Human exposure pathways to poly- and perfluoroalkyl substances (PFAS) from indoor media: A systematic review

Abstract

Background:

Human exposure to per- and polyfluoroalkyl substances (PFAS) has been primarily attributed to contaminated food and drinking water. There is information indicating other sources and pathways of exposure in residential environments, but few studies report relationships between these indoor media and human biomonitoring measurements.

Methods:

This study adapts existing systematic review tools and methodologies to synthesize evidence for PFAS exposure pathways from indoor environment media including consumer products, household articles, cleaning products, personal care products, and indoor air and dust. Studies were identified using innovative machine learning approaches and pathway-specific search strings to reduce time needed for literature search and screening. The included studies and systematic review were evaluated using tools modified specifically for exposure studies. The systematic review was conducted following a previously published protocol (DeLuca et al., 2021) that describes the systematic review methodology used in detail.

Results:

Only 7 studies were identified that measured the targeted subset of 8 PFAS chemicals in concordant household media (primarily house dust) and participant serum. Data extracted from the included studies were used to calculate exposure intake rates and estimate a percentage of occupant serum concentrations that could be attributed to the indoor exposure pathways. These calculations showed that exposure to PFOA, PFOS, PFNA, and PFHxS from contaminated house dust could account for 13%, 3%, 7%, and 25% of serum concentrations, respectively. Inhalation of PFAS in indoor air could account for less than 4% of serum PFOA concentrations and less than 2% of serum PFOS and PFNA concentrations. A risk of bias was identified due to participant profiles in most of the studies being skewed towards white, female, and higher socioeconomic status.

Conclusions:

Along with synthesizing evidence for estimated contributions to serum PFAS levels from indoor exposure media, this systematic review also identifies a consistent risk of bias across exposure study populations that should be considered in future studies. It highlights a major research gap and need for studies that measure concordant data from both indoor exposure media and participant serum and the need for continued research on exposure modeling parameters for many PFAS chemicals.

1. Introduction

Per- and polyfluorinated alkyl substances (PFAS) are synthetic commercial chemicals that are known for their water-resistant, stain-resistant, fire-resistant, and anti-stick properties. The extensive use and persistence of these chemicals in the environment has exposed humans to PFAS through a variety of routes including drinking water, food, consumer products, food packaging, cleaning products, personal care products, house dust, and indoor air. Human exposure to PFAS is mainly assessed through measurements of blood serum, in which concentrations of several legacy PFAS chemicals - perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) – have begun to decline in the United States (CDC, 2019) after voluntary and regulatory phase-outs starting in the early 2000s (U.S. EPA, 2002; U.S. EPA, 2016a; U.S. EPA 2016b).

Studies have found that the primary human exposure pathway for PFAS is ingestion of contaminated drinking water and food, especially near highly contaminated sites where fluorochemical manufacturing or use of aqueous film forming foam (AFFF) has occurred (Ericson et al., 2008; Haug et al., 2010; Hu et al., 2016; Daly et al., 2018; Domingo & Nadal, 2019). However, inhalation of dust and airborne volatiles, dermal contact with cleaning or personal care products, and ingestion from food packaging are also pathways of exposure but are not as well-represented in the literature (Vestergren et al., 2008; Lien et al., 2013; Kang et al., 2016; Boronow et al., 2019; Poothong et al., 2019; East et al., 2021).

The contribution of exposure from these lesser-studied pathways to serum PFAS concentrations is important for determining the relative source contribution (RSC), which is the proportion of total daily exposure to a chemical that is attributed or allocated to tap water. Two recent studies considered source apportionment from drinking water for PFOS and PFOA. One study using data from a national cohort of U.S. women in the Nurse’s Health Study estimated that between 11% and 14% of PFOA intake was from drinking water, and between 4.5% and 5.7% for PFOS (Hu et al., 2019). Another study using a scenario-based risk assessment model along with data collected from North America, Europe, and Japan estimated that drinking water contributed 10% of PFOA exposure, and between 3% and 10% for PFOS (Gebbink et al., 2015). Diet has been found to potentially account for > 40% of PFOA and PFOS exposures in adults (Sunderland et al., 2019). Based on these analyses, there may be another significant proportion of PFOA and PFOS exposure that remains uncharacterized.

This systematic review (SR) synthesizes evidence for exposure from indoor residential pathways to inform understanding of variability in serum PFAS levels in the general population from sources other than drinking water and diet. A subset of well-studied substances – perfluorooctanoic acid (PFOA), perfluorooctanesulfonate (PFOS), perfluorobutanoic acid (PFBA), perfluorobutane sulfonate (PFBS), perfluorodecanoic acid (PFDA), perfluorohexanoic acid (PFHxA), perfluorohexanesulfonate (PFHxS), and perfluorononanoic acid (PFNA) – were targeted in the literature for subsequent analysis. A protocol describing methodologies and tools for the systematic review was peer-reviewed and published prior to the commencement of this systematic review (DeLuca et al., 2021). This study utilizes innovative SR tools for exposure science studies, including exposure pathway-specific search strings for use in artificial intelligence screening software. It presents estimates for the percent of measured serum concentration that can be attributed to measured indoor PFAS exposure media in the included study populations.

2. Methods

The methods used in this systematic review were previously described in detail in a peer-reviewed protocol (DeLuca et al., 2021) and are based on those described in U.S. EPA’s Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments (U.S. EPA, 2019) and the Navigation Guide (Woodruff et al., 2011; Woodruff and Sutton, 2014; Johnson et al., 2014; Lam et al., 2016). This protocol is summarized briefly below. Any minor deviations from the protocol that were made during the systematic review process are stated in the following sections.

2.1. Scoping and problem formulation

The results of a broad preliminary literature search were organized into 7 major categories of known pathways of human exposure to PFAS – environmental media, home products/ articles/ materials, food packaging, personal care products, cleaning products, clothing, and specialty products. This scoping literature search indicated that exposure through environmental media, particularly through diet and drinking water, was the most well-studied pathway in the literature. Diet and drinking water have been reported to be major contributors of human exposure to PFAS (Ericson et al., 2008; Huag et al., 2010; Lorber & Egeghy, 2011; Gebbink et al., 2015; Hu et al., 2016; Daly et al., 2018; Domingo & Nadal, 2019; Zhang et al., 2019). Exposures from residential media are less represented in the literature, but could contribute to small, but chronic, doses of PFAS exposure in the general population that can have an impact on long-term human health outcomes (Vestergren et al., 2008; Lien et al., 2013; Kang et al., 2016; Boronow et al., 2019). Findings from the scoping literature search showed that 58% of the literature focused on PFAS in environmental media, particularly food and water, whereas PFAS in residential and indoor media was represented in only 13% of the findings (DeLuca et al., 2021). This discovery guided problem formulation for the systematic review, where we focus on characterizing PFAS exposures from residential and indoor media in the general population.

2.2. Population, Exposure, Comparator, and Outcome (PECO) criteria

The objective of this study was to evaluate evidence of PFAS exposure pathways in the residential environment by considering the relationship between occurrence of PFAS chemicals indoors and levels of these chemicals in serum. Specifically, this systematic review aimed to answer the question, “For the general population, what effect does exposure from PFAS chemicals via indoor media have on serum concentrations of PFAS?” To address this question, estimated serum concentrations of PFAS calculated from measurements of PFAS in these media were calculated to explain variability in participant PFAS levels from sources other than drinking water and diet. In the systematic review, we identify indoor exposure media as being consumer products, household articles, cleaning products, personal care products, or indoor air and dust. The study focuses on a subset of relatively well-studied PFAS chemicals including perfluorooctanoic acid (PFOA), perfluorooctanesulfonate (PFOS), perfluorobutanoic acid (PFBA), perfluorobutane sulfonate (PFBS), perfluorodecanoic acid (PFDA), perfluorohexanoic acid (PFHxA), perfluorohexanesulfonate (PFHxS), and perfluorononanoic acid (PFNA). Criteria found in the Population, Exposure, Comparator, and Outcome (PECO) statement guided study inclusion criteria (Table 1). Supplemental studies were tracked through the title and abstract screening step but were not carried any further during the systematic review. While the protocol indicates that additional PFAS species that are not listed in the PECO criteria but are reported in included studies would be listed in the study characteristics table, the additional species were omitted in our reporting and were not analyzed or tracked any further. The included studies were organized by chemical species and exposure media type.

Table 1.

Populations, Exposures, Comparators, and Outcomes (PECO) criteria from DeLuca et al. (2021).

PECO element	Evidence
Populations	Adults and/or children in the general population.
Exposures	Measured occurrence of PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, or PFNA in indoor exposure media, including indoor air, dust, food packaging, articles (e.g. cooking utensils), materials (e.g. clothing), cleaning products, and personal care products.
Comparators	A reference population exposed to lower levels, no exposure, or exposure below detection limits, such as the lower 10th percentile for PFAS serum concentrations from the U.S. general population from the National Health and Nutrition Exanimation Survey (NHANES).
Outcomes	Serum (or whole blood, plasma) concentration of PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, or PFNA.

2.3. Search strategy

Four databases – Web of Science, PubMed, Toxline/ToxNet, and ProQuest – were searched for relevant studies with help from U.S. EPA information specialists. Search terms for each PFAS species in the databases are found in the supplemental material of DeLuca et al. (2021). Initial searches were conducted for literature published through May 2020 for the Web of Science and PubMed databases and through April-May 2019 for the Toxline/ToxNet database. The literature search was updated during the systematic review to include recently published literature through April 2021 for the Web of Science, PubMed, and ProQuest databases. No additional studies were identified through manual reviews of the literature by experts or reviewers.

The literature identified in the database searches were imported into Sciome Workbench for Interactive computer-Facilitated Text-mining (SWIFT) Review software to pre-filter references that pertain to the study question (Howard et al., 2016). This software separates large bodies of literature into categories that can include exposure pathways, chemicals, and evidence streams. Search queries used to prioritize the literature search results in SWIFT-Review based on the PECO criteria for this study are found in the Supplemental Material of DeLuca et al. (2021). These search strategies significantly reduce the number of studies to be manually screened for inclusion or exclusion.

2.4. Study screening

Studies exported from SWIFT Review were imported into SWIFT Active Screener, a screening software that records reviewers’ inclusion/exclusion screening decisions and iteratively uses them to predict which of the remaining literature is likely to be included in the study (Howard et al., 2020). The title and abstract of each study were manually screened by two reviewers (AM, RS, ND) for relevancy to the study’s PECO criteria. Conflicts in reviewer inclusion decisions were tracked in the software and were resolved through discussion with an additional reviewer (ECH). Studies tagged for potentially supplemental information in this title and abstract screening were not tracked any further in the screening process. Studies tagged as supplemental were mostly those that reported only blood or serum PFAS measurements without corresponding indoor exposure media measurements and those that assessed exposures in a highly exposed population, either through occupational exposure or near a highly contaminated site. Other supplemental studies reported secondary exposure data instead of primary exposure data.

Studies included by the reviewers at the title and abstract level were imported into DistillerSR software (DistillerSR, 2021) to undergo full-text screening by two reviewers (JM, ND), where conflicts were also resolved through discussion with a third reviewer (ECH). Studies that were included after the full-text screening were advanced to the data extraction step in DistillerSR.

2.5. Data extraction

Data extraction of information relating to study characteristics, study design, and study context from each study was conducted and recorded in DistillerSR software by two reviewers (JM, ND). The reviewers also extracted summary PFAS measurement data from serum and any PECO defined indoor exposure media. While it was anticipated to extract mean PFAS concentration values from serum and residential media in the protocol, we found that the median was more frequently reported in the studies and was more robust to outliers. In this systematic review, the reviewers extracted median, or geometric mean if median was not reported, values from the included studies and recorded their responses in DistillerSR. One of the study authors (Kim et al., 2019) was contacted via email and provided us with median values from their datasets. Discrepancies in the two reviewers’ extracted data were resolved through discussion with a third reviewer (ECH). Due to the limited number of studies included at this stage of the systematic review, evidence maps were not produced from literature inventory summaries as was anticipated in the protocol. The limited number of studies also prompted the creation of only one study characteristics table for all PFAS species and media types, whereas separate study characteristics tables for each chemical were anticipated in the protocol.

2.6. Study evaluation

This systematic review used a modified version of the risk of bias tool for epidemiology studies described in EPA’s Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments (U.S. EPA, 2019) to evaluate each study. The risk of bias domains in the modified tool for exposure studies included exposure measurement, participant selection, and analysis and is found in the supplemental material in DeLuca et al. (2021). Potential conflict of interest is also evaluated for the included studies. Two reviewers (JM, ND) independently assessed the risk of bias for each study by giving a rating of “good,” “adequate,” or “deficient,” for each domain and recording their reasoning for each rating. Discrepancies in the ratings for each domain were resolved through discussion with a third reviewer (ECH). Due to the limited number of studies included in the review, an overall risk of bias was not determined here and instead was evaluated during the strength of evidence assessment detailed in Section 2.8.

2.7. Synthesis of evidence

Extracted medians or geometric means from the included studies were organized by PFAS species and media type. Estimated daily intakes (ng/day) were calculated using the pathway-specific equations and exposure factors in Table 2. Indoor exposure media with more than one exposure route (e.g. dust ingestion and dust absorption) were summed to determine a total daily intake for that indoor media matrix. The estimated PFAS concentration in serum that can be attributed to each indoor exposure media measurement from the studies was then calculated from a simple first-order pharmacokinetic (PK) model (Equation (1)) which assumes steady-state, as described in Lorber & Egeghy (2011),

$$C=DP/\left(k{P}^{\star }Vd\right)$$ (1)

where C is serum concentration (ng/ml), DP is daily absorbed dose (ng/kg-bw/day), kP is a first-order elimination rate in the body (1/day), and Vd is the volume of distribution (ml/kg). While validation of the first-order PK model in Lorber & Egeghy (2011) used data from only adult U.S. populations, we apply the same equation to child populations and those from outside the U.S. in this study. Values used for the variables in the PK equation for serum concentration are reported in Table 2. Values for the Vd and Kp parameters were updated since publishing of the protocol in order to prioritize data and rationale from recently published literature.

Table 2.

Exposure factors, exposure intake equations, and estimated serum concentration equation used to calculate estimated exposures from each exposure pathway included in the systematic review for both adults and children.

Exposure Factors and Intake Calculations	Value	Source
Dust ingestion, ng/day. Intake = Conc × IR × AF		Lorber & Egeghy (2011)
Concentration data (Conc): ng/g dust	Primary data (Table S1)
Ingestion Rate (IR): g dust/day	Child: 0.03 Adult: 0.02	US EPA Exposure Factors Handbook Table 5–1 (2017 update), U.S. EPA (2011)
GI Absorption Fraction (AF)	100%	Tian et al. (2016)
Dermal absorption of dust, ng/day. Intake = Conc × DL × TC × T × AF		Lorber & Egeghy (2011)
Concentration data (Conc): ng/g dust	Primary data (Table S1)
Dust Load (DL): g dust/m2	3.55	Pang et al. (2002)
Transfer Coefficient (TC): m2/hr	0.06	Cohen Hubal et al. (2006)
Time Performing Activity (T): hr/day	10	Egeghy & Lorber (2011)
Dermal Absorption Fraction (AF): %	4.8%	Egeghy & Lorber (2011)
Inhalation indoor air, ng/day. Intake = Conc × R × T × AF		Lorber & Egeghy (2011)
Concentration data (Conc): ng/g dust	Primary data (Table S1)
Inhalation Rate (IR): m3/day	Childa: 12	US EPA Exposure Factors Handbook Table 6–1 U.S. EPA (2011)
	Adult: 16
Fraction of Day Performing Activity (T)	Child: 0.792	Egeghy & Lorber (2011)
	Adult: 0.875
Lung Absorption Fraction (AF)	100%	Kennedy et al. (2004)
Body weight (kg)	Childa: 31.8	US EPA Exposure Factors Handbook Table 8–1 U.S. EPA (2011)
	Adult: 80.0
Serum Concentration Estimate, (ng/ml). Conc = DP / (kP * Vd)		Lorber & Egeghy (2011)
Daily Absorbed Intake (DP): ng/kg-bw/day	Calculated using equations above, divided by body weight
Elimination Rate in Body (kP): 1/dayb	PFOA: 0.00054	Multiplec
	PFOS: 0.00053	Multipled
	PFNA: 0.00078	Zhang et al. (2013)
	PFHxS: 0.00014	Multiplee
	PFHxA: 0.02166	Russell et al. (2013)
Volume Distribution (Vd): ml/kg	PFOA: 200	Gomis et al. (2017)
	PFOS: 230	Gomis et al. (2017)
	PFNA: 200	Ohmori et al. (2003)
	PFHxS: 230	Poothong et al. (2020)
	PFHxA: 200	Ohmori et al. (2003)

^aExposure factor for children 6 to less than 11 years old.

^bElimination rate calculated as ln(2)/half-life; half-life is average for males and females reported in respective studies.

^cZhang et al. (2013), Xu et al. (2020), Worley et al. (2017), Li et al. (2018), Olsen et al. (2009), Bartell et al. (2010).

^dZhang et al. (2013), Xu et al. (2020), Worley et al. (2017), Li et al. (2018), Olsen et al. (2009).

^eZhang et al. (2013), Xu et al. (2020), Worley et al. (2017),Li et al. (2018).

The proportion of serum PFAS concentrations attributed to indoor media exposures was calculated as a percent of the reported median or geometric mean serum PFAS concentrations from the respective studies. These results were reported for each study, grouped by media type and PFAS species. Where three or more studies measured the same indoor media and PFAS species, a median weighted by study sample size was calculated for the group of studies. Weighted medians were also calculated for groups of studies with similar study characteristics and context (e.g. location, population). A narrative summary of results includes comparisons between PFAS species by exposure media type in addition to comparisons between media types for each chemical species. All calculations were performed in R Statistical Computing Software (R Core Team, 2020). A worked example of these exposure calculations is shown in Appendix A.

2.8. Strength of evidence assessment

The systematic review used a modified version of the strength of evidence tool for epidemiology studies described in EPA’s Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments (U.S. EPA, 2019) that can be found in the supplemental material of DeLuca et al. (2021). The strength of evidence was considered “moderate” at the beginning of the assessment and either “increased strength” or “decreased strength” based on the risk of bias across studies, consistency, strength (effect magnitude) and precision, and coherence. A “neutral” response could also be chosen for each category. Two reviewers (JM, ND) independently judged the strength of evidence categories, recorded their reasoning, and designated an overall strength of evidence for the systematic review as either “high”, “medium,” or “low.” Discrepancies between the reviewer’s judgements and overall rating was resolved through discussion with a third reviewer (ECH).

3. Results and discussion

3.1. Search strategy

Fig. 1 is a flow diagram of the literature search and screening process performed during the systematic review. Relevant studies were initially identified through searches of Web of Science, PubMed, and Toxline/ToxNet that were stored in U.S. EPA’s Health and Environmental Research Online (HERO) database through May 2019 (U.S. EPA, 2021b). This initial search identified 7,334 studies that contained keywords relating to the 8 PFAS species targeted for the systematic review. The literature search update, which searched the ProQuest database in place of the ToxLine/ToxNet database, was conducted in April 2021 and identified 86 additional studies. The literature prioritization performed in SWIFT Review software filtered studies at the title and abstract level based on search strings for human evidence stream, indoor exposure pathways, and human exposure measures. This prioritization identified 486 studies to export for manual screening by reviewers.

Fig. 1.

Flow diagram showing literature search and screening steps followed during the systematic review. N indicates the number of studies screened in each step.

3.2. Study screening

The 486 studies prioritized in SWIFT Review were imported into SWIFT ActiveScreener software for title and abstract screening by reviewers, where each study was screened by at least two reviewers (AM, RS, ND). Because of the relatively small number of studies in this stage of screening, the predictive model in SWIFT Active Screener was not solely relied upon for making inclusion and exclusion decisions. All studies were manually reviewed for inclusion or exclusion, and a comparison to the predictive model showed that the same studies would have been included using either method. After inclusion and exclusion conflict resolution, 14 studies were included at the title and abstract level and were advanced to full-text screening.

The 14 studies included at the title and abstract level were imported into DistillerSR software to facilitate and record screening decisions at the full-text level by two reviewers (JM, ND). One study was not found at the full-text level and was therefore excluded from further analysis (Costopoulou et al., 2013). Another study was excluded because it included the same exposure data as was used in another study in the screening that was included (Koponen et al., 2018). Other studies were excluded at this stage because they did not report primary exposure media data or focused on a highly exposed population (Zhang et al, 2010; Beesoon et al., 2012; Rovira et al., 2019; Trasande et al., 2017; Zheng et al., 2020). After conflict resolution with a third reviewer (ECH), 7 studies were included at the full-text screening level and advanced to data extraction phase of the systematic review (Table 3). These 7 studies represent a very limited body of literature of concordant residential exposure media and biomonitoring measurements.

Table 3.

Study characteristics table for included studies in the systematic review.

Reference Name	Location	Years Sampled	Population	Reported Demographics	Media matrix	Media matrix sample size	Biomonitoring matrix	Biomonitoring matrix sample size	PFAS species analyzed in both matrices
Fraser et al., 2013 a	Boston, MA, US	2009	Adult office workers	90% white, 84% female	House dust	30	Serum	31	PFOA, PFOS, PFNA, PFBA, PFBS, PFHxS, PFHxA
Wu et al., 2015	California, US	2008–2009	Children		House dust	82	Serum	68	PFOA, PFOS, PFNA, PFHxS
			Parents of children	65–82% female, 99% with ≥ 12 years education		82		90
			Older adults			42		59
Makey et al., 2017	Vancouver, Canada	2007–2008	Pregnant women	82% white, 83% with university degree	House dust	48	Serum	50	PFOA, PFOS, PFNA
					Indoor air	39			PFOA, PFNA
Byrne et al., 2017 b	Alaska, US	2013–2014	Adults	Alaska Natives	House dust	49	Serum	85	PFOA, PFOS
Balk et al., 2019 c	Eastern Finland	2014–2015	Children	Not reported	House dust	63	Serum	54	PFOA, PFOS, PFNA, PFHxS
					Indoor air	57			PFOA, PFOS, PFNA
Kim et al., 2019 d	Seoul, Korea	2014	Adults and children (pooled)	Male/female ratio ~ 1:1	House dust	13	Serum	50	PFOA, PFOS, PFNA, PFHxS, PFHxA
Poothong et al., 2020 e	Oslo, Norway	2013–2014	Adults	74% female, 93% with ≥ 12 years education	House dust	≤ 7	Serum	61	PFOA, PFOS, PFNA, PFHxS

^aSerum and indoor air data reported in Fraser et al. (2012).

^bData reported in supplemental material.

^cDust data reported in Winkens et al. (2018). Serum data reported in Koponen et al. (2018). Indoor air data reported in Winkens et al. (2017).

^dData requested and provided by study author via email.

^eDust data reported in Padilla-Sánchez and Haug (2016). Serum data reported in Poothong et al. (2017).

3.3. Data extraction

Two reviewers (JM, ND) independently extracted study characteristics from the included studies like study population, location, years of sampling, matrices sampled, and sample sizes (Table 3). These study characteristics provide important context for interpreting the results of the systematic review, as well as give insight into gaps in the literature and research needs. The same two reviewers also independently extracted median PFAS measurement values for serum concentrations and exposure media from each of the included studies (Table S1). If the median was not reported in the study, the geometric mean was extracted and used in subsequent analyses instead. Table 3 indicates where extracted measurement data was found in the study’s supplemental materials. In some cases, our literature screening identified studies in which measurement data for both serum and exposure media were used for modeling but summary statistics for that cohort’s measurement data were reported in separate prior manuscripts. Discrepancies in extracted study characteristics, median, or geometric mean were resolved through discussion by the two reviewers. The corresponding author from one of the studies that did not report either medians or geometric means was contacted via email and graciously provided those data for use in the study (Kim et al., 2019). Only studies in which the same PFAS species were reported for both indoor media measurements and serum measurements were used in the evidence synthesis calculations and discussion of results.

3.3.1. Serum

Serum concentrations reported in the included studies were PFOA (7 studies), PFOS (7 studies), PFNA (7 studies), PFHxS (5 studies), PFHxA (1 study), and PFBS (1 study). Median concentrations in these studies ranged from 1.01 to 4.50 ng/mL for serum PFOA, 1.55–11.60 ng/mL for serum PFOS, 0.36–2.21 ng/mL for serum PFNA, and 0.22–1.70 ng/mL for serum PFHxS. Reported median serum concentrations for PFHxA and PFBS were 0.03 ng/mL and 0.04 ng/mL, respectively.

Study populations were from the United States, Canada, Finland, Norway, and Korea. Geographical variability in PFAS serum concentrations has been previously shown, where the U.S. has higher PFAS serum levels than other countries and should be considered when interpreting results of this study (Eriksson & Karrman, 2015; Jian et al., 2018). The highest median PFOA concentration in serum was reported in children in California, U.S., while the highest median PFOS and PFHxS concentrations in serum were reported in older adults from the same study in California (Wu et al., 2015). The highest median PFNA concentration in serum was reported in Alaskan Native adults in the U.S. (Byrne et al., 2017). The lowest serum concentrations of PFOS, PFNA, and PFHxS were reported in Balk et al. (2019) from children in Eastern Finland, and the lowest PFOA serum concentrations were reported in Byrne et al. (2017) from Alaskan Native adults. Serum concentrations from North American studies included here were within a similar range as the PFAS serum concentrations from nationally representative NHANES cycle 2009–2010 in the U.S. (CDC, 2013).

Sampling dates ranged from 2007 to 2015, over which period serum concentrations of PFOA and PFOS have been found to be declining in several studies from various countries (Gebbink et al., 2015; Olsen et al., 2017; Kato et al., 2011; Schroter-Kermani et al., 2013; Nost et al., 2014; Harada et al., 2010; Jian et al., 2018) Meanwhile a study in Korea found that serum PFOS levels had been decreasing through 2008, while PFOA levels were increasing (Harada et al., 2010). Serum concentrations of alternative PFAS species like PFNA and PFHxS were found to have increased or remain unchanged in Scandinavian populations (Nost et al., 2014; Glynn et al., 2015).

3.3.2. House dust

House dust data was extracted from the included studies for PFOA (7 studies), PFOS (7 studies), PFNA (6 studies), PFHxS (4 studies), PFHxA (4 studies), PFBA (2 studies), and PFBS (2 studies). Median concentrations in house dust in these studies ranged from 0.76 to 48.05 ng/g for PFOA, 0.95–62.00 ng/g for PFOS, 0.49–11.85 ng/g for PFNA, and 0.12–5.55 ng/g for PFHxS, 0.44–8.65 ng/g for PFHxA, 1.37–13.90 ng/g for PFBA, and 0.44–0.81 ng/g for PFBS. Geographic variability has also been observed in house dust concentrations, where homes in North America, particularly the United States, have been shown to have higher concentrations of PFAS chemicals than homes in other regions of the world (Egeghy & Lorber, 2011; East et al., 2021). Concentrations of PFOA and PFOS in house dust in the included studies were generally similar to those reported in other studies (East et al., 2021).

In the included studies, the highest median PFOA and PFNA concentrations in house dust were reported in the homes of older adults in California, U.S. between 2008 and 2009 (Wu et al., 2015) (Table S1). The highest median PFOS concentration in house dust was reported in the homes of pregnant women in Vancouver, Canada between 2007 and 2008 (Makey et al., 2017). The highest median PFHxA concentration in house dust was reported in Fraser et al. (2013) from homes in Boston, Massachusetts, U.S. sampled in 2009.

The lowest median PFOA concentration in house dust was reported in Byrne et al. (2017) from homes of Native Alaskans, and the lowest median PFOS concentration in house dust was reported in Balk et al. (2019) from homes in Eastern Finland. Balk et al. (2019) and Kim et al. (2019), which sampled homes in Seoul, Korea, reported PFAS concentrations in house dust that were generally lower than those from other studies for all chemical species.

3.3.3. Indoor air

Indoor air data included measurements for PFOA (2 studies), PFOS (1 study), PFNA (2 studies), and PFHxA (1 study) (Makey et al., 2017; Balk et al., 2019). Median concentrations in indoor air reported in these studies ranged from 0.02 to 0.05 ng/m³ for PFOA and 0.0024–0.0015 ng/m³ for PFNA. Median concentrations for PFOS and PFHxA were 0.0012 ng/m³ and 0.01 ng/m³, respectively. While indoor air concentrations were only reported in two of the included studies and for a limited number of PFAS chemicals, both studies reported concentrations for PFOA and PFOS that were generally consistent with those reported in other studies (East et al., 2021). PFOA had the highest concentration of the reported PFAS chemicals in indoor air for both studies (Table S1).

3.4. Study evaluation

Two reviewers (JM, ND) recorded their ratings for the 4 risk of bias domains for each study as well as their reasoning for each rating. Discrepancies in the ratings for each domain were resolved through discussion with a third reviewer (ECH) and the final ratings and reasonings are reported in Table S2. It was determined that there were no indications of a conflict of interest for any of the studies.

The exposure measurement domain was rated based on the quality of the measurement methods and the QA/QC procedures, as well as documentation of them in the studies’ methods. Many studies received a “good” rating for the exposure measurement domain, while one study received a rating of “adequate” due to the absence of a QA/QC procedure description in the article (Fig. 2). A rating of “deficient” in the exposure measurement domain was determined for one study due to its small sample size of house dust measurements and high percentage of measurements below the limit of detection for PFAS chemicals in those house dust samples.

Fig. 2.

Risk of bias ratings in four domains for each included study. Solid blue indicates rating of “good,” checkered blue indicates rating of “adequate,” and solid orange indicates rating of “deficient” for that domain.

The participant selection domain was evaluated based on whether the studies described their recruitment methods for participants and whether their studied populations were skewed toward a particular demographic that could affect interpretation of results. This domain was found to be lacking in many of the studies, with potential for bias from many studied populations being skewed towards white, female, and higher socioeconomic status (Table S2). One study that recruited their participants from women working at a public health facility was determined to be deficient in this domain because their studied population is likely more aware of chemical exposures than others in their community and potentially having lower residential and serum concentrations of PFAS (Poothong et al., 2020). Byrne et al. (2017) received a “good” rating for the participant selection domain because their participants were representative of the community being studied even though it may not have been representative on a national level.

A major finding from this systematic review is the difficulty, but importance, of recruiting diverse participants in exposure studies that represent the larger scale community or country for which the results may inform policy decisions. Previous studies have shown that PFAS concentrations in serum are influenced by race, ethnicity, sex, and age (Calafat et al., 2007; Kato et al., 2011), suggesting that the data in most of the included studies in the systematic review are inherently biased. Based on data collected from a U.S. cohort representing the general population, studies found that PFAS serum concentrations were generally lower in women than men, higher in non-Hispanic populations than Hispanic populations, and higher in participants with higher socioeconomic status than those with lower socioeconomic status (Calafat et al., 2007; Kato et al., 2011; Jackson-Browne et al., 2020). Other studies have found similar trends in community-level U.S. populations and other populations around the world (Bartolome et al., 2017; Sochorova et al., 2017; Coakley et al., 2018; Richterova et al., 2018; Kingsley et al., 2018; Nair et al., 2021; Chang et al., 2021). Based on these previous studies’ findings, the skew in our included studies’ populations in the systematic review could influence serum concentrations in an upward direction.

The analysis domain was rated based on the quality of each study’s analysis of their own measurement data and not the analysis done in the systematic review. Studies that were rated as “good” in this domain accounted for missing data or outliers appropriately and checked the distributions of their data, transforming them if needed. Studies that were rated as “adequate” either did not discuss how missing data were handled in their analysis or did not describe whether statistical assumptions were checked before running linear models or tests of significance. None of the included studies were rated as “deficient” in the analysis domain.

3.5. Synthesis of evidence

3.5.1. PFAS exposure from house dust

3.5.1.1. PFOA.

The percent of PFOA serum concentrations attributed to exposure from PFOA in house dust ranged from 0.78% to 27.63% for adults and from 13.5% to 31.74% for children (Table S3, Fig. 3). The median percent serum concentration of PFOA from all included studies, weighted by sample size, was 13.06%, which was the second largest percent serum level attributed to exposure from house dust out of the 4 chemicals for which multiple studies reported data. This weighted median is higher than the average percent source contribution of studies reviewed in De Silva et al. (2021) (6.58%), some of which were modeled exposures instead of calculations from concordant serum and house dust measurement data.

Fig. 3.

Percent serum concentration attributed to exposure from house dust ingestion and dermal contact for A) PFOS, B) PFOA, C) PFNA and D) PFHxS. Black dashed line indicates the weighted mean of all studies, weighted by sample size. Dark gray shading indicates adult populations and light gray shading indicates child populations. Note: y-axis scale for Figures A and C is different than scale for Figures B and D. Other study characteristics can be found in Table 1.

The lowest PFOA percent serum concentrations were calculated from data reported in Kim et al. (2019) and Byrne et al. (2017). Participants in these studies were adults and children (pooled) from Seoul, Korea and Alaskan Native adults from the United States, respectively. The highest percent serum concentrations attributed to exposure from house dust were calculated for children (31.74%) and parents of children (27.63%) living in California, U.S. (Table S3), both of which were paired to the same house dust samples (Wu et al., 2015). The higher percent serum concentration from house dust in children compared to the adults in the same household is consistent with previous studies indicating that childhood behavior and pathology contribute to increased chemical exposures in children (Cohen Hubal et al., 2006).

3.5.1.2. PFOS.

The percent of PFOS serum concentrations attributed to PFOS in house dust ranged from 0.39% to 15.54% for adults and from 2.09% to 10.32% for children (Table S3, Fig. 3). The median percent serum concentration of PFOS from all included studies, weighted by sample size, was 2.85%, which was the smallest percent serum level attributed to house dust out of the 4 chemicals for which multiple studies reported data. The weighted median of percent serum concentration of PFOS attributed to house dust in this study is lower than the average percent source contribution of studies reviewed in De Silva et al. (2021) (6.9%) from dust exposure. This result is consistent with previous studies showing that diet and drinking water, not house dust, are the primary sources of human exposure to PFOS (Sunderland et al., 2019).

Like PFOA, the lowest percent serum concentrations for PFOS were calculated from adults and children (pooled) living in Seoul, Korea (Kim et al., 2019) and Alaskan Native adults living in Alaska, U.S. (Byrne et al., 2017). The highest percent serum concentration for PFOS that can be attributed to exposure from house dust was calculated for pregnant women in Vancouver, Canada (Makey et al., 2017). Estimates of PFOS source contribution from house dust has been shown to be higher in participants from North America than from other regions (De Silva et al., 2021).

3.5.1.3. PFNA.

The percent of PFNA serum concentrations attributed to exposure from PFNA in house dust ranged from 0.55% to 9.50% for adults and from 7.77% to 15.21% for children (Table S3, Fig. 3). The median percent serum concentration of PFNA from all included studies, weighted by sample size, was 7.39%, which was the second smallest percent serum level attributed to exposure from house dust out of the 4 chemicals for which multiple studies reported data. Relative source contribution for PFNA from house dust (5%) was also calculated by another study from data reported Poothong et al. (2020). The percent serum concentration calculated here from the data in that study (0.89%) is lower than the previous estimate, which is likely due to the lower daily ingestion rate used in these calculations. The lower ingestion rate used here is from an updated version of the U.S. EPA’s Exposure Factors Handbook than was used in Poothong et al. (2020).

The lowest percent serum concentrations were calculated from adults and children (pooled) in Kim et al. (2019) and adults in Poothong et al. (2020). Participants in these studies were located in Seoul, Korea and Oslo, Norway, respectively. The highest percent serum concentration was calculated from children in California, U.S. (Wu et al. 2015). Percent serum concentrations of PFNA from house dust exposure calculated for child populations in the systematic review were higher than those calculated for most of the adult populations (Table S3).

3.5.1.4. PFHxS.

The percent of PFHxS serum concentrations attributed to exposure from house dust ranged from 0.73% to 41.92% for adults and from 39.92% to 54.76% for children (Table S3, Fig. 3). The median percent serum concentration of PFHxS from all included studies, weighted by sample size, was 24.99%, which was the largest percent serum level attributed to exposure from house dust out of the 4 chemicals for which multiple studies reported data. Relative source contributions for PFHxS from house dust have been reported previously in two studies, both of which were included in this systematic review (Balk et al., 2019; Poothong et al., 2020). The range in previously reported source contribution from house dust for those two studies (1% – 38%) was also mirrored in the results in this systematic review (Table S3).

The percent serum concentrations from adults and children (pooled) from Seoul, Korea (Kim et al., 2019) and adults from Oslo, Norway (Poothong et al., 2020) were significantly lower than those calculated for other included studies. The highest percent serum concentrations were in children and their parents from California, U.S. (Wu et al., 2015) and children from Eastern Finland (Balk et al., 2019). Up to 55% of the PFHxS in the serum of children in California could be from dust ingestion and dermal contact, indicating that this could be a main exposure pathway for that population. Children’s serum in Finland had relatively lower concentrations of PFHxS compared to those from California, however up to 40% of the PFHxS in children in Finland’s serum could be from exposure through dust. Previous studies have attributed PFHxS exposure to AFFF contamination of drinking water or consumer products like carpet and food packaging (Hu et al., 2019; Sunderland, 2019).

3.5.1.5. PFHxA.

Only one study included in the systematic review measured PFHxA in both house dust and serum (Kim et al., 2019). Less than 1% of the serum PFHxA concentrations in this pooled population of children and adults from Seoul, Korea could be attributed to exposure from house dust measured in that study (Table S3). Two other studies that estimated PFHxA source contribution from house dust without concordant measurements also calculated or modeled a low percentage (4%) (Gebbink et al., 2015; Poothong et al. 2020).

3.5.2. PFAS exposure from indoor air

Two of the included studies reported concordant indoor air and serum PFAS concentrations (Makey et al., 2017; Balk et al., 2019). The percent of PFOA serum concentration attributed to exposure from indoor air inhalation was 2.80% for children in Eastern Finland and 4.01% for pregnant women in Vancouver, Canada, while percent of serum PFNA concentrations was 1.28% for children and 0.29% for pregnant women (Table S4, Fig. 4). Indoor air exposure to PFOS contributed to 0.20% of serum PFOS concentrations in children. Studies that have estimated relative source contribution for these same PFAS species from inhalation, but not specifically indoor air inhalation, have found similarly low percentages (less than 6%) (Sunderland et al., 2019; De Silva et al., 2021). However, one study estimated that up to 14% of PFOA serum concentrations could be attributed to inhalation in a high exposure scenario (Trudel et al., 2008; De Silva et al., 2021).

Fig. 4.

Percent serum concentration attributed to exposure from indoor air inhalation for PFOA, PFNA and PFOS. Dark gray shading indicates adult populations and light gray shading indicates child populations. Other study characteristics can be found in Table 1.

3.5.3. Sensitivity assessment

3.5.3.1. Exposure in children versus adults.

Weighted median (weighted by sample size) percent serum concentrations attributed to exposure from house dust for PFOA, PFNA, and PFHxS (15.38%, 8.54% and 41.45%, respectively) in children were considerably higher than those for adult populations (9.83%, 2.86% and 5.98%, respectively). Percent serum concentrations for PFOS in house dust were also higher in children (2.94%) than in adults (2.61%), but by less of a margin than was observed for the other PFAS chemicals. The higher body burden of PFAS in children versus adults has been demonstrated in previous studies due to childrens’ higher ingestion rate of dust (Kato et al., 2011). The largest percent serum concentrations from house dust in children was for PFHxS, while the largest percent concentrations in adults was for PFOA.

Interestingly, parents of children in Wu et al. (2015) had elevated percent serum concentrations of PFHxS (41.92%) compared to older adults (15.49%) in the same study. Children in this study, to which the same house dust samples as the parents were paired, had the highest percent serum PFHxS concentrations of all studies where over 50% of the serum concentrations could be attributed to exposure from house dust. Balk et al. (2019) child participants had almost 40% of serum PFHxS concentrations accounted for by exposure from house dust.

However, it should be considered that the steady state assumptions made for the pharmacokinetic modeling used in this study may not be appropriate for children. A dynamic modeling approach could be considered and developed for future studies to account for chemicals with longer half-lives and exposures occurring from childhood through adulthood, when a steady state would be reached (East et al., 2021). The exposure factors used here for intake calculations from dermal absorption of PFAS through dust were not differentiated between children and adults. While this may influence the small contribution of dermal absorption to total PFAS intakes from dust used in our PK modeling, these exposure factors for children specifically could not be identified at the time of this study.

3.5.3.2. Exposure in North America versus Europe.

Results of this evidence synthesis, although based on limited availability of data, showed that weighted medians of percent PFAS serum concentrations were higher for studies from North America than for studies from Europe. Weighted median percent serum concentrations for participants from North America for PFOA and PFOS (19.05% and 3.92%, respectively) were around 60% higher than weightedmedians from Europe for PFOA and PFOS (10.05% and 2.06, respectively). PFNA and PFHxS weighted median percent serum concentrations for North America (7.91% and 30.03%, respectively) were much higher than those for Europe (0.89% and 0.73%, respectively) by over 150%. Previous studies have shown that concentrations of PFAS in serum and house dust vary with geography, where participants and their homes in North American have higher concentrations than in other parts of the world (Egeghy & Lorber, 2011; Eriksson & Karrman, 2015; Jian et al., 2018; East et al., 2021).

3.6. Strength of evidence assessment

For the strength of evidence assessment, four domains were rated as “increased strength,” “neutral,” or “decreased strength” for all studies included in the systematic review as a whole – risk of bias, consistency, strength and precision, and coherence. The risk of bias domain was rated “decreased strength” due to biases in participant selection in most of the studies. Study cohorts were mainly skewed toward higher socioeconomic status and white and female participants, which has been shown to increase serum PFAS concentrations in other populations through differences in consumer product types or brands used in the household, lifestyle, diet, and proximity to sources or contaminated sites (Calafat et al., 2007; Kato et al., 2011; Tyrrell et al., 2013; Nelson et al., 2012; Buekers et al., 2018; Jackson-Browne et al., 2020; Bartolome et al., 2017; Sochorova et al., 2017; Coakley et al., 2018; Richterova et al., 2018; Kingsley et al., 2018; Nair et al., 2021; Chang et al., 2021). One study’s participant selection was representative of the specific population being studied but was likely not generalizable to the larger national population.

Consistency was rated “increase strength” because most studies found some correlation between dust or indoor air PFAS concentrations and serum PFAS concentrations. Some studies found positive correlations between dust concentrations and adults’ serum concentrations of PFOA (Makey et al., 2017; Kim et al., 2019; Poothong et al., 2020), PFOS (Kim et al., 2019; Poothong et al., 2020), PFNA (Fraser et al., 2013; Makey et al., 2017; Kim et al., 2019; Poothong et al., 2020), PFHxS (Kim et al., 2019; Poothong et al., 2020), and PFHxA (Kim et al., 2019). Meanwhile, studies also found that there were not significant correlations between dust and adults’ serum concentrations for some PFAS species (Fraser et al., 2013; Wu et al., 2015; Byrne et al., 2017). In children, correlations were found between house dust concentrations and serum concentrations for PFOS, PFNA (Wu et al., 2015) and PFHxS (Balk et al., 2019). Concentrations of PFAS in indoor air and adult serum were also found to be significantly correlated in one study (Poothong et al., 2020).

The strength and precision domain was rated “neutral” because intakes of many PFAS chemicals attributed to indoor media in the studies were small compared to dietary intake or drinking water ingestion in most studies, which agrees with the literature (Sunderland et al., 2019). Coherence was rated as “increased strength” because there is biological significance to the finding that PFAS concentrations in indoor media are reflected in PFAS concentrations in serum, albeit a small influence, as well as exposures from indoor media being higher in children than adults due to behavior and physiology.

The strength of evidence tool developed in the protocol for this systematic review (DeLuca et al., 2021) and used above would suggest that the overall weight of evidence was high. However, the authors do not consider the strength of evidence in this systematic review to be strong due to the lack of available concordant data published in the literature, significant risk of bias from study populations that are skewed in demographics and socioeconomic statuses, and uncertainties regarding exposure calculations for PFAS chemicals. Results of evidence synthesis are also not considered robust where weighted medians for some sensitivity studies were calculated using only two data points. Additional uncertainty in the study is derived from the estimation of parameters in exposure equations, such as the elimination rate, which is still an active area of research for PFAS chemicals. Serum PFAS concentrations derived from precursor compounds were not addressed during this systematic review but could contribute to additional PFAS levels in serum attributed to indoor and residential sources. The discrepancy in results of the strength of evidence assessment using the tool versus expert judgement suggests that future systematic review protocols for exposure studies may want to incorporate considerations such as the quantity of available evidence and the uncertainties in parameterization into their strength of assessment tools.

4. Conclusion

This systematic review adapted existing tools and methodologies to investigate the importance of PFAS exposure pathways from indoor and residential media including consumer products, household articles, cleaning products, personal care products, and indoor air and dust. A small body of literature was identified in the search and screening process in which studies measured concordant indoor media, namely house dust and indoor air, and serum PFAS concentrations from adults and children in the general population. Each included study was evaluated for risk of bias in four domains, with most studies being rated as deficient in the participant selection domain due to skewed demographic and socioeconomic profiles in their study populations toward white, female, and higher socioeconomic status participants. This finding reflects a challenge for many exposure studies when recruiting a small group of participants to generalize to the larger population, and study designs should consider and address the potential for bias in their study populations whenever possible. Because of this potential for bias from participant selection in the included studies and the overall lack of available concordant data, the strength of evidence for this systematic review is not considered to be strong.

Exposure intake calculations and a simple pharmacokinetic model were used to estimate a percentage of serum concentrations that could be attributed to each indoor exposure pathway for the PFAS chemicals reported. Results of this evidence synthesis showed that exposure to PFOA, PFOS, PFNA, and PFHxS from contaminated house dust could account for 13%, 3%, 7%, and 25% of serum concentrations, respectively. Two studies that reported PFAS measurements in indoor air showed that inhalation of PFAS could account for less than 4% of serum PFOA concentrations and less than 2% of serum PFOS and PFNA concentrations.

Trends observed in previous studies and national cohort studies were also demonstrated in the evidence synthesis results here. Percent serum concentrations for children were higher than those for adults for all PFAS chemicals. PFHxS in particular was observed in higher percent serums concentrations in participants who were children or lived with children, suggested that child-specific products or materials may be increasing levels of exposure in those homes. North American studies, particularly with study populations in California, had higher percent serum concentrations than populations from other regions in the world.

While the study provides some evidence for indoor exposure media’s potential contributions to serum PFAS concentrations, it ultimately highlights the lack of concordant measurement data from indoor and residential exposure media and paired serum specimens. This large research gap limits the ability to answer the systematic review’s research question and emphasizes the need for future exposure studies with concordant PFAS measurements in serum and in media other than drinking water and diet. It also promotes the development of modeling studies that can skillfully incorporate questionnaire data where indoor exposure media are not sampled for a study population. This would help to improve understanding of the relative source contributions from various exposure pathways for human PFAS exposure and better inform management decisions. This systematic review presents the notion that there could be many residential PFAS chemicals having an impact on serum concentrations, but more data is needed to truly investigate this hypothesis.