How well does a single blood sample represent long-term exposure for epidemiological studies of PFOA among men in the general population?

Abstract

Many epidemiological studies use a single blood sample per participant to assess exposure, but it is unclear how well a single sample represents longer term exposure. We performed a simulation study using summary statistics for repeated serum PFOA measurements from several previous studies in men to generate plausible serum concentrations over time, taking within-subject correlations into account. Simulated serum concentrations for controls were categorized into quintiles at each time point, and used to determine the extent of misclassification at each time point compared to the “true” long-term average exposure. We then generated case counts by quintile needed to produce an odds ratio (OR) of 1.5 for the highest vs. lowest quintile categorized based on long term exposure, and used the same misclassification rates observed in the controls to simulate misclassified exposure quintiles for cases. Comparing long term vs. single baseline exposure measures for repeated serum samples collected within about 5–13 years of each other revealed similar effect estimates, although there was a small bias to the null. Trend tests across quintiles were mostly significant using either baseline or long-term exposure. For the general population sample of men in Norway, with 5 repeated measurements over 28 years, serum PFOA was substantially lower prior to 1987, and using either of the two earliest samples as the exposure metric, compared to the long term average, produced larger bias to the null and non-significant trend tests; however using later time points as the exposure metric resulted in only a small bias. Using data based on studies of men, single baseline serum samples represented rather well the mean of repeated samples collected up to 13 years apart, but were not always reliable surrogates for average exposure over 3 decades, during which time PFOA exposure levels in the general population have changed substantially.