Article pubs.acs.org/est
Contribution of Particle-Size-Fractionated Airborne Lead to Blood Lead during the National Health and Nutrition Examination Survey, 1999−2008 Qingyu Meng,† Jennifer Richmond-Bryant,*,‡ J. Allen Davis,‡ Jonathan Cohen,§ David Svendsgaard,‡ James S. Brown,‡ Lauren Tuttle,∥ Heidi Hubbard,§ Joann Rice,⊥ Lisa Vinikoor-Imler,‡ Jason D. Sacks,‡ Ellen Kirrane,‡ Dennis Kotchmar,‡ Erin Hines,‡ and Mary Ross‡ †
School of Public Health, Rutgers University, 683 Hoes Lane West, Piscataway, New Jersey 08854, United States National Center for Environmental Assessment, U.S. Environmental Protection Agency, 109 T. W. Alexander Drive, B243-01, Research Triangle Park, North Carolina 27711, United States § ICF International, 9300 Lee Highway, Fairfax, Virginia 22031, United States ∥ School of Architecture, The University of Texas at Austin, 1 University Station B7500, Austin, Texas 78712-0222, United States ⊥ Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, 109 T. W. Alexander Drive, C304-04, Research Triangle Park, North Carolina 27711, United States ‡
S Supporting Information *
ABSTRACT: The objective of this work is to examine associations between blood lead (PbB) and air lead (PbA) in particulate matter measured at different size cuts by use of PbB concentrations from the National Health and Nutrition Examination Survey and PbA concentrations from the U.S. Environmental Protection Agency for 1999−2008. Three size fractions of particle-bound PbA (TSP, PM10, and PM2.5) data with different averaging times (current and past 90-day average) were utilized. A multilevel linear mixed effect model was used to characterize the PbB−PbA relationship. At 0.15 μg/m3, a unit decrease in PbA in PM10 was significantly associated with a decrease in PbB of 0.3−2.2 μg/ dL across age groups and averaging times. For PbA in PM2.5 and TSP, slopes were generally positive but not significant. PbB levels were more sensitive to the change in PbA concentrations for children (1−5 and 6−11 years) and older adults (≥60 years) than teenagers (12−19 years) and adults (20−59 years). For the years following the phase-out of Pb in gasoline and a resulting upward shift in the PbA particle size distribution, PbA in PM10 was a statistically significant predictor of PbB. The results also suggest that age could affect the PbB−PbA association, with children having higher sensitivity than adults.
■
INTRODUCTION Numerous epidemiologic studies have shown that exposure to lead (Pb) can cause a broad range of adverse health effects, including impaired neurological development and function, heart disease, and chronic kidney disease.1 In epidemiologic studies, blood Pb (PbB) instead of ambient air Pb (PbA) is used as an exposure indicator. However, the national ambient air quality standard (NAAQS) for Pb is set for PbA (not PbB) to protect public health. Personal exposure to Pb has a multipathway and multimedia nature. As such, the use of PbB as a biomarker of exposure reflects total personal exposure to Pb, including Pb in air, soil, water, food, and paint-containing dust. Air-related pathways, or the pathways where Pb is airborne during some portion of its path from a source to human exposure, can result in inhalation of PbA and possible ingestion of deposited soil Pb. Therefore, assessment of the association between PbB and PbA is critical to understanding the impact of PbA on PbB and associated adverse health effects. © 2013 American Chemical Society
The associations between PbB and PbA have been quantified in very few studies, and only one older study had been conducted to characterize the relationship between PbA and PbB for the United States on a national scale.2 In addition, most of the studies reporting associations between PbB and PbA were conducted when tetramethyllead and tetraethyllead were still used as antiknock agents in gasoline in the United States and abroad. These studies observed that (1) the use of leaded gasoline accounted for more than 50% of PbB during the second National Health and Nutrition Examination Survey (NHANES II, 1976−1980);2 (2) PbB was significantly associated with PbA when mean PbA was close to 1 μg/ m3;3−6 and (3) the slope factors d[PbB]/d[PbA] of the PbB Received: Revised: Accepted: Published: 1263
September 6, 2013 December 16, 2013 December 17, 2013 December 17, 2013 dx.doi.org/10.1021/es4039825 | Environ. Sci. Technol. 2014, 48, 1263−1270
Environmental Science & Technology
■
and PbA associations ranged from 3 to 7, which corresponded to a decline in PbB of 3−7 μg/dL per 1 μg/m3 decline of PbA.3,4 The general findings from past studies might not be applicable to the current relationship between PbB and PbA. First, the ambient PbA concentration has markedly declined since the ban of leaded antiknock agents in gasoline was followed by an 89% reduction in PbA since 1980.7 This dramatic reduction is potentially problematic when the association between PbB and PbA is examined. The PbB− PbA relationship is commonly characterized with a log−log regression, and therefore the curve of d[PbB]/d[PbA] is not constant but depends on the PbA concentration. A steeper d[PbB]/d[PbA] is expected when PbA levels are smaller. Therefore, d[PbB]/d[PbA] needs to be estimated under current PbA levels. Second, ambient PbA sources and particle size distributions have changed since the ban of leaded gasoline; emissions from gasoline combustion used to be the dominant source of PbA prior to the phase-out completed in the mid-1990s.8 After the phase-out, aviation fuel emissions and industrial emissions have become major sources of PbA.8,9 The shift in emission sources of PbA has caused a change in particle size distributions of PbA. The mass median diameter of PbA has shifted from less than 2.5 μm prior to the phase-out of leaded gasoline to somewhere between 2.5 and 10 μm after the phase-out; Pb enrichment in coarse particles varies with distance from the source and the type of source.10 Lee et al.11 reported a mass median diameter less than 0.8 μm around the time when leaded gasoline was in peak use. The change in emission sources and particle size distributions suggests that there has been a change in exposure pathways and PbB−PbA associations. Third, the ambient air monitoring network has evolved over the past 30 years. The PbB−PbA associations reported in the past were based on ambient PbA measured as total suspended particles (TSP). During this time period, monitors capturing PbA in particulate matter (PM) with a 10 μm cutoff point (PM10) and PM with a 2.5 μm cutoff point (PM2.5) were also measured in various networks. Additionally, the number of TSP monitors has decreased (946 sites in 1981 versus about 270 in 2011). Moreover, there are known design flaws of the TSP sampler, in which bias related to wind speed and wind direction increase with particle size.12 PM10 and PM2.5 monitors have no directional bias and therefore may provide more accurate concentration measurements than TSP samplers. The PM10 and PM2.5 measurements are also more relevant to particle deposition in the human respiratory system. The relationship between PbB and PbA is still largely unknown for PbA in different PM size fractions. Therefore, the PbB−PbA association needs to be characterized for PbA measurements in different PM size cuts. In this study, we compare associations between PbB and PbA by using PbA measurements from TSP, PM10, and PM2.5 monitors. The NHANES PbB data obtained for 1999−2008 were merged with PbA data from the U.S. EPA Air Quality System (AQS) to explore associations. The time period examined occurred after Pb was completely phased out as an antiknock agent in gasoline and ambient PbA concentrations declined. This paper reflects a continuing effort to characterize the relationship between PbB and PbA13,14 and focuses on how the PbB−PbA associations vary with different PM size fractions in the PbA measurements.
Article
METHODS
Blood Lead and Air Lead Data. Individual PbB concentrations were obtained from NHANES for the years 1999 through 2008. This time period followed the complete phase-out of Pb as a gasoline additive in the United States. The age groups used in this work were consistent with the NHANES sample design, that is, 1−5, 6−11, 12−19, 20−59, and ≥60 years, to account for differences in behavior and Pb storage, distribution, and metabolism among young children, older children, adolescents, adults, and older adults. PbB data were not collected for children less than 1 year. PbA concentrations were obtained from the AQS as Pb in TSP, PM10, and PM2.5. Measured PbB and PbA concentrations were linked by the staff at the Research Data Center (RDC) of the Centers for Disease Control and Prevention (CDC), based on the locations of NHANES subjects and air monitors and the sampling dates of PbA and PbB. PbA monitors were linked with the NHANES subjects’ residential home locations on a Census block group level. The 2000 Census block group ID numbers were assigned to each NHANES subject. PbA monitors within 4 km of a subject’s Census block group centroid were used for subsequent analyses. If multiple PbA monitors existed within 4 km of a subject’s Census block group centroid, the sizestratified PbA concentrations across monitors were averaged. A pseudocode identifier for Census block group ID was created by RDC staff to distinguish Census block groups for the purpose of accounting for commonalities in climate and exposure patterns among residents. The actual 2000 Census block group ID numbers for each NHANES subject are unknown to the authors, because the RDC maintains strict nondisclosure policies. The University of North Carolina Institutional Review Board granted an exemption to this study. Various PbA concentration metrics were used in this work to reflect PbA exposure. Since PbB may reflect relatively recent Pb exposure, PbA metrics used in this work include concurrent PbA concentrations and the average PbA concentration over the 90 days prior to the NHANES subjects’ PbB samples. PbA is usually measured every 3 or 6 days, and the actual dates at which samples are obtained vary with monitoring site. Therefore, the concurrent PbA concentration was defined as PbA measured within 6 days of the PbB measurement. Belowdetection values and missing values were imputed for both PbB and PbA before the data sets were merged. The detailed imputation methods were described by Richmond-Bryant et al.13 From 1999 to 2008, a total of 456 TSP samplers, 155 PM10 samplers, and 548 PM2.5 samplers reported PbA concentrations to the AQS, resulting in totals of 108 126, 16 222, and 238 053 24-h average PbA concentrations reported to the AQS for TSP, PM10, and PM2.5 respectively. These measurements covered 3414, 1332, and 2554 calendar days for PbA reported as TSP, PM10, and PM2.5, respectively. For PbA in TSP, each sampler collected an average of 150 samples, with an average of 6 days between two consecutive measurements. On average, 60 samples were collected by each PM10 sampler, with an average of 6 days between two consecutive samples. For PM2.5, an average of 390 samples was collected by each monitor, and the average interval was 4 days between two consecutive measurements. A total of 25%, 34%, and 61% of the reported PbA concentrations were below the limit of detection (LOD) for TSP, PM10, and PM2.5, respectively. 1264
dx.doi.org/10.1021/es4039825 | Environ. Sci. Technol. 2014, 48, 1263−1270
Environmental Science & Technology
Article
Table 1. PbB and PbA Concentrations in the Analyses, Stratified by Particle Size and Age Groupa percentile distributionb b
age group (yrs)
type
N
geo mean
1−5 1−5 6−11 6−11 12−19 12−19 20−59 20−59 ≥60 ≥60
PbB PbA PbB PbA PbB PbA PbB PbA PbB PbA
178 178 212 212 414 414 508 508 258 258
2.3 13.5 1.7 15.1 1.2 17.1 1.5 14.9 2.4 14.6
1.9 2.7 1.9 2.9 1.9 3.1 1.9 3.2 1.9 2.7
1−5 1−5 6−11 6−11 12−19 12−19 20−59 20−59 ≥60 ≥60
PbB PbA PbB PbA PbB PbA PbB PbA PbB PbA
2150 2150 2261 2261 4787 4787 7842 7842 3510 3510
2.0 5.4 1.4 5.1 1.0 5.5 1.4 5.0 2.2 5.1
2.0 2.7 1.9 2.6 1.9 2.6 2.0 2.6 1.8 2.6
1−5 1−5 6−11 6−11 12−19 12−19 20−59 20−59 ≥60 ≥60
PbB PbA PbB PbA PbB PbA PbB PbA PbB PbA
193 193 200 200 392 392 614 614 245 245
2.4 3.0 1.7 2.5 1.1 2.8 1.6 2.8 2.3 3.3
1.9 2.8 1.9 3.0 2.0 2.5 2.0 2.6 1.9 2.3
geo SD
5th TSP 0.8 3.0 0.6 3.3 0.5 3.0 0.6 2.4 1.0 2.3 PM10 0.8 1.0 0.6 1.0 0.4 1.0 0.5 1.0 0.9 1.0 PM2.5 0.9 0.6 0.6 0.3 0.4 0.7 0.6 0.6 0.9 0.8
10th
25th
50th
75th
90th
95th
1.1 3.3 0.7 3.8 0.5 3.9 0.7 3.3 1.1 3.8
1.5 8.0 1.1 9.1 0.7 9.3 0.9 8.0 1.5 8.7
2.4 10.0 1.7 10.0 1.1 12.5 1.5 10.0 2.3 10.0
3.5 30.0 2.6 30.0 1.7 40.0 2.3 30.0 3.3 30.0
5.3 50.0 3.7 40.0 2.7 70.0 3.6 70.0 5.3 70.0
6.6 90.0 4.9 100.0 3.4 100.0 4.8 100.0 7.3 100.0
0.9 1.6 0.7 1.6 0.5 1.6 0.6 1.4 1.1 1.5
1.3 2.8 0.9 2.8 0.7 3.0 0.9 2.7 1.5 3.0
1.9 5.6 1.3 5.0 1.0 6.0 1.4 5.0 2.1 5.0
2.9 10.0 2.1 10.0 1.5 10.0 2.2 10.1 3.2 10.1
4.7 20.0 3.4 18.3 2.3 20.0 3.5 18.3 4.7 18.3
6.8 27.0 4.5 21.6 3.2 25.8 4.7 22.4 6.1 22.4
1.1 0.7 0.7 0.7 0.5 0.9 0.7 0.8 1.0 0.9
1.5 1.5 1.1 1.3 0.7 1.5 0.9 1.5 1.5 1.8
2.4 3.5 1.7 3.2 1.0 3.3 1.6 3.0 2.2 4.1
3.5 5.7 2.5 5.6 1.6 4.7 2.6 5.0 3.4 6.1
5.3 8.2 3.8 9.5 2.7 8.2 3.9 8.2 5.3 10.1
6.8 19.0 5.2 10.0 3.5 10.0 5.4 10.1 6.7 10.1
TSP, total suspended particles; PM10, particulate matter with a 10 μm cutoff point; PM2.5, particulate matter with a 2.5 μm cutoff point. bPbA (air lead) is in units of nanograms per cubic meter; PbB (blood lead) is in units of micrograms per deciliter. a
Statistical Analysis. A multilevel linear mixed effect (LME) model was developed to obtain estimates of the relationship between PbB and ambient PbA concentrations; see eq 1. Multilevel modeling was performed at the individual and Census block group levels. ln(PbBi , j) = β0 + bj + βPbA ln[t PbA j] + εi , j
intended to account for oversampling, but the weighting process often results in weights that are not inverse probabilities and so are not valid corrections. Therefore, sampling weights calculated for NHANES to adjust for oversampling segments of the population (e.g., for race/ ethnicity, age group, or urbanization) were not applied in the LME analysis. For this reason, the association between PbB and PbA was also examined with a statistical model including individual level covariates, such as demographic information and socioeconomic status (Supporting Information). Richmond-Bryant et al.13 compared models with and without covariates and found that most model pairs were not statistically significantly different from each other. Student t tests were employed to compare βPbA among PM size fractions, age groups, and averaging times by assuming our strata to be independent. This assumption is reasonable because most monitors were not co-located (so that the size distribution data would not be correlated), the likelihood of many subjects from the same age group and monitor location was low, and/or there would not be many subjects from two age groups compared for the same monitor. The acceptable probability for falsely rejecting the null hypothesis (type I error) is set at α = 0.05.
(1)
where PbBi,j is the PbB for the ith individual living in the jth Census block group, [tPbAj] is the average ambient PbA concentration obtained at the jth Census block group or over an averaging period of length t, and βPbA is the slope of ln(PbB) on ln(PbA) over an averaging period of length t (and ln denotes the natural logarithm). The ambient PbA concentration is treated as fixed effect. β0 is the overall intercept, bj is a Census block group-level random normal intercept with mean zero and variance τ2, and εi,j is a random normal variable with mean zero and variance σ2. The statistical model was fitted separately by particle size (TSP, PM10, and PM2.5), PbA temporal treatment (concurrent and 90-day average), and age group. Sampling weights were not applied in this analysis, because they may create biased estimates.15 Gelman16 also points out that poststratification application of sampling weights is 1265
dx.doi.org/10.1021/es4039825 | Environ. Sci. Technol. 2014, 48, 1263−1270
Environmental Science & Technology
Article
Table 2. Association between PbB and PbA,a Stratified by Particle Size, Age Group, and PbA Averaging Timeb current PbA concn
avg PbA concn in past 90 days
age group (yrs)
N
slope
CLLc
CLUd
1−5 6−11 12−19 20−59 ≥60
178 212 414 508 258
0.056 0.028 0.091 0.067 0.012
−0.06 −0.07 0.02 0.01 −0.07
0.17 0.13 0.16 0.13 0.10
1−5 6−11 12−19 20−59 ≥60
2150 2261 4787 7842 3510
0.104 0.073 0.064 0.025 0.069
0.07 0.04 0.04 0.01 0.05
0.14 0.10 0.09 0.04 0.09
1−5 6−11 12−19 20−59 ≥60
193 200 392 614 245
0.039 −0.009 0.026 −0.034 0.059
−0.05 −0.10 −0.05 −0.10 −0.04
0.13 0.08 0.10 0.03 0.16
P-value TSP 0.336 0.572 0.016 0.027 0.776 PM10
