Medications as a potential source of exposure to parabens in the U.S. population.

INTRODUCTION: Use of paraben-containing medications has been shown to be associated with urinary paraben concentrations among couples undergoing fertility treatment, but it is unknown whether this association is also present among the general population.

METHODS: A list of prescription medications of interest was developed based on their likelihood of containing parabens and the ability to identify users in the National Health and Nutrition Examination Survey (NHANES); alendronate, escitalopram oxalate, fluoxetine, and olanzapine were chosen. Participants reported whether they had used each medication in the past month. Linear regression models were used to compare model-based mean urinary concentrations of each paraben among users and non-users of these four medications.

RESULTS: A total of 10,302 respondents were included in the analysis, 265 (2.6%) of whom had reported using a paraben-containing prescription medication in the previous month. Users of alendronate had mean concentrations of ethyl paraben that were approximately three-fold higher than non-users (p ≥ 0.001 in unadjusted and adjusted models), which was likely due to three participants with very high concentrations. No other differences in paraben concentrations were found for any of the medications of interest (all p ≥ 0.13). Compared to non-users, a significantly greater proportion of alendronate users had butyl and ethyl paraben concentrations above the 95th percentile (17.8% and 12.3%, respectively) compared to non-users (5.0% and 5.0%, respectively; both p ≤ 0.01), despite ethyl paraben not being an expected ingredient in the brand name formulation of alendronate.

CONCLUSION: Despite previous work showing that medications can be an important source of paraben exposure, there was no clear overall evidence of associations between the use of paraben-containing medications and increases in urinary paraben concentrations among participants in NHANES 2005-2012. These results highlight the difficulties inherent in proper assessment of exposures with short half-lives based on a single cross-sectional biologic sample.