the storage container. Such knowledge should be shared with all households who store water to ensure safe, adequate water in places with water restrictions or non-continuous water supply for the benefit of child development, as well as reduction of health outcomes, such as diarrhoeal disease [29]. It is important to note that access to electricity or alternative fuel sources for boiling water may be a challenge for people living in low-income communities and alternatives may be needed, for example, solar thermal water disinfection [30].

Study strengths and limitations

Few studies have considered IDELA scores in relation to environmental factors and we consider our study among the first to do so in relation to water. A Brazilian study linked child development to community noise exposure at the child’s home and found that community noise exposure may impair behavioural and cognitive development of preschool children [23].

Another strength of this study was the parent/caregiver and child dyads that allowed us to gather multiple variables around water access and storage in households in relation to child development. Despite the relatively small sample due to loss to follow-up, we consider our results to be robust. We did not test water quality in any of the participants’ homes and this should be considered for future studies. It would also be helpful to ask about household sanitation since water and sanitation are closely linked. Future studies may also want to consider additional development tests, such as the Early Learning Measures Outcome (ELOM) [31] to support interpretation of the IDELA scores. We used a cross-sectional study design, and one might consider an alternate design such as a cohort or prospective study to gain further insight into the relationship between water access/storage and child development. Finally, we did not conduct an analysis on IDELA sub-scores to consider differential effects on these different domains but future studies with larger samples should consider doing so.