Making sense of PFAS "forever chemicals" data when most water tests come back "not detected"

What is it about?

When scientists test rivers for PFAS, a family of persistent synthetic chemicals often called "forever chemicals," most laboratory results come back below the detection limit. This means the chemical may be present but at levels too low to measure with confidence. That creates a practical problem: how do you compare contamination across different sites when most of your data points are effectively question marks? We built and tested a transparent, end-to-end workflow to handle this challenge. Using water collected from 10 sites along tributaries of the Trinity River near Fort Worth, Texas, we showed how statistical imputation, which estimates the likely values of below-detection results instead of simply replacing them with zero or an arbitrary number, changes site totals and rankings, with the largest effect at lower-concentration sites. We then stress-tested the results under five different assumptions about those non-detected values. The rankings held up: the same site (TW08, located on the West Fork Trinity River) consistently ranked highest across all methods, with elevated PFOS, PFHxS, and 6:2 FTS in its chemical mixture. Everything needed to reproduce the study, including the data, Python code, and figures, is freely available online, enabling other researchers, water managers, and regulators to apply the same workflow to their own PFAS datasets.

Featured Image

Photo by Terry Vlisidis on Unsplash

Photo by Terry Vlisidis on Unsplash

Why is it important?

As EPA Method 1633 becomes the standard for PFAS water testing across the United States, environmental agencies and water utilities will increasingly work with datasets where most results fall below reporting limits (nearly 80% in this study). Handling those non-detects poorly can distort site comparisons and misdirect limited monitoring and cleanup resources. This paper provides a ready-to-use, fully reproducible workflow, with open code and open data, that any monitoring program can adopt to generate defensible, transparent site rankings from heavily censored PFAS data. It is especially relevant for Texas, where PFAS-specific surface-water monitoring is not yet routinely required in discharge permits, and for any agency beginning reconnaissance-level PFAS surveys under emerging federal and state regulations.

Perspectives