On October 23, 2026, Circle of Blue published a report, originally carried by Inside Climate News, on a study suggesting that PFAS contamination in drinking water may be underestimated. The summary lists Alissa de Chassey as the author. The summary is truncated, and it does not describe the study's design, so this piece does not assess the findings. It addresses a narrower question that any utility or laboratory file should answer before repeating an underestimate claim: what population of results could such a claim actually describe?
The starting point is the reporting limit. The summary notes that "When PFAS concentrations in drinking water fall below laboratory reporting limits, they are..." and then breaks off. The principle is standard in trace analysis. A reporting limit is the lowest concentration a laboratory will report as a quantified value under its method and quality controls. A result below that point is recorded as a non-detect, a censored value, or a flagged result. A non-detect means the analyte was not quantified at that limit. It does not mean the concentration was zero.
Trouble starts when non-detects are converted into numbers for averaging or trend work. Substituting zero pulls averages down. Substituting a fraction of the limit or the full limit moves them differently. The size of an apparent underestimate can depend as much on that coding choice as on the samples themselves, which is why the convention used in any comparison needs to be written down.
The sample frame sets the ceiling
Any underestimate claim is bounded by what was sampled. The systems included, the source water types, whether results come from finished water at the plant or from taps in the distribution system, the seasons covered, and the years examined each define the population. A claim can be accurate for that population and still say little about a neighboring system with different source water, different treatment, or a different sampling history.
Before a utility cites such a claim, its own file should show whether its sampling locations fall inside the set the study describes. Where the summary is silent, the original report is the place to find out how samples were selected. Staff should not assume overlap because the systems are in the same region or serve similar populations.
Questions for your own file
A short review can be run against existing records without new sampling:
Which sampling points correspond to the locations the study describes, and were they finished water or distribution samples?
What reporting limit applied to each result, and did that limit change across the sampling period?
How were non-detects coded in your own file, and does that coding match the convention any external comparison uses?
Were results from more than one laboratory pooled, and does each laboratory's accreditation scope cover the analytes in question?
Which analyte list is being compared? A targeted list of named compounds and a sum measurement such as total organic fluorine answer different questions, and a claim built on one should not be read as a statement about the other.
What the claim does not change
A study, by itself, does not change a compliance figure. Compliance is set by the applicable regulatory method, the sampling schedule, and the limits in force, and those are the numbers a regulator will ask for. An underestimate claim is a reason to review the record, not a substitute for it.
Staff should not restate a study estimate as the system's own concentration. Utility correspondence should not describe health implications of the findings without direction from the appropriate health authority. This piece does not address health effects.
Keep the study date, its original source, and the sample set attached to any citation. Once the frame is named, a reviewer can see whether the claim applies to the system at hand and whether it belongs in a compliance file or only in a briefing. The original report is available from Circle of Blue's coverage of the PFAS underestimate study.