Objectives BP contractors collected nearly 25 000 personal passive dosimeter samples (about 150 000 individual exposure measurements, primarily benzene, ethylbenzene, toluene, xylene and total hydrocarbon (THC)), related to the response and cleanup of the Deepwater Horizon oil spill. BP used a sampling strategy based on compliance with applicable occupational exposure limits (OELs). Most of the measurements were below the reported limit of detection (censored). This occurred because the analytic laboratories calibrated their instruments relative to the chemicals’ OELs and they reported measurements below the lowest calibration standard (approximately 5% of the OEL) as less than the LOD. In an epidemiology study, however, all exposure levels are of interest rather than only those levels related to an OEL. Published evaluation studies on the analytical methods indicate that the methods were capable of measuring much lower concentrations than those reported. This presentation discusses the process used to recalculate the measurement data to the analytic method’s LOD.
Method Gas chromatograph output, the dosimeters’ and chemicals’ physical property data, and the slope and intercept of calibration curves were used to calculate concentrations below reported LODs.
Results The recalculation effort resulted in the reduction of all censored measurements from 92.8% to 60.2% and the THC censored measurements from 71.9% to 19.1%.
Conclusions The recalculation resulted in a substantial reduction in the number of measurements below the LOD. The degree of censoring after recalculation is well within the operating range of the statistical methods used in the GuLF STUDY to estimate exposure levels.