Abstract
Only 8% of US public schools operate their own community water systems, and thus are subject to the federal Lead and Copper Rule's regulation of water lead levels (WLLs). To date, the absence of parallel water testing data for all other schools has prevented the comparison of WLLs with schools that do not face federal regulation. This study compiled and analyzed newly available school-level WLL data that included water source (on-site well water or public utility) and pipe material data for public schools in New York State located outside of New York City. Despite direct federal regulation, schools that used water from on-site wells had a substantially higher percentage of water fixtures with elevated WLLs. Schools that used both on-site well water and iron pipes in their water distribution system had the highest percentage of elevated fixtures. Variation in water treatment practices was identified as a potential contributing mechanism, as schools that used on-site well water were less likely to implement corrosion control. The study concluded that information about water source and premise plumbing material may be useful to policymakers targeting schools for testing and remediation.
HIGHLIGHTS
Examined the prevalence of fixtures with elevated water lead levels in NY public schools outside of New York City.
Schools that sourced water from on-site wells had nearly twice as many elevated fixtures as schools that used utility water.
Schools with wells and iron pipes in their systems had the highest percentage of elevated fixtures.
Schools that used wells were less likely to treat water for corrosion control than a comparison sample of community water systems.
Graphical Abstract
INTRODUCTION
Since the adoption of the Lead and Copper Rule (LCR) in 1991, US scholars have argued for more comprehensive and unified regulation of school drinking water (Lambrinidou et al. 2010). Under the LCR, public schools fall into two regulatory categories. About 8% of schools operate their own water systems, which are typically wells located at the school sites. These school-controlled water systems must regularly test water lead levels (WLLs) in a sample of school fixtures (Cradock et al. 2019). The other 92% of schools rely on community water systems (CWS) that also serve a year-round residential population. While CWS must test WLLs under the LCR, until recently, no testing was required in schools (Lambrinidou et al. 2010). A 2021 revision to the LCR aimed to address this gap, requiring water testing at 20% of elementary schools and daycares per year; nonetheless, the vast majority of school drinking fixtures are not tested for lead on a regular basis (Rome et al. 2022).
Although direct testing and regulatory oversight might produce lower WLLs in schools that operate their own water systems, these schools face multiple technical and operational conditions associated with worse performance. School-operated water systems are smaller, on average, than CWS, and a large body of literature has demonstrated the disproportionate challenges faced by small systems, including higher marginal service costs, lower revenue, and less qualified, often part-time staff (McFarlane & Harris 2018). In contrast, larger systems can direct more resources toward maintaining compliance and returning to compliance if violations are identified (Switzer et al. 2016). For these reasons, small water systems have the higher rates of water violations than larger systems (McFarlane & Harris 2018).
Of particular concern for WLLs is whether school-operated and community water systems (CWS) differ in their ability to manage pipe corrosion, which is a common source of lead in drinking water (Deshommes et al. 2010). If schools using on-site wells are less likely to implement corrosion control, or are less effective at doing so, then premise plumbing materials are likely to affect WLLs (Gibson et al. 2020). Most notably, iron corrosion has been repeatedly linked to higher WLLs (McFadden et al. 2011; Masters & Edwards 2015; Trueman 2019). However, trace amounts of lead are found in many types of pipes, and studies have also documented the potential for lead to leach from PVC pipes (Zhang & Lin 2015), galvanized steel pipes (Clark et al. 2015; Li et al. 2020), and fixtures or fittings made of brass or bronze, which may accompany any type of pipes (Lei et al. 2018).
Recent crises in Flint, MI; Newark, NJ; and Washington, DC resulted in 18 states passing legislation requiring schools to test and publicly report school WLLs (Pakenham & Olson 2021). Because even the revised LCR requirements provide only limited insight into school WLLs, these laws offer a new opportunity to examine the performance of public water systems in schools, and how the performance differs across the water system types. This study examined the case of New York State, which tested WLLs in all drinking fixtures in all public schools in 2016.
METHODS
Data
In 2016, New York passed legislation requiring public school districts to test all water fixtures in all school buildings for lead. This testing was separate from LCR lead testing, and thus not subject to federal consequences (NYS Public Health Law Article 13, Title X). Outlets were tested using 250 mL ‘first-draw’ samples; that is, no water was run from taps before sample collection. Water was also required to be motionless in the pipes for between 8 and 18 hours prior to collection. These data were compiled by the New York State Department of Health and included the number of water fixtures in each school, the number of fixtures tested for lead, and the number that tested above 15 parts per billion (ppb), which is the action level established by the LCR. They also included street addresses, counties, and school district identifiers. About 86% of lead tests were conducted during the 2016–17 school year, but a small number occurred as late as February 2019 (Supplementary Material, Figure A1).
The lead testing data were matched to data collected during New York school building inspections in 2015. Inspections were conducted in person by teams that included at least one licensed architect or engineer. They were available online in PDF form for NY public school districts excluding New York City and included building age, water source (i.e., municipal/utility, well, or other), and pipe material/s used in the water distribution system. The building inspection items used in this study can be found in Supplementary Material, Figure A2.
These data were also merged with data from the Safe Drinking Water Information System (SDWIS). This repository contains information on public water systems across the USA, including system location, federal classification (community vs. non-transient, non-community), population served, and water treatment processes. Schools that used on-site wells were matched directly by name, with a match rate of 97%. Although the remainder of the schools could not be directly matched to their individual water systems, a plausible comparison sample of CWS in New York was created.
Samples
The sample of NY schools included 2,572 school buildings that reported both lead testing and water system data and had non-zero enrollment in the 2016–17 school year, per the National Center for Educational Statistics Common Core of Data. This excluded New York City schools, which did not publicly report water system data, and charter schools, which were not subject to lead testing legislation.
In total, 558 school buildings were omitted from the sample because they were missing water system data (N = 294), reported a water source other than a utility or on-site well (N = 5), or did not report any enrollment in grades pre-K-12 (N = 259). Omitted buildings had slightly fewer outlets tested on average, and a higher average percentage of fixtures tested for lead above 15 ppb (Supplementary Material, Table A1).
SDWIS data were used to construct a comparison sample that included 1,683 CWS in New York that served at least 100 customers. Since CWS must, by definition, serve a residential population year-round, any systems that served fewer than 100 customers (27% of all CWS in New York) were unlikely to serve both school and residential populations; fewer than 1% of schools in the lead testing data enrolled fewer than 100 students. Consistent with the lead testing sample, New York City's water system was also excluded from this analysis.
Measures
School drinking water source
Building inspectors reported each school's drinking water source as either ‘municipal or utility’ (‘utility’), ‘well,’ or ‘other.’ Five schools used a source other than utility or well water and were omitted from the analysis.
Lead in drinking water
Each school reported the number of drinking fixtures that were tested for lead as well as the number that measured above 15 ppb (the EPA action level). The central outcome was the percent of fixtures at each school that tested above the 15-ppb threshold.
Water distribution system pipe materials
Indicators were constructed for water distribution systems that used copper, iron, galvanized, PVC, or lead pipes. Many schools reported multiple pipe materials in their water distribution systems, so these indicators were not mutually exclusive.
Building construction before 1986
The data did not contain information about other plumbing components (e.g., solder, joints, fittings, or fixtures), which may also contain lead. For this reason, an indicator variable was created for building construction before 1986, when the Safe Drinking Water Act (SDWA) was amended to prohibit the use of pipes containing more than 8% lead, and solder or flux containing more than 0.2% lead.
Corrosion control
Water treatment processes for each water system were reported in the SDWIS, along with the objective behind each treatment process. An indicator was constructed for water systems that used any treatment process with the intended objective of controlling corrosion.
Other water system characteristics
Each water system in the SDWIS also reported the number of customers served, water source (groundwater, surface water, or purchased ground/surface water), and auspice (federal, local, state, private, or public/private).
Statistical analysis
Ordinary least squares (OLS) models were estimated to examine associations between the percent of elevated fixtures at each school, water source (utility or on-site well water), and water distribution system pipe materials (copper, iron, galvanized, or PVC). Additional models included interactions between on-site well water and the presence of iron pipes, galvanized pipes, and PVC pipes; copper pipes served as the omitted category.
Linear probability models were then estimated predicting the association between water system type (i.e., school with on-site water or community water system) and the use of corrosion control. Controls were introduced for population served, water source, auspice, and building date before 1986.
RESULTS AND DISCUSSION
Characteristics of NY schools by water source
About 7.4% of schools (191 of 2,572) used water from an on-site well, while the rest sourced water from a utility (Table 1). The average number of fixtures tested for lead was similar across water sources, but nearly twice as many fixtures tested above 15 ppb in schools served by on-site well water (6.5 and 10.9% for utility and on-site wells, respectively). If this testing was subject to the LCR, the average NY public school served by an on-site well would be out of compliance (i.e., more than 10% of fixtures exceeding 15 ppb), although none of these systems was cited for violation with the smaller sample of LCR samples required.
New York public school characteristics by drinking water source
Variable . | Utility . | On-site well . |
---|---|---|
Lead testing | ||
Number of outlets tested | 86.2 | 90.9 |
Percent over 15 parts per billion | 6.5 | 10.9*** |
School building characteristics | ||
Water distribution pipe materials (%) | ||
Copper | 96.5 | 95.3 |
Iron | 20.9 | 28.3* |
Galvanized | 29.8 | 24.1 |
PVC | 1.8 | 5.8*** |
Lead | 0.2 | 0.5 |
Other material | 0.6 | 0.5 |
Built before 1986 | 94.4 | 90.6* |
N | 2,381 | 191 |
Variable . | Utility . | On-site well . |
---|---|---|
Lead testing | ||
Number of outlets tested | 86.2 | 90.9 |
Percent over 15 parts per billion | 6.5 | 10.9*** |
School building characteristics | ||
Water distribution pipe materials (%) | ||
Copper | 96.5 | 95.3 |
Iron | 20.9 | 28.3* |
Galvanized | 29.8 | 24.1 |
PVC | 1.8 | 5.8*** |
Lead | 0.2 | 0.5 |
Other material | 0.6 | 0.5 |
Built before 1986 | 94.4 | 90.6* |
N | 2,381 | 191 |
Note: Sample included New York public schools that reported lead testing data in Compliance Year 2016, excluding New York City. ‘Utility’ included both privately- and publicly-owned services. Schools could report more than one type of pipe material, so the total sums to more than 100%.
*p < 0.05; **p < 0.01; ***p < 0.001.
Examination of water distribution system characteristics demonstrated that nearly all school systems contained some copper pipes (over 96 and 95% for schools using utilities and on-site wells, respectively; Table 1). Many systems used multiple pipe materials, as about half contained either iron or galvanized pipes as well. PVC pipes were substantially less common, and few schools used pipes made of any other materials. Both iron and PVC pipes were more commonly used in schools with on-site wells (iron: 28–21%; PVC: 6–2%).
Over 90% of schools were built prior to 1986, when an Safe Drinking Water Act amendment restricted the use of lead in pipes and solder in public water systems. Schools with on-site wells were somewhat less likely to have been built prior to this amendment (90.6–94.4%).
Interaction between on-site well water and water system pipe materials
The results of Ordinary least squares regression models demonstrated that schools that sourced water from on-site wells had a substantially higher percentage of fixtures with elevated WLLs (β = 4.4, p < 0.001; Table 2, Column 1). Higher rates of iron and PVC pipe used at schools with on-site wells did not account for the difference in elevated fixtures by water source (Table 2, Column 2).
Ordinary least squares estimates predicting the association between on-site school wells, pipe material, and the percentage of fixtures testing for the lead over 15 ppb
Building characteristic . | (1) . | (2) . | (3) . | (4) . | (5) . |
---|---|---|---|---|---|
On-site well | 4.4*** | 4.4*** | 2.8* | 2.9* | 1.0 |
(0.9) | (0.9) | (1.2) | (1.1) | (1.1) | |
Iron pipes | 0.1 | −0.3 | −0.4 | −1.1 | |
(0.6) | (0.6) | (0.6) | (0.6) | ||
Galvanized pipes | 0.2 | 0.3 | 0.0 | 0.1 | |
(0.5) | (0.5) | (0.5) | (0.5) | ||
PVC pipes | −0.2 | −0.9 | −1.0 | −1.8 | |
(1.6) | (1.8) | (1.8) | (1.7) | ||
On-site well and iron pipes | 5.1** | 5.1** | 4.5* | ||
(1.9) | (1.9) | (1.9) | |||
On-site well and galvanized pipes | −1.0 | −1.1 | −0.1 | ||
(2.0) | (2.0) | (1.9) | |||
On-site well and PVC pipes | 4.7 | 5.2 | 2.8 | ||
(4.0) | (4.0) | (3.8) | |||
Built before 1986 | 3.6*** | 4.7*** | |||
(1.0) | (0.9) | ||||
Constant | 6.5*** | 6.4*** | 6.5*** | 3.2*** | 2.5** |
(0.2) | (0.3) | (0.3) | (0.9) | (0.9) | |
N | 2,572 | 2,572 | 2,572 | 2,572 | 2,572 |
County fixed effects | X |
Building characteristic . | (1) . | (2) . | (3) . | (4) . | (5) . |
---|---|---|---|---|---|
On-site well | 4.4*** | 4.4*** | 2.8* | 2.9* | 1.0 |
(0.9) | (0.9) | (1.2) | (1.1) | (1.1) | |
Iron pipes | 0.1 | −0.3 | −0.4 | −1.1 | |
(0.6) | (0.6) | (0.6) | (0.6) | ||
Galvanized pipes | 0.2 | 0.3 | 0.0 | 0.1 | |
(0.5) | (0.5) | (0.5) | (0.5) | ||
PVC pipes | −0.2 | −0.9 | −1.0 | −1.8 | |
(1.6) | (1.8) | (1.8) | (1.7) | ||
On-site well and iron pipes | 5.1** | 5.1** | 4.5* | ||
(1.9) | (1.9) | (1.9) | |||
On-site well and galvanized pipes | −1.0 | −1.1 | −0.1 | ||
(2.0) | (2.0) | (1.9) | |||
On-site well and PVC pipes | 4.7 | 5.2 | 2.8 | ||
(4.0) | (4.0) | (3.8) | |||
Built before 1986 | 3.6*** | 4.7*** | |||
(1.0) | (0.9) | ||||
Constant | 6.5*** | 6.4*** | 6.5*** | 3.2*** | 2.5** |
(0.2) | (0.3) | (0.3) | (0.9) | (0.9) | |
N | 2,572 | 2,572 | 2,572 | 2,572 | 2,572 |
County fixed effects | X |
Note: Sample included New York public schools that reported lead testing data in Compliance Year 2016, excluding New York City. Standard errors in parentheses.
*p < 0.05; **p < 0.01; ***p < 0.001.
The association between pipe materials and WLLs differed for schools served by utilities and those served by wells (Table 2, Column 3). Most notably, the interaction between on-site well water and iron pipes was large and statistically significant (β = 5.1, p < 0.001), indicating that schools with both on-site wells and iron pipes had a disproportionately higher percentage of fixtures test for elevated levels of lead. The interaction between on-site well water and PVC pipes was similarly large (β = 4.7, n.s.), but much more imprecisely measured due to many fewer schools using PVC pipes, and not significantly different from zero. These interactions only partially accounted for the difference in lead by water source, as the main effect for schools served by on-site wells was still significantly different from zero (β = 2.8, p < 0.05). The interaction was not explained by differences in whether schools were built prior to 1986 (Table 2, Column 4), although schools built prior to the Safe Drinking Water Act amendment had a significantly higher percentage of fixtures test above 15 ppb (β = 3.6, p < 0.001).
New York water system characteristics by system type
Water system characteristic . | Community water systems . | Schools with on-site wells . |
---|---|---|
Water treatment includes corrosion control | 21.4 | 14.6* |
Size (%) | ||
Small ( < 500 served) | 46.8 | 38.4* |
Medium (500–5,000) | 38.7 | 61.6*** |
Large (≥5,000) | 14.3 | 0.0*** |
Source (%) | ||
Groundwater | 62.0 | 100*** |
Surface water | 12.1 | 0*** |
Purchased (ground or surface) | 25.7 | 0*** |
Auspice (%) | ||
Local government | 62.4 | 56.8 |
Private | 33.0 | 26.5 |
Public–private partnership | 2.3 | 16.2*** |
State/federal government | 2.1 | 0.5 |
N | 1,683 | 185 |
Water system characteristic . | Community water systems . | Schools with on-site wells . |
---|---|---|
Water treatment includes corrosion control | 21.4 | 14.6* |
Size (%) | ||
Small ( < 500 served) | 46.8 | 38.4* |
Medium (500–5,000) | 38.7 | 61.6*** |
Large (≥5,000) | 14.3 | 0.0*** |
Source (%) | ||
Groundwater | 62.0 | 100*** |
Surface water | 12.1 | 0*** |
Purchased (ground or surface) | 25.7 | 0*** |
Auspice (%) | ||
Local government | 62.4 | 56.8 |
Private | 33.0 | 26.5 |
Public–private partnership | 2.3 | 16.2*** |
State/federal government | 2.1 | 0.5 |
N | 1,683 | 185 |
Note: Sample included New York public water systems reporting data to the US Environmental Protections Agency in 2015. Less than 1% of community water systems were located at schools (either residential colleges or boarding schools).
*p < 0.05; **p < 0.01; ***p < 0.001.
The county-level percentage of school drinking fixtures testing for lead over 15 ppb across New York State. Black dots represent schools that use on-site wells.
The county-level percentage of school drinking fixtures testing for lead over 15 ppb across New York State. Black dots represent schools that use on-site wells.
Estimated percentage of the New York school drinking fixtures testing for lead over 15 ppb, by water source and pipe material. Error bars represent 95% confidence intervals.
Estimated percentage of the New York school drinking fixtures testing for lead over 15 ppb, by water source and pipe material. Error bars represent 95% confidence intervals.
These findings are consistent with the notion that premise plumbing is more strongly related to WLLs in schools served by on-site wells. In addition, the interaction between wells and iron pipes is consistent with small-scale observational studies that found higher levels of lead in water systems connected to iron water mains (Camara et al. 2013; Trueman 2019), as well as other research demonstrating that lead leaches into water more readily as iron levels increase (McFadden et al. 2011; Masters & Edwards 2015).
Differences in use of corrosion control by school water source
Estimated differences in the use of corrosion control treatment between community water systems and schools with on-site wells. Error bars represent 95% confidence intervals.
Estimated differences in the use of corrosion control treatment between community water systems and schools with on-site wells. Error bars represent 95% confidence intervals.
There were other notable differences between on-site wells and CWS. All schools with on-site wells used a groundwater source, whereas about a quarter of CWS purchased their water, and 12% used a surface water source. On-site wells were also much more likely than CWS to be controlled by a public–private partnership (16.2–2.3%, p < 0.001; Table 3). After accounting for differences in water source and water system auspice (Figure 3, Panel D), schools with on-site wells were 10 percentage points less likely to use corrosion control than were CWS. This means that the differences in the use of corrosion control treatment went beyond size differences across system types. One possibility is that schools that control their own water systems face similar problems to small CWS, even if they serve a larger population. In particular, staff overseeing these water systems may have poorer qualifications than their counterparts at CWS, and they may be part-time or serve multiple roles, reducing their ability to effectively implement corrosion control.
CONCLUSION
This study compiled and analyzed newly available data on lead in New York school drinking water and examined the association between schools' use of on-site wells and elevated WLLs. A strength of the study is that the data included the full population of public schools in the state, and the state legislation required sampling of all fixtures in each building, in contrast to the smaller sample of fixtures required by the LCR. The study thus provided the first comparison of schools in the same state that faced different levels of federal regulation.
Schools that used on-site wells had a higher percentage of fixtures with elevated WLLs, and those that used both on-site wells and iron pipes in their water distribution systems had the highest levels of lead. Systems with on-site wells were less likely to implement corrosion control, which may have contributed to differences in WLLs across systems.
Findings were consistent with recent large-scale studies in Virginia and North Carolina documenting high levels of lead in private wells (Pieper et al. 2015; Gibson et al. 2020) and demonstrated that students in schools that use wells may be exposed to the same water lead risks as private well users not subject to regulation or testing under the LCR (Pieper et al. 2018). Better understanding the organizational capacity and treatment factors that may be associated with the differences described in the study is an important area for future research. Despite facing direct regulation and regular testing, these schools had a higher portion of fixtures with elevated WLLs than did schools served by CWS.
A limitation of the study is that state legislation required the testing of each fixture only once. Previous research has demonstrated substantial fixture-level sampling variation, including variation over time, by water temperature, and by period of stagnation (Triantafyllidou et al. 2007).
As more school water testing data become available, both from the newly revised LCR and state laws, there will be more opportunities to examine the distribution of school WLLs in other states. This study demonstrated the value of collecting information about both water source and premise plumbing materials for understanding factors that influence school WLLs. Additionally, the collection of additional information, such as water pH, use of brass components, and local soil composition, should be considered. These data could play a vital role in assisting state and district policymakers to make decisions about where to target limited resources.
ACKNOWLEDGEMENT
Special thanks to Professor John A. Higgins, whose feedback improved the manuscript immensely.
AUTHOR CONTRIBUTIONS
All authors contributed to the study conception, design, and preparation of the initial draft and subsequent revisions. Data analysis was performed by S.L.. All authors read and approved the final manuscript.
FUNDING
The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.
DATA AVAILABILITY STATEMENT
All relevant data are available from an online repository or repositories (https://github.com/LathamS12/Lead_and_wells_in_NYS).
CONFLICT OF INTEREST
The authors declare there is no conflict.