We evaluated predictors of plasma concentrations of dichlorodiphenyldichlor
oethylene (DDE), a metabolite of dichlorodiphenyltrichloroethane (DDT), and
polychlorinated biphenyls (PCBs) in a group of 240 women, controls from a
breast cancer case-control study nested in the Nurses Health Study. We cons
idered personal attributes such as age, serum cholesterol, region of reside
nce, adiposity, lactation, and dietary intake. DDE levels increased 0.17 pp
b/year of age (p = 0.0003), and PCBs increased 0.08 ppb (p = 0.0001). DDE a
nd PCBs increased 0.20 (p = 0.02) and 0.13 ppb (p = 0.001), respectively, p
er 10 mg/dl serum cholesterol. Women living in the western United States ha
d higher levels of DDE (mean = 11.0 ppb; p = 0.003), and women in the North
east and Midwest had higher levels of PCBs (mean = 5.6 ppb; p = 0.0002) as
compared to women from other parts of the country (mean DDE = 6.3; mean PCB
s = 4.5 ppb). Levels of DDE could not be predicted from consumption of meat
, fish, poultry, dairy products, vegetables, fruits, and grains. There was
a positive association between fish consumption and PCB concentrations amon
g women in the Northeast and Midwest. Using data from the cases in the nest
ed case-control study to assess the predictive ability of the models, we co
nfirmed that the most reliable predictors of DDE were age and serum cholest
erol, and the most important predictors of PCBs were age, serum cholesterol
, and residence in the Midwest or Northeast. The null results for the major
ity of the food variables suggest that specific dietary factors, other than
fish, are not currently a substantial contributor to human exposure to DDE
and PCBs.