How would the NHANES data be utilized in PFAS risk assessment? For example, NHANES 2013-2014 data reported the 95th percentile for PFBS at or below the level of detection (0.1 ng/mL), and PFBS is no longer included in subsequent testing. Would this influence grouping that included PFBS, as an example?

This LOD is quite high. PFBS is difficult to identify and quantify analytical. Not a priority substance.
Taking all data from whatever dataset would result in median values of zero for most PFAS including others that are perceived as more important than PFBS (if not in a hotspot; see also EFSA reports).
Robust analysis for PFAS is not set yet.
I do not recommend to include PFBS (or PFBA, nor the penta SA or CA).

NHANES is useful for assessing the health risk of PFAS for which blood is the best biological matrix, at background exposure levels. The recent data from ATSDR in contaminated communities clearly showed that NHANES fails to represent exposure in hotspots. Also, it is possible that urine is the best matrix for measuring exposure to PFAS with a short biological half-life.

The NHANES data set provides useful information on exposure trends, but in my view, it has limited utility for conducting risk assessments, and in particular, any site-specific risk assessments. I do not believe that a lack of data on a specific PFAS excluded from the NHANES data set (e.g. PFBS) should influence any PFAS grouping decisions that are an outcome of this SciPinion exercise.

PFBS is considered to be very persistent and very mobile (vPvM) so would likely be regulated (and grouped with other perfluoroalkyl acids, PFAAs) under European Union (EU) legislation. PFBS has been designated a Substance of Very High Concern (SVHC) in the EU based on considerations of its persistence, mobility, and toxicity, with probable serious adverse effects to human health and the environment.

That said, the NHANES data could indeed be used as justification to exclude certain PFAAs such as PFBS from grouping for mixture risk assessment used for e.g. setting drinking water thresholds in other jurisdictions. I understand the scientific justification (i.e. its rapid elimination from humans) for excluding PFBS from a subgrouping of PFAAs, but personally prefer to err on the side of caution. Levels of PFBS could build up in some aquifers because of its extremely high persistence.

NHANES has been shown to be unable to detect chemicals that only reach a small fraction of the general population. As a result, chemicals with exposed populations that are limited to communities near manufacturing sites or occur as a result of specific occupations (firefighters and chemical plant employees) would not be well captured by NHANES.

National surveys such as NHANES are unlikely to generate useful information of PFAS unless large numbers of individuals (>5% of an age group) received doses that are detectable. This is only likely to occur for exposures related to consumer products, where the products are frequently used by large fractions of the population. PFAS containing consumer products are to my understanding limited to clothing and weather-resistant materials that use polymer PFAS. Polymer PFAS substances are poorly absorbed and during use do not degrade and release lower molecular weight PFAS. As a result, I would not expect PFAS to be detectable in NHANES.

The negative findings from NHANES could be used to confirm that exposure to a PFAS substance was not widespread. It could not be used to confirm the existence or nonexistence of exposed small populations.

In the past PFBS has been used in the manufacture of the consumer product called Scotchgard. However, it is my understanding that it was used as a telomer. In the final product, it existed as a side chain on a non-fluorinated polymer. Such polymers are poorly absorbed and are stable over the life of the product.

This finding would not be relevant to listing PFBS as PFAS.

This could be used in a problem formulation as part of a justification for excluding PFBS from a quantitative assessment.

Good point, although NHANES alone should not be used as data source. NHANES is a national sample, but some PFAS problems in drinking water are limited to local sources; for example, GenX exposure is high in North Carolina, and NHANES data will not reflect GenX exposure in that population. Therefore, besides NHANES data, a literature survey of all other reliable and published studies should be included to obtain comprehensive exposure information. When data are sporadic, a worst case scenario may be used.

The potential for human exposure is a key criteria for determining whether a chemical should be evaluated. The NHANES data, which represents exposure from all routes, indicates which compounds humans are highly exposed to and which they are not highly exposed to. With reference to PFBS, the NHANES data provides little support for widespread human exposure by any route. The half-life of PFBS in humans is approximately one month, so widespread exposure would result in measureable blood levels. In the UCMR 3, 4920 public water systems reported results and only 8 (0.16%) of those reported that levels of PFBS were greater than or equal to the minimum reporting level. I would think that the NHANES data and the UCMR3 data would suggest that PFBS would be of low priority, and if assessed, would not be a major driver in the assessment.

In the broadest of possible terms, NHANES data can be used to confirm "exposure". The concept of biomonitoring equivalents and dose reconstruction is important in further refining the potential quantitative value of NHANES data, especially for chemicals that either have a short half-life, or whose pharmacokinetic data are clouded by uncertainty. For chemicals with shorter versus longer half-lives, the absence of chemical detection is less confirmatory of exposure than for chemicals with longer half-lives. The half-life of PFBS may temper some expectations of the likelihood that the absence of PFBS from a sample indicates a lack of exposure.

I am not statistician, but ... regardless, PFBS should be included in the grouping strategy considered for PFAS. If exposure becomes an issue, then there are statistical means available to assess the likely distribution of non-detects. For example, assuming a normal or lognormal distribution and understanding that only 5% of the samples exceeded a quantified level of detection (blood concentration), the distribution of non-detects can be estimated.
But, for PFAS chemicals, the issue of biological half-life, toxicokinetics, and animal to human differences in toxicokinetics remains a major hurdle in a quantified estimate of risk based on serum concentration data.

There are a number of ways to deal with non-detects. Using one half the detection limit as a default in models is one common practice.

Another approach is to consider eliminating a chemical or group from the assessment. This needs to be approached with some caution. The note above does not say that PFBS was not detected at all; rather that the 95 percentile of observations was at a non-detect. NHANES data generally have sufficient numbers of subjects to examine the 99 percentile. For application to a large population (such as the US "general" population), the tail of the exposure distribution may contain a considerable number of persons.

For some screening use of NHANES data see
Lesa L. Aylward, Christopher R. Kirman, Rita Schoeny, Christopher Portier, Sean M. Hays. Evaluation of Biomonitoring Data from the CDC National Exposure Report in a Risk Assessment Context: Perspectives across Chemicals. Environ. Health Perspect. 121(3):287-294, 2013.

Lesa L. Aylward, Santhini Ramasamy, Sean M. Hays, Rita Schoeny, Christopher R. Kirman. Evaluation of urinary speciated arsenic in NHANES: Issues in interpretation in the context of potential inorganic arsenic exposure. Reg. Toxicol. Pharmacol. 69: 49–54, 2014.

It depends. Some populations still have detectable levels of PFBS (women firefighters - see Trowbridge et al., 2020) associated with specific activities such as the handling of aqueous film forming foams. Similarly, populations living near specific sources of PFAS may also have concentrations elevated about NHANES reported levels. For example, Kotlarz et al. (2020) reported that people drinking water sourced from a river contaminated with PFAS had higher concentrations of PFOA and PFOS than those surveyed through NHANES even though levels of PFOA and PFOS in this river were relatively low. Failure to detect a specific PFAS in biomonitoring efforts of the general population says nothing about specific and potentially highly exposed subpopulations.