The East Bay Diesel Exposure Project: a biomonitoring study of parents and their children in heavily impacted communities

Diesel exhaust (DE) exposures vary widely across California, with low-income communities and communities of color often experiencing disproportionately higher exposures. DE is associated with adverse health effects, including asthma, cancer, and cardiovascular disease, and these harms can be exacerbated by co-occurring environmental and social stressors. Despite statewide emission reductions, highly impacted neighborhoods near traffic, rail, and maritime sources (e.g., West Oakland) can experience elevated air pollution. The California Environmental Contaminant Biomonitoring Program prioritized DE in 2009, and, after review of evidence on exposure, toxicity, and measurement feasibility, identified urinary metabolites of 1-nitropyrene (1-NP)—particularly 6-OHNP and 8-OHNP—as viable biomarkers of DE exposure. Prior work has shown positive associations between urinary 1-NP metabolites and airborne 1-NP among border commuters and higher metabolite levels in urban versus agricultural settings. Building on this literature, the East Bay Diesel Exposure Project (EBDEP), launched in 2017, is the first study to evaluate DE exposures in families with young children, pairing urinary biomonitoring with measurements of 1-NP in household dust and indoor air to characterize exposures and variability and to support mitigation efforts in heavily burdened communities.

Prior studies have identified 1-nitropyrene (1-NP) as a diesel-specific nitro-PAH and validated its urinary metabolites (6-OHNP, 8-OHNP) as biomarkers of DE exposure. Positive correlations between airborne 1-NP and urinary metabolites were observed among frequent border crossers at San Ysidro, CA. A pilot study suggested higher urinary metabolites in urban (Oakland) children compared with an agricultural cohort (Salinas Valley). Occupational studies (e.g., taxi drivers in China, traffic workers in Peru) reported metabolite levels substantially higher than in general populations. Ambient and neighborhood-scale monitoring indicates spatial heterogeneity and seasonality in nitro-PAHs and black carbon, with elevated concentrations near traffic corridors and during fall/winter inversions.

Design and setting: Community-based biomonitoring and environmental sampling among 40 parent–child pairs (n=80) residing in Oakland, Richmond, and nearby San Francisco East Bay communities (Jan 2018–Feb 2019). Recruitment targeted areas with a range of DE potential using CalEnviroScreen 3.0 diesel PM indicator; higher exposure neighborhoods near the Port of Oakland and interstate highways were prioritized, with some lower-exposure areas included for comparison. Eligibility: English-speaking families with toilet-trained children aged 2–10 years intending to remain at their residence for ≥4 months. Two sampling rounds were conducted per family, separated by a mean of ~4 months (range 0.5–8 months). IRB approvals were obtained; informed consent (and child assent for ages 7–10) included options for results return and sample donation. Data collection procedures: On day 1 of round 1, staff enrolled participants, measured anthropometrics, administered an exposure questionnaire, provided GPS loggers for parent and child, instructed on daily time-activity diaries, conducted a home walk-through, deployed indoor air monitoring equipment, and collected a vacuum bag/bagless dust sample when available. On day 4, a follow-up questionnaire was completed, urine samples were collected from both participants, floor sweepings were collected if no vacuum was available, and the air sampler was retrieved. Round 2 repeated these procedures except dust was not resampled. Questionnaires and home walk-through: Collected demographics and potential determinants of 1-NP exposure: occupational diesel use; work/school/childcare locations and times; tobacco smoke exposures; combustion sources and appliances at home; vehicle type and attached garage; stove fan use; housing characteristics; heating system; ventilation and portable air cleaners. Urine sampling and laboratory analysis: Two sampling schemes were used: (a) 25 families provided one or more spot urine samples (often first morning void) on day 4 of each round; (b) 15 families collected daily spot samples for four consecutive days in each round. Samples were refrigerated, retrieved on day 4, and transported on cold packs to the California Department of Public Health Environmental Health Laboratory (CDPH EHL) for aliquoting, specific gravity (SG) via refractometer, and creatinine by Jaffe reaction; aliquots were stored at −80°C. Aliquots (target 100 mL; less if low volume) and field blanks were shipped on dry ice to the University of Washington (UW) laboratory, stored at −20°C until analysis. Urinary 6-OHNP and 8-OHNP were quantified by HPLC–MS/MS after enzymatic hydrolysis, solid phase extraction with blue rayon, and alumina cleanup. Of 335 urine samples and 19 field blanks analyzed, some measurements were excluded due to chromatographic interferences or poor internal standard recovery, yielding valid measurements of 6-OHNP in 155 (children) and 138 (parents) and 8-OHNP in 168 (children) and 150 (parents); paired valid measurements numbered 138 (6-OHNP) and 164 (8-OHNP). QC included field blanks, internal standard–spiked lab blanks, benchmark urine, and analyte-spiked benchmark urine; LODs (based on 100 mL) were 15.5 pg/L (6-OHNP) and 21.2 pg/L (8-OHNP); average spike recoveries were 126% (6-OHNP) and 142% (8-OHNP); field blanks were <LOD. Concentrations were SG-adjusted using a reference SG of 1.017 (NHANES 2007–2008 adult median). Dust sampling and analysis: In round 1, vacuum bag/bagless dust (n=30) or floor sweepings if no vacuum (n=10) were collected, stored at −20°C, and shipped to UW. Methods were developed for dust 1-NP: samples were sieved to 150 µm; 200 mg of sieved dust was spiked with deuterated 1-NP internal standard, extracted with methylene chloride by sonication (60 min), centrifuged, concentrated, cleaned via silica gel SPE, eluted (methylene chloride:pentane 35:65), evaporated, reconstituted in 75:25 ethanol:20 mM sodium acetate pH 5.5, filtered (0.2 µm PTFE), and analyzed by two-dimensional HPLC–MS/MS. QC used lab blanks and spiked benchmark dust; spike recovery mean 105% (CV 8%); replicate CV 8%; LOD set to the lowest calibration standard (15 fg/mg); interferences/insufficient mass led to 36 valid measurements from 33 homes. Indoor air sampling and analysis: An Aerosol Black Carbon Detector (ABCD) was deployed in homes for each 4-day period per round, logging black carbon at 5-second intervals while drawing ~111 cc/min (mean volume 0.49 m³ per deployment). Filters were shipped to UW for 1-NP analysis via two-dimensional HPLC–MS/MS after deuterated internal standard spiking, methylene chloride sonication, evaporation, reconstitution, filtration. System stability was monitored with a mid-level calibration standard; QC included clean blank filters and spiked filters (average recovery 100%, CV 3%); LOD defined as mean lab blank + 1 SD (0.164 pg/filter). Of 80 possible filters, 74 valid measurements were obtained; 6 were missing/invalid due to single-round participation or QC issues. Air concentrations were reported in pg/m³ based on sampled volume. Statistical analysis: Conducted in R 3.6.1 and SAS 9.4. Concentrations <LOD were imputed as LOD/√2. Weighted Pearson correlations (weights = inverse of number of repeats per individual/family) were computed among log-transformed urinary metabolites (parent–child pairs) and between metabolites and environmental 1-NP (air, dust). Random-effects models estimated geometric means (GMs) and 95% CIs for metabolites (parents vs children) and tested paired differences. Linear mixed-effects models with subject random intercepts evaluated bivariate associations between urinary metabolites and race/ethnicity, family income, and season (fall/winter vs spring/summer). For short-term variability (n=15 families with daily samples; 215 samples), intraclass correlation coefficients (ICCs) were computed from mixed models including fixed effects for sampling round. Associations with CalEnviroScreen 3.0 diesel PM were tested by comparing census tracts ≥90th percentile vs <90th percentile: t-tests for dust (log-transformed), mixed models for indoor air and urinary metabolites, with and without seasonal covariates.

- Detection and concentrations: At least one urinary 1-NP metabolite was detected in 96.6% of urine samples; 1-NP was detected in 97% of dust samples and 74% of indoor air samples. - Urinary metabolite levels: Parent 6-OHNP GM (95% CI): 240 pg/L (180, 310); parent 8-OHNP GM: 150 pg/L (120, 190). Child 6-OHNP GM: 150 pg/L (110, 200); child 8-OHNP GM: 130 pg/L (100, 170). Maximums: parent 6-OHNP 7800 pg/L, parent 8-OHNP 5800 pg/L; child 6-OHNP 4000 pg/L, child 8-OHNP 3200 pg/L. - Parent–child comparison: 6-OHNP levels were significantly higher in parents than children (random-effects model, p=0.005); 8-OHNP lower in children vs parents but not significantly. Within individuals, 6-OHNP and 8-OHNP were strongly correlated (r=0.84, p<0.0001). - Parent–child correlations: 8-OHNP weakly correlated between parents and children (weighted r=0.28, p<0.001); 6-OHNP not significantly correlated (r=0.10, p=0.24). Correlations were stronger in families with stay-at-home/working-from-home parents (8-OHNP r=0.41, p<0.001; 6-OHNP r=0.24, p=0.04) than in families where parents worked away from home (8-OHNP r=0.20, p=0.064; 6-OHNP r=−0.03, p=0.83). - Seasonality: Urinary metabolites were higher in fall/winter than spring/summer. Mixed models indicated significant seasonal differences for 8-OHNP (children p=0.017; parents p<0.01) and borderline for 6-OHNP (children p=0.070; parents p=0.053). - Short-term variability: ICCs indicated low to moderate reproducibility over 4 consecutive days within rounds: parents ICC=0.39 (6-OHNP) and 0.43 (8-OHNP); children ICC=0.41 (6-OHNP) and 0.38 (8-OHNP). Within-subject variance comprised 57–62% of total variance; between-subject 38–43%. - Dust and indoor air: Dust 1-NP GM (95% CI): 380 pg/g (200, 730), DF 97%. One home near multiple freeway intersections had an extreme dust value (910,000 pg/g). Indoor air 1-NP GM: 0.41 pg/m³ (0.36, 0.45), DF 74%; median 0.43 pg/m³; 95th percentile 0.87 pg/m³; max 1.2 pg/m³. Indoor air 1-NP was higher in fall/winter (GM 0.47 pg/m³ [0.41, 0.55]) than spring/summer (0.35 [0.30, 0.41]) (p<0.01). - Air–dust relationships: Log-transformed indoor air and dust 1-NP from round 1 were moderately correlated (Pearson r=0.54, p<0.01 including the extreme dust value; r=0.45, p<0.05 excluding it). - Environmental–biomarker correlations: In round 1, children’s urinary 8-OHNP weakly correlated with 1-NP in dust (weighted r=0.22, p=0.02) and indoor air (weighted r=0.22, p<0.01). No other significant correlations between environmental 1-NP and urinary metabolites were observed. - Socioeconomics and demographics: Parents in the highest income group (>$75,000) had higher urinary metabolites than lower-income groups; significant differences were detected (6-OHNP high vs low income; 8-OHNP high vs middle income) after Tukey adjustment. No associations with race/ethnicity were found. - CalEnviroScreen comparisons: Homes in census tracts with diesel PM scores ≥90th percentile had higher dust 1-NP (t-test p=0.027, extreme dust value excluded) and higher indoor air 1-NP (mixed model p=0.017) than homes below the 90th percentile. No significant differences in urinary metabolites between these tract groups were detected, with or without adjusting for season. - Context: EBDEP parents’ urinary metabolites were approximately twice those measured in the CARE-LA study of LA County adults; levels were lower than in occupationally exposed populations (e.g., Chinese taxi drivers, Peruvian traffic workers).

This study addressed the research question of characterizing diesel exhaust exposure in disproportionately impacted East Bay communities by combining biomonitoring of urinary 1-NP metabolites with measurements of 1-NP in household dust and indoor air among parent–child pairs. The high detection frequencies and elevated geometric means confirm widespread exposure to DE-related nitro-PAHs in both adults and young children. The higher metabolite levels in parents relative to children suggest differential exposure patterns or metabolism, warranting further investigation. Seasonal differences, with higher levels in fall/winter, align with known atmospheric conditions (e.g., inversions) that increase near-source pollutant concentrations, reinforcing the importance of temporal context in exposure assessment. The moderate correlation between indoor air and dust 1-NP, and the observed associations of children’s 8-OHNP with both matrices, indicate that multiple exposure pathways (inhalation, non-dietary ingestion, dermal) may contribute to children’s internal dose. The elevated dust and indoor air 1-NP in census tracts with higher modeled diesel PM burdens support the validity of CalEnviroScreen prioritization for environmental sampling, though the lack of corresponding differences in urinary metabolites suggests that personal mobility and short metabolite half-lives may obscure census-tract-level contrasts in biomonitoring. High within-subject variability and low ICCs over short timescales highlight the need for repeated biomonitoring to characterize typical exposures. Overall, integrating biomonitoring with hyperlocal environmental measurements provides a more nuanced understanding of exposure disparities and can inform targeted mitigation near traffic, rail, and maritime sources.

EBDEP is the first study to assess diesel exhaust exposure in families with children as young as two years by pairing urinary 1-NP metabolites with household dust and indoor air measurements. The study demonstrates widespread exposure, higher parental 6-OHNP compared with children, seasonal elevation in fall/winter, and high short-term temporal variability necessitating repeated sampling. Environmental concentrations of 1-NP in dust and indoor air were higher in areas with high CalEnviroScreen diesel PM scores, and children’s urinary 8-OHNP showed weak associations with indoor air and dust, suggesting multiple exposure pathways. These findings underscore the value of combining biomonitoring with hyperlocal environmental monitoring to characterize exposure disparities and support the design and evaluation of mitigation strategies in disproportionately burdened communities. Future work will investigate determinants of exposures across microenvironments (residence, work, school/childcare), account for meteorological modifiers, and leverage community air monitoring to better link sources to internal dose.

- Sample size was relatively small (40 families), limiting power to detect associations with demographics and to fully assess seasonal contrasts; recruitment and scheduling challenges prevented the originally intended seasonal balance. - Urinary metabolites exhibited high within-subject short-term variability (low ICCs), reflecting short biological half-lives (<24 h), which complicates characterization of longer-term exposure from limited spot samples. - Indoor air detection frequency (74%) may have been reduced by the low flow rate/air volume of the ABCD samplers, potentially limiting sensitivity. - Dust was sampled only in the first round, precluding assessment of temporal changes in dust 1-NP within homes. - Some urinary measurements were excluded due to chromatographic interferences or poor internal standard recovery; two samples were lost due to vial breakage. - The lack of association between urinary metabolites and CalEnviroScreen diesel PM scores may reflect misalignment between tract-level modeled emissions (4 km² grid) and individuals’ time–activity patterns across microenvironments. - Generalizability is constrained to similar urban communities and to the study period; most parents were female and highly educated, which may not reflect all impacted populations.