Cost-Effectiveness of Installing Water Dispensers in Massachusetts’ Schools to Improve Nutrition Environments, Health, and Health Equity

Abstract

Background

Evidence suggests that providing children with water on school lunch lines promotes healthy weight and is cost-effective.

Objective

Estimate reach, cost, cost-effectiveness, and impact on weight status and obesity-related health equity of a strategy to install chilled, filtered water dispensers on school lunch lines in Massachusetts’ schools.

Design

A cost-effectiveness analysis was conducted in 2022–2023 using the Childhood Obesity Intervention Cost-Effectiveness Study (CHOICES) microsimulation model (2020–2029), estimates from published studies, publicly available national and state surveillance, programmatic, administrative data, and experts’ assumptions.

Participants/Setting

Schools participating in Massachusetts’ school drinking water lead testing program with elevated lead concentrations in areas with disproportionate health risks

Main Outcome Measures

Reach, cost, cost per quality-adjusted life year (QALY) gained, and cases of obesity prevented overall and by income, race, and ethnicity

Statistical Analyses

Means and 95% uncertainty intervals (UIs) for outcomes were estimated by running the model 1000 times, accounting for uncertainty from data sources and population projections.

Results

This strategy is projected to improve drinking water access for 265 000 (95% UI, 259 000–273 000) students in 304 schools, prevent 525 (95% UI, 192–880) cases of childhood obesity in 2029, cost $4.48/student/year (95% UI, $3.54–$5.43) to implement, and is likely cost-effective ($72700 (95% UI, $40 000–$234 000) per QALY). Larger projected reductions in obesity prevalence for Hispanic/Latino and Black students compared with white students were observed (7.2 [95% UI, 5.2–10.3], and 4.7 [95% UI, 2.9–7.4], respectively), and among students from families with lowest incomes (<130% Federal Poverty Level) compared with the highest incomes (>350% Federal Poverty Level) (2.6 [95% UI, 1.9–3.5]).

Conclusions

A strategy to provide chilled, filtered water dispensers on lunch lines in Massachusetts schools is projected to improve school nutrition environments, increase water consumption, prevent excess weight gain, improve health equity, and be cost-effective.

ALL STUDENTS DESERVE ACCESS TO SAFE AND appealing drinking water at school. Drinking water is essential for health and can positively affect physiologic functions and cognition.¹ However, many children may not consume water at a level recommended for health,^2,3 and inadequate hydration status differs by race, ethnicity, and gender.³ Students drink more water when schools provide access to water at lunch at no charge,^4–7 and they consume fewer sugary drinks.^5,8 However, some students may not drink water while at school because of mistrust of water as safe or water sources as clean and safe^9,10 or because drinking water sources are not readily accessible.¹¹ Evidence indicates that improving school water access can help prevent excess weight gain¹² and obesity in children.^13,14 In 2021, most adolescents in Massachusetts reported consuming sugary drinks,¹⁵ which contribute to excess weight gain and other chronic diseases.^16,17 Studies suggest that some schools in Massachusetts also do not provide adequate access to drinking water, including in areas where meals are served.^18,19 Thus, policies and strategies to ensure appealing access to quality drinking water are needed to ensure that this option, which is healthier than sugary drinks, is available to meet hydration needs.²⁰

Many schools in Massachusetts have identified water quality issues in their drinking water²¹ and the need to improve drinking water infrastructure to improve quality access to water.^18,19 For example, lead exists in some schools’ tap water because of lead sources in plumbing, fittings, and fixtures in school buildings with older infrastructure.²¹ However, there are effective ways to mitigate this preventable harmful exposure to lead in drinking water and improve the physical environments of school buildings via drinking water testing, remediation actions to reduce lead by replacing fixtures and fittings,²² and strategies to increase drinking water access.²³ Since 2016, the Massachusetts Department of Environmental Protection has operated the voluntary statewide Assistance Program for Lead in School Drinking Water to ensure quality drinking water by testing for lead in school drinking water.²¹ Through this program, water samples are taken from building fixtures used for drinking or food preparation and tested for lead, and, where detected, program staff connect schools with resources to facilitate remediation.²¹ Consultation also may include referral to other state-sponsored program opportunities for funding to replace lead-containing fixtures with chilled, filtered water dispensers.²¹ However, many eligible schools have not participated in the water quality testing program in Massachusetts.²⁴

Recent evidence outlines the health benefits and cost savings for policies that can reduce lead in drinking water.²⁵ Although not implicated in excess weight gain, exposure to lead is indicated to have negative impacts on cognition, attention, and hyperactivity in children and adolescents.²⁶ Meanwhile, other evidence suggests that, after installing water dispensers on school lunch lines, the resulting healthier weight status and related national obesity-related health care cost savings could support half the upfront costs of water dispenser installation.²⁷ However, these studies did not assess the potential impacts of the installation of water dispensers on health equity outcomes. Therefore, the objectives of this study were to estimate the reach, cost, cost-effectiveness, and impact on weight status and obesity-related health equity of a strategy to install chilled, filtered water dispensers on school lunch lines in Massachusetts.

METHODS

This study used economic evaluation and cost-effectiveness analysis to estimate the impacts of a strategy to install chilled, filtered water dispensers on or near school cafeteria lunch lines in Massachusetts K–8 public schools. Data were collected via a review of estimates from published studies, publicly available national and state surveillance, programmatic and administrative data, and expert assumptions. Estimates of the intervention strategy’s reach, costs, and effects (Fig 1) were inputs to the Childhood Obesity Intervention Cost-Effectiveness Study (CHOICES) microsimulation model.

Figure 1.

Parameters used in cost-effectiveness analysis to estimate the reach, effect, and cost of implementing a strategy to provide safe chilled water on lunch lines in Massachusetts (United States) schools, 2020–2029.^{5,6,14,27–37}

The CHOICES model is a stochastic, discrete-time, individual-based microsimulation model developed in Java³⁸ and accessed via custom-built standalone software to project obesity-related outcomes for an open virtual population under scenarios with and without the intervention.^39–41 This cost-effectiveness analysis used a societal perspective, and a 10-year time horizon (2020–2029) was used because of its relevancy for policymaking. Analyses were conducted in 2022–2023. This study followed the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) reporting guidelines.⁴² The Harvard TH Chan School of Public Health Institutional Review Board determined this study to be Not Human Subjects Research.

Overview of Project

The CHOICES Project at the Harvard TH Chan School of Public Health has used cost-effectiveness analysis to compare the costs and outcomes of policies and programs promoting improved nutrition or increased physical activity in schools, early care and education, out-of-school time settings, communities, and clinics.^{27,39,43–50} The CHOICES process involves engaging community partners, public health practitioners, and researchers; reviewing evidence to identify effective strategies for promoting healthy weight and their estimated population reach and cost; and simulation modeling to project cost-effectiveness outcomes.³⁹ CHOICES has worked with national, state, and local partners to assess the cost-effectiveness of several policies and programs for the US population^{27,39,43–47} and state and local populations.^49–51 In Massachusetts, the CHOICES Project, the Massachusetts Department of Public Health (MDPH), and the Massachusetts Department of Elementary and Secondary Education (DESE) collaborated through the Massachusetts-CHOICES Project, a training, technical assistance, and modeling initiative that aimed to develop a playbook of strategies to promote healthy weight and advance health equity and study how cost-effectiveness metrics are used by decision-makers.⁵²

Setting

Massachusetts is a US state with 7 million residents, 19% of whom are children younger than 18 years of age.⁵³ Two-thirds (67.6%) of Massachusetts residents are white, not Hispanic or Latino; 6.5% are Black or African American, not Hispanic or Latino; 12.6% are Hispanic or Latino; 7.2% are Asian, not Hispanic or Latino; and 6.1% identify as another race or multiracial.⁵³ Over 900 000 students attend public schools in Massachusetts. Of these, 51.5% are white, 10.2% are Black or African American, 25.9% are Hispanic or Latino, 7.5% are Asian, and 4.9% identify as another race or as multiracial. Additionally, 42% of students are from households with low income.

About the Water Dispensers in Schools Strategy

The modeled strategy includes the installation of chilled, filtered water dispensers (ie, bottle filling stations) on or near school cafeteria lunch lines in K–8 schools in Massachusetts, providing potable drinking water access where meals are served in schools in alignment with federal nutrition standards.⁵⁴ This study assumed that implementation would be supported via Massachusetts’ existing voluntary lead testing program,²¹ that partners at MDPH and DESE would provide outreach, technical assistance, and support school districts’ participation in the existing School Water Infrastructure Grant Program (SWIG), or in identifying other funding opportunities to support the installation of chilled, filtered water dispensers in schools. It was assumed that school districts would be prioritized for outreach and implementation based on existing criteria used by Massachusetts agencies to identify geographic areas in the state with disproportionate health risks based on social determinants of health.^33,34 Water dispensers would be installed at schools with viable plumbing to fill cups or bottles with filtered drinking water.

Effect

Estimates used to model the effect of the strategy on body mass index (BMI) and student water consumption are shown in Figure 1. The effect of the strategy on BMI was estimated using evidence from a natural experiment among more than 1 million K–8 students in New York City.¹⁴ Using a difference-in-difference analysis, the study found that installation of chilled water dispensers on school cafeteria lunch lines was associated with a BMI z-score difference of −0.022 (95% CI, −0.035, −0.008) among girls and −0.025 (95% CI, −0.038, −0.011) among boys, compared with BMI z-score changes among students at schools where water dispensers were not yet installed. On request, the study authors shared corresponding results using changes in BMI as outcomes from this natural experimental study (Figure 1) for use in a cost-effectiveness analysis (e-mail communication, October 2016). Before installation, mean student BMI z-scores in schools that installed water dispensers were not systematically different from schools that did not install water dispensers during the study period.¹⁴ Changes in students’ water consumption after the installation of water dispensers were also estimated, assuming that 22% of students at schools where water dispensers are installed drink water from the dispensers on any given day,⁶ and on average, consume more ounces of water per day (mean ± standard deviation; 1.42 ± 0.44 ounces).²⁷

Population Reach

Estimates, descriptions, and sources used to estimate population reach are shown in Figure 1. Massachusetts public schools serving grades K–8 were identified, and attendance in grades K–8 by race and ethnicity was estimated based on data from the Massachusetts Department of Elementary and Secondary Education.²⁸ Schools included in the eligible population were those eligible for fixture replacement because of findings of elevated lead levels through participation in the state drinking water testing program. These schools were identified using publicly available data from April 2016 through February 2022 from the Massachusetts Executive Office of Energy and Environmental Affairs.³² Districts and schools already awarded funding in the SWIG program were identified using publicly available data from the Massachusetts Clean Water Trust^29–31 and excluded from the eligible population. It was assumed that schools with elevated lead levels would be prioritized for strategy implementation based on existing guidance used by Massachusetts agencies to focus efforts in areas with disproportionate health risks based on social determinants of health.^33,34 Student enrollment in eligible schools by race and ethnicity²⁸ was used to estimate the population reached.³⁴ An average benefit for all children attending schools where water dispensers would be installed was assumed.

Implementation Costs

A societal perspective⁵⁵ was used to estimate the costs of implementing the strategy over 10 years. The incremental costs to implement the approach were estimated based on estimates from a prior economic evaluation²⁷ with additional guidance from members of relevant Massachusetts state governmental departments (ie, MDPH and DESE). This included the incremental costs of administering and managing the program at the state and school district levels; training school food service directors; purchasing, installing, cleaning, and maintaining the dispensers; increased tap water; electricity; disposable cup usage; and ongoing lead testing and remediation. Estimates of these costs and sources are shown in Figure 1. Costs are reported in 2019 US dollars (USD), adjusted for inflation using the US Bureau of Labor Statistics Consumer Price Index.⁵⁶ Cost outcomes are discounted at 3% per year⁵⁵ unless otherwise noted.

Modeling Framework

The CHOICES microsimulation model^39–41 was used to project outcomes if the intervention strategy were in place vs not implemented over a 10-year time horizon (2020–2029). The CHOICES microsimulation model simulated an open cohort of the Massachusetts population to estimate the impact on population health outcomes related to excess weight gain and costs, accounting for births and deaths expected to occur over the timeframe studied.³⁹ Fifty virtual populations of 1 million people each were simulated, accounting for sampling uncertainty based on national, state, and local data and individual-level characteristics such as dietary intake, body growth, and behaviors such as smoking.³⁹ Parameters, assumptions, and sources used in the microsimulation model are shown in Figure 2.

Figure 2.

Microsimulation model parameters used in cost-effectiveness analysis of implementing a strategy to provide safe chilled water on lunch lines in Massachusetts (United States) schools, 2020–2029, CHOICES v4.6.1a.^{39–41,57–84}

Researchers calibrated the model to empirical data.⁴⁰ Weight trajectories were calibrated to match objectively measured data from BMI screenings in Massachusetts public school districts (2009–2016)⁸⁵ by including statewide weight status prevalence estimates as calibration targets in the simulation.⁸⁶ Additional details on the CHOICES model are provided elsewhere.^39,83

Outcomes

Projected outcomes included cases of obesity prevented in 2029, along with population reach, quality-adjusted life years (QALYs) gained, health care costs saved because of reductions in excess weight, and net cost per QALY gained over 10 years. QALYs were estimated using health-related quality of life weights associated with 3 BMI categories (under/healthy weight, overweight, obesity) by sex and age group.^83,84,87 Health care cost savings were projected based on published estimates of medical expenditures by BMI unit and age.^41,83,84,87 Incremental cost-effectiveness ratios of costs per QALY gained were estimated as the net cost difference (implementation costs minus health care cost savings) per health benefit gained in terms of QALYs. Thresholds ranging from $50 000 to $200 000 per QALY gained are commonly used to classify acceptable cost per health benefit (ie, cost-effectiveness)⁸⁸ and were considered for this study. Health care cost savings were projected based on published estimates of medical expenditures by BMI unit and age.⁴¹ Costs and health outcomes are discounted at 3% per year⁵⁵ unless otherwise noted.

To estimate the strategy’s impact on health equity by race, ethnicity, and income, cases of obesity prevented in 2029 (total and per 100 000) were compared in 4 race and ethnicity groups (Black or African American, not Hispanic/Latino; Hispanic/Latino; White, not Hispanic/Latino; All Other Races, not Hispanic/Latino) and 4 income groups (household incomes ≤130% of the federal poverty level [FPL], 131%–185% FPL, 186%–350% FPL, and greater than 350% FPL). Income groups representing ≤130% and 131% to 185% FPL correspond with eligibility criteria for free and reduced-price meals, respectively, via the National School Lunch and School Breakfast Programs.⁸⁹ All Other Races includes people who identify as American Indian/Alaska Native, Asian, Native Hawaiian or Pacific Islander, Multi-racial, or another race or ethnicity. More detailed race and ethnicity groups were not examined because of limited sample sizes in national- and state-level health surveillance systems.^63,65,66

Statistical Analysis

Probabilistic sensitivity analysis was performed by simultaneously sampling all parameter values shown in Figures 1 and 2 from specified distributions. Means and 95% uncertainty intervals (UIs) were estimated for all outcomes by running the model 1000 times using Monte Carlo simulations programmed in Java and taking 2.5 and 97.5 percentile values from simulated data, accounting for stochastic and parameter uncertainty.^39,90

RESULTS

The projected reach, cost, impact on population health, and cost-effectiveness outcomes attributable to the implementation of the strategy to provide chilled, filtered water on lunch lines in Massachusetts schools from 2020 to 2029 are shown in Table 1. If implemented in Massachusetts, this strategy is projected to improve drinking water access for 265 000 (95% UI, 259 000–273 000) students in 304 schools in Massachusetts over 10 years. The strategy is projected to prevent 525 cases of obesity (95% UI, 192–880) in the final year of 2029 alone and cost, on average, $4.48 per child (95% UI, $3.54–$5.43) to implement each year. It is likely cost-effective, at a net cost of $72700 per quality-adjusted life year gained (95% UI, $40 000–$234 000). It would cost an estimated total of $11.9 million (95% UI, $9.38 million to $14.4 million) to implement the strategy over 10 years, and $344 000 (95% CI, $85 800–$632 000) in health care costs related to excess weight would be saved.

Table 1.

Projected reach, cost, impact on population health, and cost-effectiveness outcomes due to implementation of a strategy to provide safe chilled water on lunch lines in Massachusetts (United States) schools, 2020—2029, based on simulation modeling and cost-effectiveness analysis

Outcomea	Mean (95% UI)b
Population reachc	265 000 (259 000—273 000)
Increase in servings of water consumed per person per year, 12-oz servingsc	43 (17—69)
Cases of obesity prevented in 2029c	525 (192—880)
QALYsd gained	158 (57—261)
Average annual implementation cost per persone	$4.48 ($3.54—$5.43)
Average annual implementation cost	$1.19 million ($938 000—$1.44 million)
Total implementation cost	$11.9 million ($9.38 million to $14.4 million)
Health care costs saved	$344 000 ($85 800—$632 000)
Healthcare costs saved per $1 invested	$0.03 ($0.01—$0.06)
Net cost per QALY gained	$72 700 ($40 000—$234 000)

^aCosts and health outcomes are summarized over 10 years and are discounted at 3% annually⁵⁵ except where noted. Costs are reported in 2019 US dollars.

^bUI = uncertainty interval. The 95% uncertainty interval is a central range in which 95% of model results fell when the model was run 1000 times, taking into account uncertainty from data sources and population projections.

^cNot discounted 3% annually.

^dQALY, quality-adjusted life year.

^eAverage annualized cost to implement the strategy over the model period (10 years) per person reached over the model period (ie, Average annual implementation cost/Population reach).

Projected costs to implement the strategy, summarized by payer and activity, are shown in Table 2. Over 10 years, schools are projected to bear responsibility for most implementation costs (72% of total costs), followed by the state government (22% of total costs) and school districts (6% of total costs). The largest implementation costs are projected to be for maintaining and cleaning water dispensers over time (50% of total costs), increasing utilities and disposable cup usage over time (21% of total costs, for which schools are responsible), the initial purchasing and installation of water dispensers (19% of total costs, for which the state government is responsible). Smaller shares of costs include administering and managing the water dispenser program at the school district (6%) and state government (3%) levels, testing lead levels in drinking water and addressing any future water quality issues (1%), and training for school food service directors on cleaning and maintaining the chilled water dispensers (<1%).

Table 2.

Projected costs to implement a strategy to provide safe chilled water on lunch lines in Massachusetts (United States) schools, by payer and activity, 2020—2029, based on simulation modeling and cost-effectiveness analysis

Payer	Activity	Average annual implementation cost per person (2019 USD)a	Percentage of costa
State government	Total	$0.99	22%
	Purchase and install chilled water dispensers	$0.85	19%
	Program administration and management, technical support, and training	$0.14	3%
School districts	Total	$0.27	6%
	Grant application and dispenser installation, post-installation water testing, and communications	$0.27	6%
Schools	Total	$3.22	72%
	Train school food service directors on cleaning and maintaining the chilled water dispensers	$0.003	<1%
	Maintain and clean water dispensers	$2.26	50%
	Increase utilities and disposable cup usage	$0.92	21%
	Test lead levels in drinking water and take actions to address identified issues	$0.03	1%
Total a		$4.48	100%

^aCosts are discounted at 3% annually⁵⁵ and reported in 2019 US dollars (USD). Values may not sum to total because of rounding.

Projected health impacts by race, ethnicity, and income are shown in Table 3 and graphically, including comparative impacts by subgroup in Figure 3. Projected health impacts over 10 years were greatest among youth who were Black or African American, Hispanic/Latino, or from households with lower incomes. Of the projected 525 cases of obesity prevented overall in 2029 because of implementation of the strategy, more than half (225) were estimated to be among Hispanic or Latino youth, and nearly 3 in 10 among youth from households with the lowest incomes (130% FPL or less). Compared with impacts among White, non-Hispanic or Latino youth, the strategy is projected to prevent 4.71 (95% UI, 2.91–7.42) times more cases of obesity per 100 000 people among Black or African American youth and 7.24 (95% UI, 5.23–10.3) times more cases of obesity per 100 000 people among Hispanic or Latino youth. Youth of all other races (including Asian, Native Hawaiian, Pacific Islander, American Indian, Alaska Native, and 2 or more races) are also projected to have a 2.45 (95% UI,1.60–3.75) times greater health impact compared with White youth. Compared with impacts among people from households with incomes 351% FPL or greater, the strategy is projected to prevent 2.61 (95% UI, 1.44–3.12) times more cases of obesity per 100 000 people among youth from households with incomes 130% FPL or less, 2.16 (95% UI, 1.44–3.12) times more among youth from households with incomes 131%–185% FPL, and 1.65 (95% UI, 1.18–2.17) times more among youth from households with incomes 186%–350% FPL.

Table 3.

Projected reach and impact on population health by race, ethnicity, and income attributable to implementation of a strategy to provide safe chilled water on lunch lines in Massachusetts (United States) schools, 2020—2029, based on simulation modeling and cost-effectiveness analysis

	Population reach over 10 years Mean (95% UI)a	Cases of obesity prevented in 2029 Mean (95% UI)a	Cases of obesity prevented per 100 000 people in 2029 Mean (95% UI)a
Total	265 000 (259 000 to 273 000)	525 (192—880)	8 (3—14)
Race and ethnicity
Black or African American, not Hispanic or Latino	32 400 (30 300—34 400)	75 (22—135)	17 (5—32)
Hispanic or Latino	96 900 (94 100—101 000)	225 (80—389)	27 (10—46)
White, not Hispanic or Latino	104 000 (101 000—107 000)	169 (57—292)	4 (1—6)
All other races, not Hispanic or Latinob	31 700 (30 500—32 900)	56 (18—100)	9 (3—16)
Household income as a percentage of the federal poverty level (FPL)
130% FPL or less	73 200 (70800—76 500)	153 (53—259)	14 (5—24)
131%—185% FPL	25 600 (24 200—26 800)	52 (19—93)	12 (4—21)
186%—350% FPL	57 800 (55 200—59 800)	118 (41—203)	9 (3—16)
351% FPL or greater	108 000 (105 000—112 000)	201 (70—335)	5 (2—9)

^aUI = uncertainty interval. The 95% UI is a central range in which 95% of model results fell when the model was run 1000 times, taking into account uncertainty from data sources and population projections.

^b“All other races” includes people who identify as American Indian/Alaska Native, Asian, Native Hawaiian or Pacific Islander, Multi-racial, or another race or ethnicity not represented in the categories shown. Although each of these groups represents distinct populations with differences in health opportunities, risks, and outcomes, they are summarized together because of limited data in national- and state-level surveillance systems.^63,65,66

Figure 3.

Projected reach and impact on population health by race, ethnicity, and income attributable to implementation of a strategy to provide safe chilled water on lunch lines in Massachusetts (United States) schools, 2020–2029, based on microsimulation modeling.

DISCUSSION

In this study, researchers estimated the reach, cost, cost-effectiveness, impact on weight status, and obesity-related health equity of a strategy to install chilled, filtered water dispensers on school lunch lines in Massachusetts schools. Results indicate that improving quality drinking water access where needed in lunchroom areas is projected to improve drinking water access for 265 000 students in 304 schools in Massachusetts, prevent 525 cases of childhood obesity in 2029, cost $4.48/year/student, and is likely to be cost-effective ($72700 per QALY).⁹¹ Other projected health-related benefits estimated by cost-effectiveness analysis include increased consumption of drinking water and healthy weight status. Increasing drinking water consumption is especially important because many children and adolescents do not meet recommended levels of adequate daily water intake,² and inadequate hydration may impact cognition.^92,93 Additionally, reducing the proportion of children and adolescents with obesity is a national public health priority.⁹⁴ Results from this model build on the existing evidence from studies that suggest that when schools provide free access to water at lunch, students drink more water,^4–7 consume fewer sugary drinks,^5,8 and have healthier weight status^12–14 by providing additional details on cost and cost-effectiveness of potential implementation scenarios. An implementation strategy that focuses on placement of new water dispensers on or near the lunch line also may be an effective way to ensure that more Massachusetts schools can provide access to free, quality drinking water in the areas where meals are served as required by federal guidance for schools participating in the National School Lunch Program.⁵⁴

The model projects that there will be obesity-related health equity benefits from this approach, an important consideration for some decision-makers.^95–97 The CHOICES microsimulation model does not estimate health or health equity impacts related to potential reductions in exposure to lead or other contaminants in drinking water. However, in the outreach strategy used in this implementation model, the Massachusetts state governmental department partners would prioritize the promotion of installing filtered water dispensers in schools that identify elevated concentrations of lead in their drinking water in school districts located in health-priority communities (ie, those with poorer health outcomes among their populations).³³ Because the intervention modeled would be focused on engaging schools within health-priority communities, it will reach school settings in communities that also had a larger share of households with lower incomes, limited English proficiency, or individuals identifying as Black, Indigenous, or people of color, a strategy that the model findings estimate would improve obesity-related health equity by household income, race, and ethnicity. Features of the built environment, including school buildings serving students, staff, and community members, are critical social determinants of health.⁹⁶ School environments with elevated levels of lead in drinking water⁹¹ or limited access to quality drinking water in meal service areas¹⁹ could contribute to unequal and unjust differences in exposure to contaminants and negatively impact other health outcomes by limiting quality access to drinking water. These health disparities, or differences in health or health-related exposures between more- and less-advantaged groups, are shaped by policies, environments, and actions.⁹⁷ The pursuit of greater health equity includes reducing differences in exposure to less-healthy environments between groups.⁹⁷

School drinking water quality and access are determined by a complex network of policies, practices, and actors who have different information and areas of responsibility regarding water systems, plumbing infrastructure, maintenance, and operational factors that may interact to impact quality and access for students and staff.^35,98,99 State agencies set policies and administer programs that oversee drinking water. School facility and operations leaders are accountable for ensuring quality in building infrastructure and operations, whereas school nutrition service leaders are responsible for food service for students.⁹⁹ Monitoring and regulating contaminants in drinking water, such as lead, usually falls in the purview of environmental protection or public health agencies, whereas education agencies support healthy school environments.^35,98,99 These types of cross-sector collaboration strategies, aligning multi-sectoral services and organizational operations, have proven possible in other settings.¹⁰⁰

The approach in this study is modeled to represent the shared responsibilities amongst these partners at the state, school district, and individual school level, reflected in costs by payer for activities necessary to achieve expected positive health outcomes. The model projects that this strategy will have an annual implementation cost of less than $4.50 per student. This cost per student is comparable to prior cost-effectiveness studies looking at similar nationally implemented school-based water interventions²⁷ but lower than a separate cost–benefit analysis.¹⁰¹ Upfront costs for water quality testing and the purchase and installation of new water fixtures would be shared by state government (22% of total costs) and school districts (6%) through existing federally funded programs and grants, but over 10 years, maintenance and other supporting activities would accrue at the school level (72% of total costs). The costs schools would bear for increased utility fees and cup usage would be budgetary costs, and the costs of maintaining and cleaning water dispensers would be opportunity costs, but are within the scope of usual school-supported functions for building operations and maintenance.

This strategy is likely cost-effective ($72700 per QALY) in preventing child obesity at commonly accepted thresholds. Although no single threshold is used to define cost-effectiveness, commonly used thresholds range from $50 000 to $200 000 per QALY gained.⁸⁸ Data on implementation costs and cost-effectiveness are relevant inputs for organizational decision-makers.¹⁰²

Study Considerations

The proposed strategy builds on existing voluntary lead testing and remediation programs to reduce lead concentration in drinking water in school settings in Massachusetts.²⁴ This allows for the use of specific, local data on school eligibility, student characteristics, and associated costs. In other states, the use of higher “trigger” levels (ie, >1 parts per billion) or differing burdens of lead in drinking water⁹¹ makes some findings less generalizable to other locations in the United States. Researchers assumed that replacement fixtures with filtration in eligible schools would be certified to remove lead and copper from drinking water (but not fluoride) and be placed in cafeteria areas on or near the lunch lines, providing potable drinking water access where meals are served in schools in alignment with federal nutrition standards.⁵⁴ Schools may have required additional replacement fixtures to comply with other standards, such as plumbing codes,¹⁰³ which are not accounted for here. Researchers assumed that the impact of the strategy on weight status from an earlier study¹⁴ would be generalizable to the Massachusetts student population and be maintained over the model period. The estimates of the total benefits in this study are likely conservative because of the lack of inclusion of potential health-related benefits of better hydration² with fluoridated tap water,¹⁰⁴ reduced exposure to lead,²⁵ and any similar benefits that might accrue for school staff. These outcomes are not evaluated as part of the CHOICES microsimulation model, which focuses on the costs and benefits resulting from changes in body weight.^39–41 The use of different base years or time horizons was not evaluated and could impact outcome estimates. This study examines the impacts of using known, effective practices to reduce lead concentrations in drinking water²² (ie, testing and remediation actions) that are not universally practiced^91,98 and are estimated to be cost-saving.²⁵

CONCLUSIONS

A strategy to provide chilled, filtered water dispensers on lunch lines in Massachusetts schools is projected to improve school nutrition environments, increase water consumption, prevent excess weight gain, and improve health equity, and it is likely to be cost-effective. Further research could evaluate the implementation strategies that might lead to further uptake of potentially cost-effective strategies to improve water access and variations in implementation costs and health outcomes.

RESEARCH SNAPSHOT.

Research Question:

What is the reach, cost, cost-effectiveness, and impact on weight status of a strategy to improve access to quality drinking water in school nutrition environments in Massachusetts’ schools?

Key Findings:

A cost-effectiveness analysis of providing chilled, filtered water dispensers on lunch lines in Massachusetts schools is projected to improve drinking water access for 265 000 (95% UI, 259 000–273 000) students in 304 schools, prevent 525 (95% UI, 192–880) cases of childhood obesity in 2029, cost $4.48/year/student (95% UI, $3.54–$5.43), and likely is cost-effective ($72 700 (95% UI, $40 000–$234 000) per QALY).