ABSTRACT
Tap water lead testing programs in the U.S. need improved methods for identifying high-risk facilities to optimize limited resources. In this study, machine-learned Bayesian network (BN) models were used to predict building-wide water lead risk in over 4,000 child care facilities in North Carolina according to maximum and 90th percentile lead levels from water lead concentrations at 22,943 taps. The performance of the BN models was compared to common alternative risk factors, or heuristics, used to inform water lead testing programs among child care facilities including building age, water source, and Head Start program status. The BN models identified a range of variables associated with building-wide water lead, with facilities that serve low-income families, rely on groundwater, and have more taps exhibiting greater risk. Models predicting the probability of a single tap exceeding each target concentration performed better than models predicting facilities with clustered high-risk taps. The BN models' Fß-scores outperformed each of the alternative heuristics by 118-213%. This represents up to a 60% increase in the number of high-risk facilities that could be identified and up to a 49% decrease in the number of samples that would need to be collected by using BN model-informed sampling compared to using simple heuristics. Overall, this study demonstrates the value of machine-learning approaches for identifying high water lead risk that could improve lead testing programs nationwide.
Subject(s)
Drinking Water , Lead , Humans , Child , Lead/analysis , Bayes Theorem , Child Care , Water , Decision MakingABSTRACT
Objectives. To evaluate lead levels in tap water at licensed North Carolina child care facilities. Methods. Between July 2020 and October 2021, we enrolled 4005 facilities in a grant-funded, participatory science testing program. We identified risk factors associated with elevated first-draw lead levels using multiple logistic regression analysis. Results. By sample (n = 22 943), 3% of tap water sources exceeded the 10 parts per billion (ppb) North Carolina hazard level, whereas 25% of tap water sources exceeded 1 ppb, the American Academy of Pediatrics' reference level. By facility, at least 1 tap water source exceeded 1 ppb and 10 ppb at 56% and 12% of facilities, respectively. Well water reliance was the largest risk factor, followed by participation in Head Start programs and building age. We observed large variability between tap water sources within the same facility. Conclusions. Tap water in child care facilities is a potential lead exposure source for children. Given variability among tap water sources, it is imperative to test every source used for drinking and cooking so appropriate action can be taken to protect children's health. (Am J Public Health. 2022;112(S7):S695-S705. https://doi.org/10.2105/AJPH.2022.307003).
Subject(s)
Drinking Water , Lead , Child , Child Care , Humans , Lead/analysis , North Carolina , Water/analysis , Water SupplyABSTRACT
Exposure to traffic-related fine particulate matter air pollution (tr-PM2.5) has been associated with adverse health outcomes including preterm birth and low birthweight. In-vehicle exposure to tr-PM2.5 can contribute substantially to total tr-PM2.5 exposure. Because average commuting habits of women differ from men, a research gap is estimating in-vehicle tr-PM2.5 exposures for women commuters. For 46 women commuters in the Washington, D.C. metro area, we measured personal exposure to PM2.5 during all vehicle trips taken in a 48-h sampling period. We also measured 48-h integrated PM2.5 chemical constituents including black carbon and zinc. We identified trip times using vehicle monitors, specifically on-board diagnostics data loggers and dashboard cameras. For 386 trips, we estimated associations between PM2.5 exposure and trip characteristics using linear mixed models accounting for participant, day, and time of day. Additionally, we estimated associations between rush hour trip PM2.5 and 48-h integrated PM2.5 chemical constituents using linear models. Exposure to PM2.5 during trips was 1.9 µg/m3 (95% confidence interval (CI): 0.9, 2.9) higher than non-trip exposures and rush hour trip exposures were 3.2 µg/m3 (95% CI: 1.8, 4.6) higher than non-trip exposures on average. We did not find differences in PM2.5 exposure by trip length. Although concentrations of tr-PM2.5 chemical constituents were generally positively associated with rush hour trip PM2.5, associations were weak indicating that other settings contribute to total tr-PM2.5 exposure. Our study demonstrates the utility of combining vehicle monitors and personal PM2.5 monitors for estimating personal exposure to tr-PM2.5. Future work will investigate whether additional data collected by vehicle monitors, such as traffic and speed, can be leveraged to better understand tr-PM2.5 exposure among commuters.
Subject(s)
Air Pollutants , Air Pollution , Premature Birth , Air Pollutants/analysis , Air Pollution/analysis , Environmental Exposure/analysis , Environmental Monitoring , Female , Humans , Infant, Newborn , Male , Particulate Matter/analysis , Pregnancy , Vehicle Emissions/analysis , WashingtonABSTRACT
Particulate matter (PM) concentrations have decreased dramatically over the past 20 years, thus lower method detection limits (MDL) are required for these measurements. Energy-dispersive X-ray fluorescence (XRF) spectroscopy is used to quantify multiple elements simultaneously in the U.S. Environmental Protection Agency (EPA) Chemical Speciation Network (CSN). Inductively-coupled plasma mass spectrometry (ICP-MS) is an alternative analysis with lower MDL for elements. Here, we present a side-by-side comparison of XRF and ICP-MS for elements in PM2.5 samples collected via the EPA's CSN. For ICP-MS, a simple extraction and ICP-MS analysis technique was applied to a wide variety of samples to minimize effort and cost and serve as a feasibility test for a large monitoring network. Filter samples (N = 549) from various urban locations across the US were analyzed first analyzed via XRF at UC Davis and then ICP-MS at RTI International. Both methods measured 29 of the same elements out of the 33 usually reported to CSN. Of these 29, 14 elements (Na, Mg, Al, K, Ca, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Pb) were found to be frequently detected (i.e. had more than 10% of values above both XRF and ICP-MS MDL). ICP-MS was found to have lower MDL for 26 out of 29 elements, namely Na, Mg, Al, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, As, Se, Rb, Sr, Zr, Ag, Cd, In, Sn, Sb, Cs, Ba, Ce, Pb; conversely, XRF had lower MDL for 3 elements, namely, P, K, Zn. Intra-method quality checks using (1) inter-elemental inspection of scatter plots using a priori knowledge of element sources and (2) scatter plots of routine versus collocated measurements reveal that ICP-MS exhibits better measurement precision. Lower detection limits for element measurements in nationwide PM monitoring networks would benefit human-health and source apportionment research.Implications: We demonstrate that ICP-MS with adilute-acid digestion method would significantly improve the element detection rates and thus be avaluable addition to the current analysis techniques for airborne PM samples in anationwide monitoring network. In this paper, we show that a hybrid method of elemental analysis for airborne particulate matter (PM) would significantly improve the detection rates for elements in PM. This would be a valuable addition to the current analysis techniques for airborne PM samples in nationwide and other large-scale monitoring networks, such as the EPA's Chemical Speciation Network (CSN). The techniques explored in this study (i.e., X-ray Fluorescence Spectroscopy or XRF and Inductively Coupled Plasma-Mass Spectrometry or ICP-MS) are relevant to the PM monitoring and regulatory community audience of JAWMA, especially agencies and states that are already involved in CSN. In addition, our results outline considerations that give insight on factors to consider for other large-scale and long-term ambient air monitoring efforts.
Subject(s)
Lead , Particulate Matter , United States , Humans , United States Environmental Protection AgencyABSTRACT
Personal exposure to PM2.5, and the elemental composition therein, may vary greatly from ambient measurements at fixed monitoring sites. Here, we characterized the differences between personal, indoor, and outdoor concentrations of PM2.5-bound elements, and predicted personal exposures to 21 PM2.5-bound elements. Personal-indoor-outdoor PM2.5 filter samples were collected for five consecutive days across two seasons from 66 healthy non-smoking retired adults in Beijing (BJ) and Nanjing (NJ), China. Personal element-specific models were developed using liner mixed effects models and evaluated by R2 and root mean square error (RMSE). The mean (SD) concentrations of personal exposures varied by element and city, ranging from 2.5 (1.4) ng/m3 for Ni in BJ to 4271.2 (1614.8) ng/m3 for S in NJ. Personal exposures to PM2.5 and most elements were significantly correlated with both indoor and outdoor (except Ni in BJ) measurements, but frequently exceeded indoor levels and fell below outdoor levels. Indoor and outdoor PM2.5 elemental concentrations were the strongest determinants of most personal elemental exposures, with RM2 ranging from 0.074 to 0.975 for indoor and from 0.078 to 0.917 for outdoor levels, respectively. Home ventilation conditions (especially window opening behavior), time-activity patterns, meteorological factors, household characteristics, and season were also key factors influencing personal exposure levels. The final models accounted for 24.2 %-94.0 % (RMSE: 0.135-0.718) of the variance in personal PM2.5 elemental exposures. By incorporating these crucial determinants, the modeling approach used here can improve PM2.5-bound elemental exposure estimates and better associate compositionally dependent PM2.5 exposures and health risks.
Subject(s)
Air Pollutants , Air Pollution, Indoor , Air Pollutants/analysis , Particulate Matter/analysis , Environmental Exposure/analysis , Environmental Monitoring , Cities , China , Air Pollution, Indoor/analysis , Particle SizeABSTRACT
Limited information is available regarding chemical water quality at the tap in Guatemala City, preventing individuals, water utilities, and public health authorities from making data-driven decisions related to water quality. To address this need, 113 participants among households served by a range of water providers across the Guatemala City metropolitan area were recruited as participatory scientists to collect first-draw and flushed tap water samples at their residence. Samples were transported to the U.S. and analyzed for 20 metals and 25 per- and polyfluoroalkyl substances (PFAS). At least one metal exceeded the Guatemalan Maximum Permissible Limit (MPL) for drinking water in 63% of households (n = 71). Arsenic and lead exceeded the MPL in 33.6% (n = 38) and 8.9% (n = 10) of samples, respectively. Arsenic was strongly associated with groundwater while lead occurrence was not associated with location, water source, or provider. One or more PFAS were detected in 19% of samples (n = 21, range 2.1-64.2 ppt). PFAS were significantly associated with the use of plastic water storage tanks but not with location, water source, or provider. Overall, the high prevalence of arsenic above the MPL in Guatemala City tap water represents a potential health risk that current water treatment processes are not optimized to remove. Furthermore, potential contaminants from premise plumbing and storage, including lead and PFAS, represent additional risks requiring further investigation and public engagement.
Subject(s)
Arsenic , Fluorocarbons , Water Pollutants, Chemical , Fluorocarbons/analysis , Guatemala , Humans , Metals , Water Pollutants, Chemical/analysisABSTRACT
The U.S. Environmental Protection Agency (EPA)'s PM2.5 Chemical Speciation Network (CSN) and the Interagency Monitoring of Protected Visual Environments (IMPROVE) network use X-ray fluorescence (XRF) analysis to quantify trace elements in samples of fine particles less than 2.5 microns in aerodynamic diameter (PM2.5). Methods for calculating uncertainty values for XRF results vary considerably among laboratories and instrument makes and models. To support certain types of modeling and data analysis, uncertainty estimates are required that are consistent within and between monitoring programs, and that are independent of the laboratories that performed the analyses and the analytical instrumentation used. The goal of this work was to develop a consensus model for uncertainties associated with XRF analysis of PM2.5 filter samples. The following important components of uncertainty are included in the model described herein: variability in peak area, calibration, field sampling, and attenuation of X-ray intensity for light elements. This paper includes a detailed analysis of how attenuation uncertainties for light elements are derived. For the remaining uncertainty components included in the model, an approach and recommendations are presented to ensure that laboratories performing this type of analysis can use similar equations and parameterizations. By applying this uniform approach, it is illustrated how the uncertainties reported by the CSN and IMPROVE network laboratories can be brought into very good agreement. The proposed method is best applied at the time of data generation, but retrospective estimation of uncertainties in existing data-sets is also possible. This paper serves to document the equations used for calculating the uncertainties in speciated PM2.5 data currently being posted on EPA's Air Quality System database for the PM2.5 CSN program.
Subject(s)
Air Pollutants, Occupational/analysis , Particulate Matter/analysis , Algorithms , Calibration , Data Interpretation, Statistical , Environmental Monitoring/methods , Environmental Monitoring/statistics & numerical data , Filtration , Particle Size , Spectrometry, X-Ray Emission , United States , United States Environmental Protection AgencyABSTRACT
Zinc deficiency and excess can result in adverse health outcomes. There is conflicting evidence regarding whether excess or deficient zinc in the diet can contribute to carcinogenicity. The objective of this study was to characterize zinc carbonate basic for use as a source of dietary zinc in a rodent toxicity and carcinogenicity study investigating the effects of zinc deficiency and excess. Because of the complex chemistries of zinc carbonate basic compounds, inconsistent nomenclature, and literature and reference spectra gaps, it was necessary to employ multiple analytical techniques, including Karl Fischer titration, combustion analysis, inductively coupled plasma-optical emission spectrometry, X-ray diffraction, infrared spectroscopy, X-ray fluorescence spectrometry, and thermogravimetric analysis to characterize the test article. Based on the collective evidence and through the process of elimination, the test article was found to be composed mainly of zinc carbonate basic with zinc oxide as a minor component. The zinc content was determined to be 56.6% (w/w) with heavy metals such as arsenic, cadmium, mercury and lead below the limit of quantitation of less than or equal to 0.01%. The test material was stable at ambient temperature. Based on the work described in this manuscript, the test article was suitable for use as a source of zinc in studies of deficiency and excess in the diet.