^{a}Quartiles were defined by the estimated cumulative PFOA serum concentration among the thyroid, kidney, or testicular cancer cases at the time of cancer diagnosis. ^{b}A proportional hazards regression model was run for each cancer; each model was adjusted for time-varying smoking, time-varying alcohol consumption, sex, education, and stratified by 5-year period of birth year. Time began at age 20 years if the person’s 20th birthday was in 1952 or later; otherwise time began at the age the person was in 1952; time ended at the age of cancer diagnosis, age at the last follow-up survey, or age on December 31st 2011, whichever came first. ^{c}*p*-Value is for linear trend test in the log rate ratios across quartiles; *p*-Values were calculated using exposure category midpoints and inverse variance weighting in a no-intercept linear regression model. ^{d}*p*-Value is from the continuous log estimated cumulative PFOA serum concentration models. |