The sensitivity and the specificity of the analytical technique used in human biomonitoring were mentioned repeatedly. The importance of this issue may be illustrated here numerically. Suppose in the numerical example presented in Slide 15, the percent urinary recovery is not 70%, but around 3 to 7%. The chemical in a 24-hour urine volume of 1 to 1.2 liter is typically measured in 5 to 10 ml aliquots. If the error of quantifying the contents in the 10 ml aliquot were 20%, then altogether the true dose could be under- or over-estimated 2- to 3-fold. That is, the calculated dose of Ch-A (Chemical A) in Slide 15, at a percent recovery of 5% supposedly yielding 0.1 mg of Met-A (Metabolite A) in 24 hours, could be reported as 2.3 mg [= (0.08 mg for a 20% quantification error) x (2/1 for difference in molecular mass)/(7% for recovery from a study)], when its true value is supposed to be 4.0 mg [= (0.1 mg) x (2/1 for difference in molecular mass)/(5% for the expected recovery)].

It must be remembered that in some cases, only spot urine samples are collectable or available. Some simulation techniques such as physiologically-based pharmacokinetic (PBPK) modeling mentioned earlier (Lecture 3) can be used to interpret spot urine sample results (see, e.g., Dong et al., 1994, 1996). Even with PBPK modeling, the endogenous urinary creatinine levels are often used or required as a means to gauge the correction to a 24-hour or longer urine output. Yet as stated in Slide 17, the urinary creatinine level varies with time as well as with some physiological factors such as sex, age, stress, and diseases.