Water promotion for overweight prevention in schoolchildren: For whom does it work best? – Results of a randomized controlled trial

Abstract

Introduction

Promoting water consumption as an alternative to sugar-sweetened beverages (SSB) helped prevent overweight among students in a trial of the Water First intervention. We investigated the heterogeneity of effects on weight and beverage consumption by child socio-demographic, family, and health-related characteristics.

Methods

Elementary schools were randomized to serve as controls or receive the Water First intervention, which combined environmental and educational approaches to promote water consumption. In this secondary analysis, outcomes were Body Mass Index z-score, and frequency of water and SSB consumption at the 15-month follow-up. The moderating effects of baseline, sex, Hispanic ethnicity, number of caregivers, home connectedness, and acculturation were examined with mixed-effect linear models by testing an interaction with intervention status.

Results

Overall, 1250 4^th-grade students were included. For each additional unit of baseline SSB consumption, the 15-month frequency of SSB consumption was 5% (proportional increase [PI]: 0.95; 95% confidence interval [CI]: 0.91,0.98) lower in the intervention than the control group. The effect of Water First in increasing water consumption was estimated to be lower in students speaking English and Spanish at home (PI: 0.80; 95%CI: 0.61,1.04), compared to those with a lower acculturation status, speaking only Spanish. There were no other important moderators.

Conclusions

Water First had largely consistent effects across socio-demographic, family, and health-related characteristics. However, effects were more pronounced for students with higher baseline SSB consumption and lower acculturation. Water First is therefore a valuable program for promoting healthier beverage consumption and preventing overweight in children from diverse backgrounds.

Introduction

In the United States (U.S.), one in three children meet criteria for overweight and almost one in five for obesity¹, with children from disadvantaged backgrounds and ethnic minority groups being disproportionally affected². This is a major public health concern because childhood obesity is known to track into adulthood and to be a risk factor for a range of metabolic conditions such as dyslipidemia, type 2 diabetes and hypertension^3–5. The consumption of sugar-sweetened beverages (SSB) is an important contributor to overweight and obesity^6,7. Added sugar intake from beverages may induce lower satiety compared to solid food, resulting in only partial compensation during meals and higher total caloric intake⁸. However, despite decreasing consumption patterns over time, SSB intake remains elevated in the United States, accounting for 10% of daily energy intake of children and adolescents on average⁹.

Interventions promoting water consumption as an alternative to SSB have been shown effective for preventing overweight in school children^10–12. However, health promotion may not be equally beneficial in different subgroups because of various needs and circumstances, for example related to health status, food environment and social support^13,14. Few studies have examined how the effect of beverage interventions might vary in preventing overweight across groups of children from different backgrounds or with different health characteristics or behaviors^11,15–18. Ebbeling et al.¹⁵ and Sichieri et al.¹⁷ showed that children with baseline excess weight tended to benefit more from such interventions than children with healthy weight, while Muckelbauer et al.¹¹ did not find a different impact by weight status. Another study found that replacing SSB with noncaloric beverages in adolescents with overweight or obesity was effective in reducing body mass index (BMI) among Hispanic participants, but not among non-Hispanic ones¹⁶. Beyond ethnic background, the degree of acculturation - that is, the extent to which individuals retain their cultural practices versus adopt those of the host country - might also play a role. A school-based nutrition and physical activity intervention found greater improvements in weight status at two-year follow-up among more acculturated children¹⁹. Differences in beverage intervention effects according to gender were mixed^11,17,18. Systematic reviews of broader lifestyle interventions focusing on diet and/or physical activity found that Hispanic children and those with lower parental involvement might benefit less from such health promotion²⁰, while girls tended to respond better than boys, as did children with unhealthier baseline of the behaviors targeted by the interventions¹³.

Examining heterogeneity in the effects of health interventions is key to addressing health equity, and can guide adaptations of interventions to improve effectiveness prior to broader implementation in the community^13,21. This secondary analysis of the Water First trial investigated how children’s socio-demographic, family and health characteristics and behaviors modify the effect of a school-based intervention on participants’ weight status and beverage intake. Water First is a program that promotes water consumption by enhancing access to safe, appealing drinking water and promoting water intake in elementary schools. The program was developed using participatory research methods and was previously tested in public schools serving predominantly Hispanic students. It was effective for increasing the frequency of water intake and preventing overweight after 15 months, but did not significantly change SSB intake, prevalence of obesity, or BMI z-score¹⁰. Significant changes in obesity rates may require more intensive clinical interventions²². Water First had differential effect on weight status that could help explain the lack of a linear effect on BMI z-score. Following the current literature, it was hypothesized that girls, Hispanic-identifying students, those with a supportive family environment, indicated by the number of caregivers and higher family connectedness, those with higher baseline BMI z-score or SSB consumption, as well as those with lower baseline water consumption would benefit more from the intervention than their counterparts.

Methods

Study Sample

Four cohorts of 6 to 8 elementary public schools serving low-income students (≥50% free/reduced-price meal-eligible) in northern California were annually included between 2016 and 2019 for a total of 26 schools. In each cohort, schools were randomized with a 1:1 allocation ratio to either receive the intervention or serve as controls (i.e. continue their usual school practices). All fourth-grade students speaking English or Spanish and who did not have a health condition precluding intake of water were eligible for participation in the study. Fourth-grade students were targeted because they tend to consume more SSBs than younger children²³, were at a developmental stage to participate in educational lessons and take home activities, and provide more reliable survey data than younger children, while being easier to track over the 15-month study period than older students who may move to middle or junior high schools²⁴. The study was approved by the institutional review boards at the University of California, San Francisco, and Stanford University and registered at clinical-trials.gov (NCT03181971). Written informed consent was obtained from parents of all participating students. Assent was obtained from all student participants.

Assessment took place in schools at baseline, and at 7- and 15-month follow-up. Follow-up of the 2019 cohort could not be conducted because of COVID-19 pandemic school closures. Therefore, the current secondary analysis only focuses on the 18 schools with complete assessments recruited in the first three years of the study.

Intervention

The Water First intervention was developed using participatory research methods, with the aim to promote water consumption among children in schools. A community advisory board including representatives from the California Department of Education, schools, and families reviewed the intervention material for relevance. Hispanic-identifying students constituted the largest ethnic group in the targeted schools²⁴ and Water First was culturally tailored to this population by translating the material into Spanish, including examples of culturally relevant food and beverages, and having Spanish-speaking staff as point of contact. Listening sessions were conducted with families from Hispanic background to ensure cultural appropriateness of the intervention. The program spanned over a school year and aimed to increase water accessibility by providing safe and appealing water sources in strategic school locations, including the cafeteria where cups were provided, and by distributing personalized reusable bottles to students and teachers in 4^th grade classes. Additionally, it promoted water consumption through educational lessons and activities that highlighted the health and environmental benefits of drinking tap water. Families were involved through family engagement activities, such as assessing the sugar intake from drinks available at home or setting common goals for healthier beverage consumption.

Measures

All variables were student-reported, except for height and weight that were measured by trained research staff.

Outcomes were drawn from the 15-month follow-up assessment. The primary outcome was BMI z-score calculated from the height and weight of students measured with standardized protocols²⁵ and based on the U.S. Centers for Disease Control and Prevention growth curves²⁶. Although Water First did not significantly change BMI z-score¹⁰, it was selected as the primary weight outcome instead of overweight status because its continuous nature helps maximize the statistical power of the analysis (especially important with analyses of interactions) and because the absence of a main effect does not preclude the existence of interactions. Following the trial’s protocol²⁴, secondary outcomes were the frequency of water and SSB consumption over the past week assessed from a validated questionnaire²⁷. SSB included any beverage with added sugar, such as non-diet soda, fruit drink, sports drink, energy drink, sweetened coffee and tea, flavored milk, and flavored water.

Based on prior findings from beverage interventions^11,15–18, primary moderators were sex, self-identified Hispanic ethnicity, and the baseline level of each outcome (BMI z-score, frequency of water and SSB consumption), measured as previously described. Exploratory moderators, for which existing evidence is lacking in beverage interventions, included the average number of caregivers across the household(s) children lived in, home connectedness, and acculturation-level level. Home connectedness was measured as a score based on two items of the home connectedness subscale from the Healthy Kids Resiliency Assessment²⁸. Answers to the questions “In my home, there is a parent or some other adult who cares about my schoolwork” and “In my home, there is a parent or some other adult who wants me to do my best” were assessed on a 4-point Likert scale and coded from not at all true = 0 to very true = 3 and averaged across the two items²⁹. Language was used as a proxy for acculturation in Hispanic-identifying students, as deemed acceptable when a detailed assessment is unfeasible³⁰. Language(s) spoken at home was categorized into English only (high acculturation), both Spanish and English (medium acculturation), and Spanish only (low acculturation)³¹. This measure exclusively focused on Hispanic-identifying students to target the largest immigrant population in the study, while also distinguishing the effect of acculturation from comprehension of Water First intervention materials which were available in English and Spanish, but not in other languages.

Because of previous research documenting strong relationships with beverage intake and weight status, children’s age at baseline, self-reported sex, screen time and physical activity were selected, a priori, as control variables, all adapted from widely used instruments^34,35. Average daily screen time was calculated by summing up self-reported hours spent playing video games, watching videos and doing other screen activities on the previous day and on typical weekend day (i.e., (weekday × 5 + weekend day × 2)/7)³². Physical activity was measured as the number of times in the prior week that students reported being engaged in an activity that made them breathe hard or sweat³³.

Statistical Analysis

A statistical analysis plan was developed for this study and uploaded to the Open Science Framework (OSF) website under the Water First project (https://osf.io/f8hzd/). All analyses were performed with R (4.3.2) following the intention-to-treat principle. Mixed-effect linear regression models were performed with random intercepts at the school and class levels to control for the clustered design. BMI z-score and frequency of water and SSB consumption were included as dependent variables; beverage outcomes were log-transformed because of their skewed distribution. The effect of each moderator on each outcome was separately estimated by adding an interaction term of the intervention status and the moderator of interest. Because the intervention was found to be more effective in overweight children than in children with healthy weight and obesity¹⁰, a non-linear moderating effect of the baseline BMI z-score on the 15-months BMI z-score was tested with a quadratic term on the corresponding interaction. The quadratic term was dropped from the final model if its p-value was larger than 0.05. Models were adjusted for age, screen time, physical activity, and all moderators except acculturation. For the presentation of results, coefficients from the log-transformed models were back-transformed to the original outcome scale with exponentiation, allowing for interpretation as proportional increase (PI) in the outcome. Interactions significant at a 5% level were judged to be statistically significant and further investigated with analysis stratified by the corresponding moderator and/or estimated marginal means³⁶. Correction for multiple testing was not performed to preserve statistical power for interaction tests, which tend to have low power.

Multiple imputation by chained equations was performed using the R mice package³⁷, with 20 imputed datasets following the guidelines provided by White et al.³⁸. The imputation model included all study variables, the cohort year, as well as BMI z-score, and frequency of water and SSB consumption at 7-month follow-up. Interaction and quadratic terms were passively imputed to keep consistency between variables.

Results

A total of 1250 students from 18 schools and 56 classes were included in the study after exclusion of one student with missing height and weight at all timepoints. Mean baseline age of students was 9.6 years (standard deviation: 0.4) and 593 (47.4%) were female. As a result of randomization of schools, 677 (54.2%) students received the Water First promotion program (Table 1). Exploratory analysis including the level of acculturation were performed in Hispanic-identifying students only (n=873).

Table 1.

Characteristics and behaviors of school students participating in the Water First study by intervention status.

Study variable	Total (n=1250)	Intervention (n=677)	Control (n=573)
Baseline age (year, n=1250)	9.6 (9.3 to 9.9)	9.6 (9.3 to 9.9)	9.6 (9.3 to 9.8)
Sex (n=1250)
Male	657 (52.6)	359 (53.0)	298 (52.0)
Female	593 (47.4)	318 (47.0)	275 (48.0)
Ethnicity (n=1242)
Non-Hispanic	369 (29.5)	183 (27.0)	186 (32.5)
Hispanic	873 (69.8)	492 (72.7)	381 (66.5)
Baseline BMI z-score (n=1249)	1.0 (0.1 to 1.7)	1.0 (0.1 to 1.7)	1.0 (0.1 to 1.7)
Baseline water consumption (times/day, n=1249)	5.1 (2.6 to 8.0)	5.4 (2.8 to 8.1)	4.9 (2.5 to 8.0)
Baseline SSB consumption (times/day, n=1249)	2.0 (1.0 to 4.0)	2.0 (1.0 to 4.1)	2.0 (1.0 to 3.9)
Number of caregiversa (n=1191)	2.0 (2.0 to 3.0)	2.0 (2.0 to 3.0)	2.0 (1.5 to 3.0)
Home connectednessab (scale range 0–3, n=1194)	3.0 (2.5 to 3.0)	3.0 (2.5 to 3.0)	3.0 (2.5 to 3.0)
Acculturationac (n=824)
English only spoken at home	239 (19.1)	135 (19.9)	104 (18.2)
Spanish and English spoken at home	470 (37.6)	269 (39.7)	201 (35.1)
Spanish only spoken at home	115 (9.2)	54 (8.0)	61 (10.6)
Physical activity (times/week, n=1245)	3.5 (1.5 to 5.5)	3.5 (1.5 to 5.5)	3.5 (1.5 to 5.5)
Screen time (hours/day, n=1249)	2.7 (1.4 to 4.7)	2.7 (1.5 to 4.9)	2.6 (1.3 to 4.5)

BMI: body mass index; SSB: sugar-sweetened beverage.

Age, BMI z-score, water and SSB consumption, number of caregivers, home connectedness, physical activity, and screen time are median (quartile 1 to quartile 3). Sex, ethnicity, and acculturation are number (%).

^aCollected at 7-month follow-up

^bAdaptation of the home connectedness subscale from the Healthy Kids Resiliency Assessment; higher score indicates higher connectedness ²⁸

^cAcculturation measured in Hispanic-identifying students only.

Of the full sample, 204 (16.3%) students had incomplete information, mainly due to missing follow-up surveys (n=197, 15.8%), which was more frequent among older students and in the intervention group (Appendix Table 1). Analyses presented hereafter were performed on imputed data to limit potential biases arising from selective non-response. Depending on the outcome, the intraclass coefficient ranged from 0.01 to 0.04 for the school level and from 0.01 to 0.05 for the classroom level. Parameter estimates for the models for each outcome without interaction terms are presented in Appendix Table 2. Model diagnostics, including visual assessments of linearity, residual normality, and homoscedasticity were deemed acceptable (Appendix Figures 1-3).

The effect of Water First on students’ BMI z-score, the primary outcome, did not differ across the examined moderators, whether primary or exploratory (all p-values > 0.1; Table 2).

Table 2.

Moderating effects of students characteristics and behaviors on the effect of the Water First intervention (n=1250).

Moderator	BMI z-score	Frequency of water consumption	Frequency of SSB consumption
	β (95% CI)	PI (95% CI)	PI (95% CI)
Primary
Sex: Female (Ref. Male)	0.02 (−0.06, 0.09)	1.02 (0.84, 1.23)	0.98 (0.78, 1.22)
Ethnicity: Hispanic (Ref. Non-Hispanic)	−0.03 (−0.12, 0.06)	1.09 (0.87, 1.35)	1.19 (0.92, 1.54)
Baseline BMI z-score	0.01 (−0.02, 0.05)	1.05 (0.96, 1.14)	1.03 (0.94, 1.13)
Baseline frequency of water consumption (times/day)	0.00 (−0.01, 0.01)	1.01 (0.99, 1.03)	1.00 (0.98, 1.03)
Baseline frequency of SSB consumption (times/day)	0.00 (−0.01, 0.01)	1.02 (0.98, 1.05)	0.95 (0.91, 0.98) **
Exploratory
Number of caregivers	−0.01 (−0.05, 0.03)	0.98 (0.89, 1.07)	1.06 (0.95, 1.19)
Home connectedness scorea	0.02 (−0.05, 0.08)	1.01 (0.86, 1.18)	1.10 (0.90, 1.34)
Acculturation: Spanish and English speaking (Ref. Spanish only)b	−0.03 (−0.13, 0.07)	0.80 (0.61, 1.04)°	0.82 (0.62, 1.10)
Acculturation: English speaking only (Ref. Spanish only)b	0.02 (−0.12, 0.16)	0.92 (0.63, 1.35)	0.91 (0.60, 1.39)

Boldface indicates statistical significance (p<0.05).

^**p-value < 0.01,

^°p-value < 0.1.

BMI: body mass index; CI: confidence interval; PI: proportional increase; SSB: sugar-sweetened beverage.

Results are estimates of the interaction term between each moderator and the intervention status (β), 95% confidence intervals and p-values from mixed-effect linear regression models adjusted for age, physical activity, screen time, and all the above moderators except acculturation. Beverage outcomes are log-transformed and coefficients back-transformed to the original outcome scale, allowing for interpretation as proportional increase. Missing data imputed with multiple imputation by chained equation.

^aHome connectedness score adapted from the home connectedness subscale from the Healthy Kids Resiliency Assessment, higher score indicates higher connectedness²⁸

^bAcculturation measured in Hispanic-identifying students only (n=873).

When looking at secondary outcomes, the effect of Water First on follow-up frequency of SSB consumption varied by baseline SSB consumption (PI: 0.95; 95% confidence interval [CI]: 0.91, 0.98; Table 2); there was a 5% decreased frequency of SSB consumption at follow-up in the intervention versus control group for each additional unit of baseline SSB consumption (Table 2 and Figure 1). In absolute terms, students who consumed SSBs 5 times per day at baseline had a 0.28 lower daily frequency of SSB consumption at follow-up in the intervention group compared to the control group, whereas the difference was negligible for those with a baseline frequency of 2 times per day.

Figure 1.

Predicted frequency of sugar-sweetened beverage (SSB) consumption at 15-month follow-up according to intervention status and baseline SSB consumption (n=1250). Estimated marginal means and 95% confidence intervals from mixed-effect linear regression models. P-value for the interaction between baseline SSB consumption and intervention status: 0.005.

In exploratory analyses, results indicated that the acculturation status may moderate the effect of Water First on follow-up water consumption in the Hispanic-identifying subsample (PI_{Spanish and English speaking}: 0.80; 95% CI: 0.61, 1.04, p-value: 0.091; Table 2). Although the interaction term did not meet the traditional 5% significance threshold, a moderating effect could not be ruled out in view of the meaningful effect size, the small p-value and the likely underpowered subsample of Hispanic participants (n=873). Therefore, the moderating effect of acculturation status was further investigated. The intervention appeared to be more effective in increasing the frequency of water consumption in students speaking Spanish only at home compared to those speaking both English and Spanish (Figure 2). When stratifying the analysis by acculturation status, Hispanic-identifying students speaking Spanish only at home reported a 50%% (PI: 1.50; 95% CI: 1.20, 1.86) higher frequency of water consumption at 15-month in the intervention compared with the control group, corresponding to an absolute increase of 1.55 times per day. This difference was smaller in children speaking English and Spanish (PI: 1.18; 95% CI: 1.01, 1.37, absolute increase: 0.78 times/day) or English only at home (PI: 1.32; 95% CI: 0.95, 1.84, absolute increase: 1.15 times/day).

Figure 2.

Predicted frequency of water consumption at 15-month follow-up according to intervention status and acculturation level in Hispanic-identifying students (n=873). Estimated marginal means and 95% confidence intervals from mixed-effect linear regression models. P-value for the interaction between acculturation level and intervention status: 0.091.

The remaining interaction terms between the Water First intervention status and the examined moderators were non-significant and mostly centered around zero (Table 2, Appendix Figures 4-11). Estimates from complete case analysis were of similar magnitude (Appendix Table 3).

Discussion

The Water First intervention was previously shown to be effective in increasing the frequency of water intake and preventing overweight in elementary school children¹⁰. This secondary analysis further showed that the effect of Water First on students’ BMI z-score and frequency of water and SSB consumption was homogeneous across socio-demographic, family, and health characteristics, except for baseline SSB consumption. Exploratory findings additionally suggested differences according to the acculturation status in Hispanic-identifying students, although not significant.

The effect of Water First on BMI z-score, the primary outcome for the current analysis, was not moderated by any of the examined student characteristics. This extends previous findings, which showed that the Water First intervention did not significantly impact students’ BMI z-score¹⁰, by demonstrating that this result holds across various socio-demographic, family, and health characteristics. Furthermore, students’ response to the Water First intervention neither varied by sex nor by baseline BMI z-score or water consumption, which adds to the limited and mixed findings from the literature on water promotion^11,15,17,18. Interestingly, there was no support for a moderating effect of Water First by students’ reports of home connectedness or by the number of caregivers in the home. This could be because the Water First intervention has a primary focus on the school environment rather than the home. Although it was expected that students with higher family support would benefit more from the intervention²⁰, it is reassuring to observe that the program was effective regardless of the family circumstances.

The effectiveness of Water First was similar among both Hispanic and non-Hispanic students, which might be attributed to the careful participatory approach used to develop the program and to its tailoring to Hispanic culture²⁴. In their meta-analysis Ling et al. found that Hispanic populations tended to benefit less from lifestyle interventions and called for culturally sensitive interventions to address this issue ²⁰. The current findings are encouraging by indicating that Hispanic minority groups may experience equal improvements from health interventions that are culturally and linguistically tailored.

However, the reported exploratory findings suggest that intervention effects may have differed across Hispanic-identifying students who are not a homogeneous group: Water First increased the frequency of water consumption to a 20% lesser extent in students speaking English and Spanish at home, than in their less acculturated counterparts speaking only Spanish, although this finding did not reach statistical significance. In a U.S. study focusing on Hispanic populations, perceived tap water safety was shown to be lower in less acculturated individuals, while most of them stated that they would buy less bottled water if they knew tap water was safe³⁹. Accordingly, in the current study, students with stronger ties to their country of origin might have been more accustomed to drinking bottled beverages in lieu of tap water due to concerns about water safety, making the promotion and provision of safe tap water particularly effective in increasing their water consumption. In contrast, acculturated students may have already been aware of local tap water safety and had established drinking habits accordingly, making them less likely to be influenced by the intervention.

On one hand, these findings indicate a need to better tailor water promotion to acculturated students. Drawing on previous research effectively promoting water consumption^40,41, this could be done through the development of community-based programs ensuring access to safe tap water at home and leveraging norms around its consumption at the family and community level. On the other hand, the current study suggests that providing culturally tailored information in individuals’ native language, combined with environmental changes, may effectively influence beverage habits in unacculturated populations who may have lower tap water trust and water consumption⁴². Furthermore, dietary choices serve as a mean for children to express their belonging within a peer group⁴³. Hence, school-based interventions that promote positive beverage norms might additionally provide unacculturated students with a healthy opportunity to integrate into the local school culture. However, this finding should be interpreted cautiously and requires validation in future studies that are fully powered to test this hypothesis.

In line with the study hypothesis, Water First was more effective in reducing the frequency of SSB consumption in students with higher baseline intake. This echoes findings from a systematic review, which showed that the baseline level of energy balance-related behaviors, such as diet and physical activity, was a common moderator in interventions aimed at preventing overweight¹³. It suggests that the intervention was most successful in limiting SSB consumption among those who needed it the most. Beyond the potential for overweight prevention, this could lead to long term health benefits, as higher SSB intake is associated with an increased risk of type 2 diabetes and cardiovascular diseases over time⁴⁴. By contrast, it also indicates that reducing the SSB intake of moderate consumers might require more intensive interventions, such as limiting the overall accessibility of SSB or providing personalized counseling at the family level.

Limitations

Findings of this study should be interpreted considering its limitations. First, the trial sample size was not specifically powered for modelling interactions, especially in the Hispanic-identifying subsample, and was additionally reduced due to data collection restrictions during the COVID-19 pandemic. To limit the potential issue of statistical power, researchers opted for a p-value threshold of 0.05 without correction for multiple testing, which could increase the risk of false positive findings. They also prioritized the use of continuous outcomes and moderators in the models and kept the number of categories to a minimum for categorical moderators, such as ethnicity and acculturation. However, it would have been insightful to investigate the effect of Water First on other racial or ethnic groups, for which the intervention was not specifically tailored, such as Asian-identifying students who accounted for 13.9% of the sample¹⁰. Second, most of the study variables were self-reported and are thus subject to recall and social desirability bias⁴⁵. However, the observed trends for water consumption were consistent with direct observation of water station use at the school sites, thereby increasing confidence in this measure. Third, because of data availability, the language spoken at home was used as an imperfect proxy for acculturation. Future research should incorporate more comprehensive indicators where possible³⁰. Finally, the Water First trial was conducted in public elementary schools serving a relatively high proportion of low-income students in California; results might thus not be generalizable to other populations. The study also presents major strengths, including the experimental design, the objective measure of anthropometric data, as well as the investigation of various socio-demographic, family, and health-related moderators.

Conclusions

The effect of the Water First multilevel intervention on elementary school students’ BMI z-score and water and SSB consumption was mostly consistent across socio-demographic, family, and health-related characteristics. However, differences were observed according to baseline SSB consumption with the intervention having the greatest impact on students with the less healthy behaviors. Furthermore, the effect of Water First appeared to be more pronounced in less acculturated Hispanic-identifying students speaking Spanish only at home, compared to those speaking both English and Spanish, although not significantly. These findings highlight the value of community-based, culturally tailored approaches to effectively engage the target population. Findings also highlight the need to address tap water safety concerns and social norms regarding tap water consumption in acculturated populations, for example with multilevel interventions at the family and community levels ^40,41. Overall, the increase in water intake was mostly consistent across demographic and behavioral groups. This study strengthens confidence in Water First as a promising intervention to improve water consumption patterns and ultimately help prevent overweight for students.