Serum concentrations at birth and during the first 10 years of life could only be estimated reliably for participants born after 1989 (n = 4,787), because contaminant releases during this time period were the dominant source of exposure, providing the most reliable exposure estimates.
Most earlier studies assigned exposure estimates to groups of individuals residing in the same city or close to the same pollution monitor, thus providing less differentiation.
Given the ƒ K at p k, different exposure estimates (most probable, error minimizing, etc., estimates) can be calculated at each spatiotemporal node of the appropriate mapping grid, depending on the objectives of the study.
This analysis shows that national exposure estimates are most influenced by reported concentrations in imported tuna, swordfish, and shrimp; Pacific pollock; and Atlantic crabs.
The most sustained association with postnatal exposure estimates was a modest statistically significant association of estimated serum PCB-153 levels through age 6 months with greater DSM-IV Hyperactive-Impulsive index and Conners' ADHD index behaviors at age 8 years (quantile regression models at the 75th percentile), but these associations also attenuated over time.
When concentration estimates were rounded to integers, most exposure cells had a value of zero in most months (331,035 of 504,960), indicating low smoke-specific PM2.5.
Another potential limitation is that we averaged global exposure estimates across the three monitoring stations for most pollutants.
Hence, current approaches to estimate individual TRAP exposures (LUR, dispersion model, geostatistical methods) have two consistent limitations: (i) TRAP surrogates are based on non-specific pollutant measures; (ii) modeled estimates predict concentrations outside the home while most exposure occurs indoors.
Four options exist, if there are potential reasons for concern (RQProduct > 1, Figure1): (i) the PEC and/or PNEC estimate of the most risky compound might be refined by providing additional ecotoxicological data and/or exposure estimates.
Nevertheless, error in the exposure estimates is likely independent of disease status and would most likely, although not certainly, bias exposure outcome associations to the null.
