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

Abstract

Background

Human exposure to per- and polyfluoroalkyl substances (PFAS) has been primarily attributed to contaminated food and drinking water. However, additional PFAS exposure pathways have been raised by a limited number of studies reporting correlations between commercial and industrial products and PFAS levels in human media and biomonitoring. Systematic review (SR) methodologies have been widely used to evaluate similar questions using an unbiased approach in the fields of clinical medicine, epidemiology, and toxicology, but the deployment in exposure science is ongoing. Here we present a systematic review protocol that adapts existing systematic review methodologies and study evaluation tools to exposure science studies in order to investigate evidence for important PFAS exposure pathways from indoor media including consumer products, household articles, cleaning products, personal care products, plus indoor air and dust.

Objectives

We will systematically review exposure science studies that present both PFAS concentrations from indoor exposure media and PFAS concentrations in blood serum or plasma. Exposure estimates will be synthesized from the evidence to answer the question, “For the general population, what effect does exposure from PFAS chemicals via indoor media have on blood, serum or plasma concentrations of PFAS?” We adapt existing systematic review methodologies and study evaluation tools from the U.S. EPA’s Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments and the Navigation Guide for exposure science studies, as well as present innovative developments of exposure pathway-specific search strings for use in artificial intelligence screening software.

Data Sources

We will search electronic databases for potentially relevant literature, including Web of Science, PubMed, and ProQuest. Literature search results will be stored in EPA’s Health and Environmental Research Online (HERO) database.

Study eligibility and criteria

Included studies will present exposure measures from indoor media including consumer products, household articles, cleaning products, personal care products, plus indoor air and dust, paired with PFAS concentrations in blood, serum or plasma from adults and/or children in the general population. We focus on a subset of 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).

Study appraisal and synthesis methods

Studies will be prefiltered at the title and abstract level using computationally intelligent search strings to expedite the screening process for reviewers. Two independent reviewers will screen the prefiltered studies against inclusion criteria at the title/abstract level and then full-text level, after which the reviewers will assess the studies’ risk of bias using an approach modified from established systematic review tools for exposure studies. Exposure estimates will be calculated to investigate the proportion of blood, serum or plasma) PFAS concentrations that can be explained by exposure to PFAS in indoor media.

1. Introduction

Per- and polyfluorinated alkyl substances (PFAS) are a class of synthetic chemicals that have been in commercial production since the 1940s and are known for their water-resistant, stain-resistant, fire-resistant and anti-stick properties. These chemicals can be found in industrial facilities, drinking water, living organisms, commercial household products, and food packaging (U.S. EPA, 2018). Because of their extensive use and lengthy persistence, the presence of PFAS in the environment is ubiquitous and most people in the United States have been exposed to them (Kato et al., 2015). Human exposure to PFAS is associated with various adverse health outcomes including increased cholesterol levels (Nelson et al., 2010; Geiger et al., 2014; Zeng et al., 2015; Liu et al., 2018), developmental effects in children (Liew et al., 2018), reproductive effects (Knox et al., 2011; Lopez-Espinosa et al., 2011; Petersen et al., 2018), thyroid hormone disruption (Kim et al., 2018), liver and kidney disease (Stanifer et al., 2018; Bassler et al., 2019), and immune system depression and toxicity (Granum et al., 2013; Pennings et al., 2016).

Human exposure to PFAS is primarily measured through blood serum concentrations. The half-lives of several common PFAS chemicals can range between 3 days and 35 years (Kudo, 2015). Long perfluorinated-chain PFAS are generally thought to have longer half-lives than short perfluorinated-chain PFAS (Kudo et al., 2001; Huang et al., 2019). Serum concentrations of long-chain PFAS – perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA) – have recently declined in the U.S. (CDC, 2019) due to agreements between the U.S. Environmental Protection Agency (U.S. EPA) and major PFAS industry companies for voluntary phase outs beginning in the early 2000s (U.S. EPA, 2002), regulatory actions taken by some of the States, changes in manufacturing processes, inclusion of the perfluorinated carboxylates and sulfonates as a pollutant in international negotiations (UNEP, 2017), and health advisories issued by the U.S. EPA in 2016 (U.S. EPA, 2016a; U.S. EPA 2016b). Serum concentrations of other PFAS have not demonstrated comparable declines (CDC, 2019). The difference in toxicity between exposures to long-chain PFAS and short-chain alternatives is not yet clear, but recent evidence suggests they can be similarly hazardous (Gomis et al., 2018).

It has been reported that human exposure to PFAS mainly occurs via ingestion pathways such as diet and drinking water, particularly in areas proximal to highly contaminated sites (Ericson et al., 2008; Haug et al., 2010; Hu et al., 2016; Daly et al., 2018; Domingo & Nadal, 2019). However, exposures are also linked to inhalation of dust and airborne volatiles, dermal contact with cleaning or personal care products, and ingestion from food packaging (Vestergren et al., 2008; Lien et al., 2013; Kang et al., 2016; Boronow et al., 2019; Poothong et al., 2019). While the relationships between exposure from drinking water and food ingestion and serum concentrations have been well-studied and the importance of these exposure pathways established, evidence of exposure through indoor environment pathways is not as well-represented in the literature. A systematic understanding of the relationship between indoor exposure media and serum concentrations of PFAS could better inform our understanding of “background” exposure levels coming from pathways that do not account for the majority of the general population’s exposure. Background exposure levels are important to understand and quantify in future risk assessments and chemical management decisions because they provide a baseline for which regulations of highly contaminated exposure pathways can be reasonably set to.

This systematic review (SR) aims to synthesize evidence for exposure from indoor pathways that can explain variability in serum PFAS levels in the general population from sources other than drinking water and diet. Each PFAS chemical in 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) – will be reviewed, analyzed, and presented in the systematic review individually. The protocol generally follows the methodology from the 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., 2016a) (Figure 1). However, it also presents innovative SR methodologies for exposure science studies, including the development of exposure pathway-specific search strings for use in artificial intelligence screening software. Elements of the protocol were piloted to assess the relevance of study evaluation criteria to exposure science studies and identify evidence synthesis approaches applicable to the research question.

Figure 1.

Workflow of systematic review protocol adapted from the EPA ORD Staff Handbook for Developing IRIS Assessments (U.S. EPA, 2020). Scoping, Problem Formulation, and Literature Steps (green bracket and arrows) are an iterative process throughout the beginning of protocol development. Steps under red bracket are described throughout the protocol.

2. Methods

2.1. Scoping and Problem Formulation

Potential human exposure pathways to PFAS chemicals begin with the manufacturing of these chemicals and PFAS-containing products and extend into theproduct use and end-of-life phases (Table 1). The manufacturing phase of PFAS or PFAS-containing products can release the chemicals into the environment through ambient air, wastewater, or sludge. The use of products by consumers can also release PFAS through volatilization, abrasion to particulates, and transfer from packaging materials or textiles. The resulting exposure media, including outdoor air, drinking water, soil, biota, and house dust, can then take several routes to human exposure through inhalation, ingestion, and dermal contact. Because of the lengthy persistence of many PFAS chemicals in the environment, end-of-life phase exposures can result from contaminated wastewater, landfill leachate, and incineration emissions.

Table 1.

Human Exposures to PFAS from Manufacturing and Use of Consumer Products

Source	Release & Transfer	Exposure Media	Exposure Route
•Industrial PFAS production •Industrial product manufacturing •Product Use •Product Recycle •Product Disposal	•Manufacture Phase: Ambient Air Effluent Wastewater Sludge Biosolids Soils and plant uptake •Consumer Use Phase: Volatilization to indoor air Abrasion to particulates Cleaning to wastewater Transfer from packaging Transfer from textiles Transfer from animal products •End of Product Life: Wastewater Landfill, leachate Incineration	•Outdoor: Air Surface Water Ground Water Soil Biota •Indoor: Air Dust Surfaces Drinking Water Food	•Volatile inhalation •Particulate inhalation •Dust ingestion •Drinking water ingestion •Dietary ingestion •Dermal contact

A broad preliminary literature search for known pathways of human exposure to PFOA and PFOS was first conducted to facilitate problem formulation for the systematic review. The results from this scoping literature search were categorized into 7 major PFAS pathways: Media, Home Products/Articles/Materials, Food Packaging, Personal Care Products, Cleaning Products, Clothing and Specialty Products (Figure 2A). Exposure through environmental media, particularly through diet and drinking water (Figure 2B), was the most well-studied pathway category in the scoping literature search. Diet and drinking water have been reported in the literature as major pathways for PFAS exposure (Ericson et al., 2008; Huang et al., 2010; Hu et al., 2016; Daly et al., 2018; Domingo & Nadal, 2019; Zhang et al., 2019).

Figure 2.

Scoping literature search results for PFAS exposure through A) major pathway categories and B) sub-categories for environmental media pathways. References were tagged in SWIFT-Review literature prioritization software using search strings found in Appendix B. Numbers indicate the percentage of references tagged in each category or subcategory. References may overlap into multiple categories.

Indoor media exposures such as inhalation of airborne volatiles, ingestion of dust, ingestion as a result of transfer from food packaging, and dermal contact with indoor articles, materials, cleaning products, and personal care products, have been less represented in the literature thus far (Figure 2A). These less studied pathways 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). Therefore, this systematic review aims to characterize the importance of and relative contribution of PFAS exposures from indoor media pathways in the general population.

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

The objective of the systematic review is to answer the question, “For the general population, what effect does exposure from PFAS chemicals via indoor media have on serum concentrations of PFAS?” Exposure estimates calculated from indoor media concentrations will be used to help explain variability in participant PFAS levels from sources other than drinking water and diet. Indoor exposure media include consumer products, household articles, cleaning products, personal care products, plus indoor air and dust. The study will focus on a subset of 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). The Population, Exposure, Comparator, and Outcome (PECO) (Table 2) is a systematic review tool used to convey the study inclusion criteria and focus the literature search. An additional table (Table 3) provides categories of supplemental literature for studies that would not meet PECO criteria nor be analyzed during the systematic review but will be tracked during the screening process because they may provide important context for interpretation of results. Where additional PFAS species are reported within studies that do meet the PECO criteria and are included in the systematic review, those chemical species will be listed in the study characteristics table but not analyzed further. Using the same PECO criteria, literature will be organized by chemical species and evidence synthesis will be performed separately for each group of literature.

Table 2.

Populations, Exposures, Comparators, and Outcomes (PECO) criteria

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.

Table 3.

Categories of potentially relevant supplemental material

Category	Evidence
Additional PFAS species	Studies reporting serum (or whole blood, plasma) concentration and indoor media measurements for PFAS species other than PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, or PFNA.
Biomonitoring measurements	Studies reporting only serum (or whole blood, plasma) concentration of PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, or PFNA that do not include indoor media measurements from a related environment.
Indoor media measurements	Studies reporting only indoor media measurements of PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, or PFNA that do not include serum (or whole blood, plasma) concentration from a related population.
Impacted populations	Studies that measure indoor media and serum (or whole blood, plasma) concentration from a population that is known to be highly exposed to PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, or PFNA, either through contaminated drinking water contamination or occupational exposure.
No original data	Studies that are related to PECO criteria, but do not do primary data collection.

2.3. Search Strategy

With the help of EPA information specialists, we search for relevant literature in three databases:

• Web of Science (Thomson Reuters)

• PubMed (National Library of Medicine)

• ToxNet/ ToxLine (National Library of Medicine)

  • Retired database, used for search results stored through May 2019

• ProQuest

  • Will be used to update ToxNet/ToxLine search results post-May 2019

Search terms for each database include PFOA, PFOS, PFBA, PFBS, PFDA, PFHxA, PFHxS, PFNA, their respective synonyms, and their CAS numbers (Appendix A). Literature is not filtered by date or language. Before completing the systematic review, all databases will be searched again for recent publications that follow the initial database searches. Literature search results are stored in EPA’s Health and Environmental Research Online (HERO) database (https://hero.epa.gov/hero/). The literature search strategies described above are designed to be broad, but like any search strategy, studies may be missed. Thus, in addition to the database searches, these sources are used to identify studies that may have been missed based on the database search:

• Manual review of the reference list from final or publicly available draft assessments (e.g. EPA IRIS, ATSDR Toxicological Profile profile) or published reviews identified as supplemental material.

• Manual review of the reference list of studies screened as PECO-relevant after full-text review.

• References identified by expert consultants or during peer-review.

Literature identified in the database searches are 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). In brief, SWIFT Review has pre-set literature search strategies (“filters”) developed by information specialists. The pre-set filters can be customized by users as needed to separate literature by specific exposure pathways, chemicals, or evidence streams. The filters function like a typical search strategy where studies are tagged as belonging to a certain category if the terms in the search strategy appear in title, abstract, keyword or medical subject headings (MeSH) fields content. Pre-set search queries in SWIFT Review can be used to organize literature into lists that pertain to the PECO criteria, which can be extracted from the larger original list of literature to reduce the time and effort needed for screening.

Additional search queries were developed for pathway-specific filtering (e.g. environmental media, personal care products, household articles, etc.) and PECO criteria (e.g. biomonitoring) in SWIFT Review (Appendix B). Search queries were refined through expert consultation on exposure keywords and manual inspection of the database search results. These search strategies were used for the scoping and problem formation step in this protocol. It is anticipated to also be used verbatim during the systematic review to significantly reduce the number of studies to be manually screened for inclusion or exclusion. Studies not retrieved using these filters will not be considered further. Studies that include search terms from the human exposure measures tag, human evidence stream tag, and one or more of the indoor exposure pathway tags in the title, abstract, keyword, or MeSH fields will be exported as a RIS file for screening in SWIFT-ActiveScreener, as described below.

2.4. Study Screening

An overview of inclusion and exclusion criteria for study screening is found in Appendix C. The prioritized literature exported from SWIFT Review will be imported into SWIFT Active Screener. This software iteratively uses the reviewers’ inclusion/ exclusion inputs and a statistical model to predict which of the remaining literature is likely to be included in the study, expediting the screening process (Howard et al., 2020). In the software, two reviewers will independently screen the titles and abstracts of studies in a subset of the imported literature for relevance to the study’s PECO criteria and manually indicate whether the study should be included or excluded.

Concurrently, the software will use data from the reviewers’ inclusion or exclusion decisions to predict the selected outcome of the remaining literature in the subset. When SWIFT Active Screener indicates that it is likely to predict 95% of the reviewers’ inclusion decisions, a percent comparable to human error rates (Cohen et al., 2006; Bannach-Brown, 2018; Howard et al., 2016; ), the reviewers can decide to refrain from manually screening the rest of the studies. The software will indicate whether there are conflicts between the reviewers’ inclusion or exclusion decisions, and any conflicts will be resolved through discussion or consultation with a third reviewer if needed (Cohen Hubal et al., 2020).

Studies that the SWIFT ActiveScreener software predicts will be included in the systematic review at the title and abstract level will be imported into DistillerSR software and undergo a full-text screening for PECO criteria relevance by at least two reviewers. Reviewers at this stage will also tag the studies by the PFAS chemical or chemicals of which there is primary data reported. Studies that do not fully meet the PECO criteria, but are considered contextually relevant to the study question, will be annotated as supplemental throughout the screening process as outlined in Table 3. Screening strategies may be updated and refined as needed during the systematic review.

2.5. Data Extraction

After studies are identified for inclusion during the full-text screening, primary data extraction of variables relating to study characteristics (e.g. study year, study country), study design (e.g., exposure media type, sample collection), and study context (e.g. population, years over which samples were collected, age range of participants) for each PFAS chemical listed in the PECO will be performed independently by at least two reviewers. Discrepancies in the extracted data will be resolved through discussion between the two reviewers. Literature inventory summaries from this extracted data will be used to create evidence maps organized by PFAS species and exposure media type to 1) identify gaps in the literature and future research needs for assessments, and 2) produce user-friendly outputs for the body of literature (Wolffe et al., 2019).

In addition to study characteristics, the reviewers will also extract summary PFAS measurement data from serum (or blood, plasma)and any targeted indoor exposure media that meets the PECO criteria. The summary measurement data of interest are primarily the mean concentration values for each PFAS species in the PECO statement. However, when mean values are not reported in the study the median, range, and sample size may be used to calculate an estimated mean value. If these summary statistics are not reported in a study (e.g. studies that report only interquartile ranges), or they are not reported for individual PFAS species (e.g. total PFAS concentration), the data will not be extracted and will be excluded from further analysis. Metadata, such as biomonitoring media type, unit of measurement, time period over which data was collected, and laboratory measurement methodology, may be documented alongside extracted PFAS measurement data in order to assure that sound comparisons and/or combinations of data are made in the analysis. All extracted data will be stored, managed, and organized by PFAS chemical in DistillerSR software (Cohen Hubal et al., 2020). Conflicts in data extraction will be resolved through discussion or a third reviewer if needed. Any missing data or study information will be requested from the study authors by email or phone using contact information provided in the study record.

A study characteristics table, including summary findings, was developed to aid in the extraction of study data during the systematic review and provide readers with an overview of all studies included in the systematic review. The study characteristics table, including summary findings, was piloted using three studies (Fraser et al., 2013; Wu et al., 2014; Kim et al., 2019) identified as relevant to the PECO criteria with both dust and serum PFOA measurements (Appendix D). The systematic review will include a separate study characteristics table for each PFAS chemical listed in the PECO statement, each of which will be organized by indoor matrix type.

2.6. Study Evaluation

A standard methodology for risk of bias assessments for systematic reviews in the exposure sciences is not yet well agreed upon (Bero et al., 2018). In our systematic review, we will evaluate each study using 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 make our risk of bias assessment tool more relevant to exposure studies, we modified the EPA tool by incorporating criteria and considerations from Mandrioli et al. (2018), VanNoy et al. (2018), and Bero et al. (2018). Risk of bias domains in the modified tool include exposure measurement, participant selection, and analysis (Appendix E). The potential for a conflict of interest is also assessed for each study. The risk of bias tool was piloted using three studies (Fraser et al., 2013; Wu et al., 2014; Kim et al., 2019) in order to target the domains and ratings criteria most relevant to exposure studies. The results of this piloting are shown in both a table format (Appendix F) and a graphical format (Appendix G).

Two reviewers will independently assess the risk of bias for each study, giving a rating of “good,” “adequate,” or “deficient,” for each domain. A summary of criteria for rating each risk of bias domain as “good” is found in Table 4. Studies will be rated “good” for exposure media measurement if they report detailed and validated (i.e. QA/QC procedures) methods for measuring the PFAS chemical(s) in indoor media(s) of interest. Participant selection will be rated as “good” if the recruitment strategy is described in detail and there is minimal concern that selection bias occurred. A study will be rated “good” for analysis if they report the analysis methods and robustness of quantitative results in detail. A table describing the complete criteria for all ratings for risk of bias domains can be found in Appendix E.

Table 4.

Summary definitions of criteria for rating risk of bias domains as “good.”

Risk of bias domain	Summary criteria for “good” rating designation
Exposure measurement	Detailed and validated methods for measuring PFAS chemical(s) in indoor media(s) of interest and serum (or blood, plasma) are described, including QA/QC procedures and limit of detection.
Participant selection	Recruitment procedure is described and there is minimal concern for selection bias.
Analysis	Analysis methods and robustness of quantitative results are presented in detail, including appropriate confidence intervals, evaluation of variability, and handling of missing data.

Adapted from U.S. EPA’s Systematic Review Protocol for the PFBA, PFHxA, PFHxS, PFNA, and PFDA IRIS Assessments (draft, 2019), Mandrioli et al. (2018), VanNoy et al. (2018), and Bero et al. (2018).

Reviewers will discuss the tool and conduct a pilot exercise with several studies to ensure understanding and promote consistency in the use of the tool prior to commencement of the systematic review. Any discrepancies in ratings between the two reviewers will be resolved through discussion or by a third reviewer if needed. After all domain ratings are agreed upon for each study, the two reviewers will designate the risk of bias in the study as a whole as “high confidence,” “medium confidence,” “low confidence,” or “uninformative” (U.S. EPA, 2019). This overall study evaluation will be determined based on the reviewers’ judgement across the individual domain ratings and the likely impact that the domain ratings have on the study’s outcome results (U.S. EPA, 2019). The reviewers will record the reasoning for their overall study evaluations, and any discrepancies will be resolved through discussion. Summary figures for all risk of bias ratings and overall study evaluation will be included in the systematic review (Higgins et al., 2011).

2.7. Synthesis of Evidence

Summary measurement data will be grouped by the 8 targeted PFAS chemicals in the systematic review and then categorized into 6 indoor media types – indoor air, house dust, food packaging, household articles and materials, cleaning products, and personal care products. Separate analyses will be performed and reported for each PFAS chemical following the procedure described below.

Extracted or estimated mean PFAS concentrations from exposure media in each study will be used to calculate daily intakes for the indoor exposure route(s) from each media type using equations found in Lorber & Egeghy (2011). Daily intakes (ng/day) will be calculated by multiplying the mean concentration by pathway-specific and media-specific exposure factors for adults found in EPA’s Exposure Factors Handbook (U.S. EPA, 2011). Indoor media with more than one exposure route (e.g. dust ingestion and dust absorption) will be summed to determine a total daily intake for that indoor media matrix.

Following the calculation of daily intakes for all indoor media measured in the included studies, the PFAS concentration in serum (or blood, plasma) attributed to that exposure will be estimated using pharmacokinetic (PK) models. Then, the proportion of serum (or blood, plasma) PFAS concentrations that can be accounted for by indoor exposure pathways will be calculated as percent of the reported or estimated mean measured serum (or blood, plasma) concentrations from each respective study. These percent of measured serum (or blood, plasma) concentration results will be reported for each study, grouped by media type. If three or more studies measure the same media type and have similar study characteristics and context (e.g. location, population, and sample collection methodology), the mean and standard deviation of results within a media group will be reported. Both the means and standard deviations between media groups will be considered in a non-quantitative comparison to explore which indoor exposure pathways could be contributing most to serum (or blood, plasma) PFAS concentrations. The narrative summary of results will include comparisons between PFAS species by exposure media type, as well as between exposure media types by PFAS chemical species as the quantity of extracted data allows. Potential sensitivity analyses include further stratification by sex, age group, and location.

The synthesis of evidence procedure for calculating daily intakes, estimated serum concentrations, and percent of measured serum concentrations explained by exposure to PFAS in indoor media was piloted using three studies (Fraser et al., 2013; Wu et al., 2014; Kim et al., 2019) that reported PFOA measurements from both house dust and serum. A simple first-order PK model developed by Lorber & Egeghy (2011) that employs exposure factors from the EPA’s Exposure Factors Handbook (U.S. EPA, 2011) was used to estimate PFOA serum concentrations from exposure to the chemicals in house dust for these pilot studies. A summary of findings table reports the calculated daily intakes, estimated serum concentrations, and percent serum concentrations from this pilot exercise (Appendix H). All calculations are performed in R Statistical Computing Software (R Core Team, 2013).

The results from our pilot studies in Appendix H showed a large variation in the percent of measured PFOA serum concentrations that could be contributed to daily exposure through house dust, ranging from 0.27% (Kim et al. 2019) to 30.95% (Wu et al. 2014). The mean percent measured serum concentration attributed to dust exposure from all three house dust studies was 19.30% with standard deviation of 16.62%. From the study characteristics table (Appendix D), we can begin to offer an interpretation of these widely varying results between studies. The low percent measured PFOA serum concentration calculated from Kim et al. (2019) could be a function of geography, being the only study in our pilot from outside of the U.S., or a function of a relatively small sample size (13). The large difference between the concentrations of PFOA in house dust in the U.S.-based studies and the Korea-based study suggests that the geographic effect may be due to the use of varying household products or materials being in used. However, the mean reported serum PFOA concentration of the Kim et al. (2019) study is higher than that of the Wu et al. (2014) study, indicating that some exposure pathway other than house dust is largely contributing to serum concentrations in the Kim et al. (2019) participants.

2.8. Strength of Evidence Assessment

The strength of the evidence used to identify important indoor exposure pathways to PFAS in the systematic review will be judged using a modified version of a qualitative 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). The strength of evidence will be considered “moderate” at the beginning of the assessment (Johnson et al., 2014) and be subsequently considered to have “increased strength” or “decreased strength” based on several consideration categories including risk of bias across studies, consistency, strength (effect magnitude) and precision, and coherence. A “neutral” response can also be chosen where the strength of evidence is neither increased nor decreased. The consideration criteria for rating each category is shown in Appendix I. Two reviewers will independently judge the strength of evidence categories and designate the overall strength of evidence as “high”, “medium,” or “low.” Discrepancies between the reviewer’s judgements and overall rating will be resolved through discussion. Reasoning for each category judgement and the overall strength of evidence rating will be recorded in the systematic review.