A Study of Reverse Causation: Examining the Associations of Perfluorooctanoic Acid Serum Levels with Two Outcomes

Environ Health Perspect. 2017 Mar;125(3):416-421. doi: 10.1289/EHP273. Epub 2016 Aug 16.


Background: Impaired kidney function and earlier menopause were associated with perfluorooctanoic acid (PFOA) serum levels in previous cross-sectional studies. Reverse causation, whereby health outcomes increase serum PFOA, may underlie these associations.

Objective: We compared measured (subject to reverse causation) versus modeled (unaffected by reverse causation) serum PFOA in association with these outcomes to examine the possible role of reverse causation in these associations.

Methods: In cross-sectional analyses, we analyzed PFOA in relation to self-reported menopause among women (n = 9,192) 30-65 years old and in relation to kidney function among adults > 20 years old (n = 29,499) in a highly exposed Mid-Ohio Valley cohort. Estimated glomerular filtration rate (eGFR, a marker of kidney function) and serum PFOA concentration were measured in blood samples collected during 2005-2006. Retrospective year-specific serum PFOA estimates were modeled independently of measured PFOA based on residential history and plant emissions. Using measured and modeled PFOA in 2005 or 2006 (predictor variables), cross-sectional associations were assessed for eGFR and menopause (yes/no). We also analyzed measured PFOA (dependent variable) in relation to the number of years since menopause.

Results: Menopause and eGFR were significantly associated with measured (trend tests: p = 0.013, p = 0.0005, respectively) but not with modeled serum PFOA (p = 0.50, p = 0.76, respectively). Measured PFOA levels increased for the first 7 years after menopause (trend test, p < 0.0001), providing further evidence that the observed association between measured PFOA and menopause is subject to reverse causation for this outcome.

Conclusion: Our results support the conjecture that in previous studies, earlier menopause and reduced kidney function are the causes rather than the results of increased measured serum PFOA. These results suggest caution in using biomarkers in cross-sectional studies. Citation: Dhingra R, Winquist A, Darrow LA, Klein M, Steenland K. 2017. A study of reverse causation: examining the associations of perfluorooctanoic acid serum levels with two outcomes. Environ Health Perspect 125:416-421; http://dx.doi.org/10.1289/EHP273.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Biomarkers / blood
  • Caprylates / blood*
  • Environmental Exposure / statistics & numerical data*
  • Female
  • Fluorocarbons / blood*
  • Humans
  • Menopause / blood
  • Middle Aged
  • Water Pollutants, Chemical / blood*


  • Biomarkers
  • Caprylates
  • Fluorocarbons
  • Water Pollutants, Chemical
  • perfluorooctanoic acid