, 2006) up to MG-132 in vivo 271% in P’an’s most highly-exposed subset (1981). Although the very high values obtained by P’an may be explained by

the rapid increase in saliva lead levels at blood lead levels >50 μg/L (Koh et al., 2003), there is still a great deal of unexplained variation between studies in the literature. This is most likely due to the lack of a standardised sample collection or preparation method for the analysis of lead in saliva; the wide variety of different procedures employed. The method presented in this study, using a new sampling device and a nitric acid digestion step to release protein-bound lead in the matrix, obtained a mean blank-corrected recovery from 10 μg/L spiked saliva of 65.9% (SD: 1.83 μg/L, n = 13). This demonstrates an improvement on the recovery of 30–35% reported by Morton et al. (2014), where a comparable ICP-MS method was used, but with Proteasome structure a different sampling device and no acid-digestion step. This may account for the higher levels of saliva lead observed by

this study than Morton et al. (2014) (median: 17.1 μg/L and 7.3 μg/L, respectively), despite the sample cohort reported in this paper showing lower blood lead levels (median: 8.34 μg/dL and 20 μg/dL, respectively). However, the StatSure device did exhibit a drawback – a significant level of lead contamination was shown to emanate from the device, with a mean result from blank saliva of 2.86 μg/L (SD: 1.13 μg/L, n = 10) using the device, compared to 0.38 μg/L (SD: 0.36 μg/L, n = 10) by direct analysis, i.e. the device contributed 2.48 μg/L of lead to the saliva result. The results from blank saliva aliquots sampled using the device also showed a higher degree of variation than those analysed directly. An investigation of the lead concentration of the device components showed that this contamination originated in the sampling paddle. For this study of occupationally-exposed lead

workers, the median saliva lead was 17.1 μg/L, and therefore the effect of this contamination would be relatively small. However, this would be of concern for the measurement of lower-level environmental exposures. The manufacturers of the device Thymidine kinase have been made aware of the authors’ findings and will endeavour to ensure that this contamination is not present in future batches. Additional analyses will be necessary to confirm this. A weak but significant correlation (r = 0.457) was observed between the log(saliva lead) and log(blood lead) results from the 105 paired samples analysed. This is a stronger correlation than that observed between the same variables by Barbosa et al. (2006) (r = 0.277) or by Nriagu et al. (2006) (r = 0.156), and slightly stronger than the correlation observed by Koh et al. (2003) between log(saliva lead) and blood lead results (r = 0.41). A further study by Thaweboon et al. (2005) reported a poor correlation between saliva lead and blood lead.