09.02.2014 Views

home edit2 whole TSD November 2002 PDF format - OEHHA

home edit2 whole TSD November 2002 PDF format - OEHHA

home edit2 whole TSD November 2002 PDF format - OEHHA

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

A risk assessment was also conducted using an adjustment for the strong interaction between arsenic<br />

and smoking observed in several occupational cohorts. The prevalence of smoking was independent of<br />

the level of arsenic exposure in the Anaconda cohort (Welch et al., 1982), but may have been higher<br />

than in the general population. Also, there appeared to be no reason for smokers to be distributed<br />

differently among the exposure levels in the Tacoma cohort, hence smoking was assumed to be<br />

independent of arsenic exposure in this cohort as well.<br />

Each dose-specific crude SMR was adjusted taking the low-dose SMR in each study as the baseline.<br />

Next, a nonsmokers’ SMR and a smokers’ SMR were derived. From the nonsmokers’ SMR,<br />

observed and expected deaths among nonsmokers were inferred. Finally, a regression model was fitted<br />

to the inferred nonsmokers’ data to find the slope of the line relating cumulative arsenic dose to excess<br />

relative risk. This procedure was applied to the data of Enterline et al. (1987a) under the assumption<br />

that the interaction between smoking and arsenic varies as a function of dose, and that the joint effects at<br />

low doses are multiplicative.<br />

The MLE for β was 2.30 × 10 -4 using the data on nonsmokers from the study by Enterline et al.<br />

(1987a). An 95% UCL was estimated and used in evaluating unit risks.<br />

Risks were evaluated separately by sex and for four smoking categories: never, former, light (< 1<br />

pack/day) and heavy smokers. Unit risks for these categories range from 400 to 8,400 per million<br />

persons, with upper bounds ranging from 630 to 13,000 per million.<br />

The staff of DHS recommended that the range of risk for ambient exposures to arsenic be based on the<br />

95% UCL predicted from fitting a linear model to the human data adjusted for interaction with smoking.<br />

The staff of DHS further recommended that the overall unit risk, 3.3 × 10 -3 per µg/m 3 , be considered<br />

the best estimate of the upper bound of risk.<br />

Oral<br />

A generalized multistage procedure with both linear and quadratic dose assumptions was used to<br />

predict the prevalence of skin cancer as a function of arsenic concentration in drinking water (d) and age<br />

(t), assuming exposure to a constant dose rate since birth. F(t,d) represents the probability of<br />

developing skin cancer by age t after lifetime exposure to arsenic concentration d. The procedure used<br />

is expressed as follows: F(t,d) = 1 - exp[-g(d) H(t)], where g(d) is a polynomial in dose with nonnegative<br />

coefficients, and H(t) is (t-w) k , where k is any positive real number, and t > w for induction<br />

time w. The cancer potency calculation was based on skin cancer incidence data for Taiwanese males<br />

(Tseng et al., 1968) because their skin cancer prevalence rates were higher than the females studied.<br />

The calculation was also based on several assumptions listed below.<br />

1. The mortality rate was equal for both diseased (skin cancer) and nondiseased persons.<br />

2. The population composition (with respect to skin cancer risk factors) remained constant over time,<br />

implying that there was no cohort effect.<br />

68

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!