The element that displayed the greatest creatinine-corrected variation in relation to the mean (in terms of GCV) was lead. In fact, lead displayed the greatest inter-individual GCV and intra-individual GCV of the 31 elements. Creatinine-corrected boron, cobalt, caesium, copper and selenium displayed the lowest intra-individual GCV, indicating that day-to-day variation of these elements in individuals are low in comparison to the other elements (after adjusting for gender). These elements are considered ‘essential’ elements and it is likely that the smaller variation is as a result of regulation of these elements in the body. When inter-individual variation

was investigated, scandium, selenium and titanium were found to exhibit the lowest inter-individual GCV, indicating Afatinib in vivo Alectinib concentration that creatinine-corrected concentrations of these elements varied least between individuals (after adjusting for gender), of the 31 elements. For those elements where a reduction in variability was seen, creatinine correction may

be beneficial. The effectiveness of creatinine correction was investigated further by fitting a mixed effects model to uncorrected data (on the natural log scale) with ln(creatinine) treated as a fixed effect in the model. For some elements, the coefficients for ln(creatinine) were not found to be significantly different from the value 1 and there was no significant difference in the within-person variability when compared to when using the creatinine-corrected data (Al, As, Ba, Cd, Co, Ga, Ge, Mo, Ni, Pb, Zn). For these elements, this result indicated that the creatinine corrected values were effective in reducing some of the variation in elemental concentrations due to urine dilution. For Be, Br, Cr, Ru, Ta and V, although there was no significant difference in GCVintra between the corrected and uncorrected

data, a significant reduction was seen in the model where ln(creatinine) was treated as a fixed effect with an estimated coefficient. This is analogous to adjusting for creatinine by dividing the elemental concentrations by a power (the estimated coefficient) of creatinine. The statistical analysis showed that this led to significantly lower intra-individual variation for those elements many than both corrected and uncorrected concentrations. The 95th percentiles of 61 elements in urine samples have been reported. Elements for which we have reported 95th percentile values but for which there is no available comparison are Br, Ce, Er, Ga, Gd, Ge, Hf, Ho, Ir, La, Rb, Rh, Ru, Sc, Sr, Ta, Th, Ti and Yb. The mixed effect modelling provides valuable information on the variation of elemental concentrations by accounting for correlations between repeat samples and modelling the intra-individual and inter-individual variability.