Review: Characteristics of Spatial Distribution, Health Risk Assessment, and Regulation of PFAS in Global Drinking Water

We systematically reviewed original studies (January 1, 2000 to February 25, 2025) on PFAS in drinking water from PubMed and Web of Science using keywords including “PFAS” with “drinking water” or related terms. Studies were eligible if they provided original or summary data on PFAS concentrations in drinking water. Exclusion criteria were: 1) reporting only total PFAS without compound-specific concentrations, 2) omitting detection/quantitation limits while including non-detectable/non-quantifiable values, or 3) lacking both raw measurements and adequate summary statistics (defined as requiring either mean ± standard deviation or two or more percentiles). The review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (11). Data extracted included country, sampling date, sample size, target PFAS compounds, and concentrations. PFAS concentrations were aggregated nationally by compound, assuming a log-normal distribution.

Risk assessment followed the U.S. EPA’s environmental health risk assessment framework (12) and the Technical Guide for Environmental Health Risk Assessment of Chemical Exposure (WS/T 777-2021) (13) through four steps:

Evaluate potential harm of stressors to humans and ecosystems.

Assess non-carcinogenic risks by quantifying exposure–effect relationships using Formula (1). The reference dose [RfD, mg/(kg·d)] was derived from the U.S. Risk Assessment Information System (RAIS) (https://rais.ornl.gov/). The No Observed Adverse Effect Level [NOAEL, mg/(kg·d)] was used when available; otherwise, the Lowest Observed Adverse Effect Level (LOAEL) was applied. Uncertainty factors (UF_i) were incorporated.

Determine frequency, timing, and levels of contact with the stressor using Formula (2): ADD, average daily dose [mg/(kg·d)]; c, PFAS concentration (mg/L); IR, daily water intake (L/d). EF, exposure frequency (365 d/a); ED, exposure duration (1); BW, body weight (kg); AT, averaging time (d; calculated as EF×ED for chronic effects). We calculated the population exposure parameter $ BW\sim (59.96,4.16) $, $ \ln (IR)\sim N(6.50,0.82) $ based on age-stratified and general population data from the U.S. EPA Exposure Factors Handbook, assuming normal and log-normal distributions, respectively (14).

Calculate the hazard quotient (HQ, unitless), with HQ ≥1 indicating potential health risk (acceptable or low if <1).

We performed 10,000 Monte Carlo simulations to estimate HQ values at the 50th and 95th percentiles using probabilistic risk quotient methodology.

A total of 122 studies from 37 countries across six continents were included by searching the PubMed and Web of Science databases (Figure 1). Among 5,600 water samples analyzed, 102 PFAS compounds were detected (Supplementary Table S1). Figure 2A classifies PFAS into high-concern (>20 studies) and low-concern (≤20 studies) compounds with ≥30% detection rates. PFOA and PFOS received the highest research attention (102 and 100 studies, respectively) and showed the highest detection frequencies (64.69% and 60.72%) (Figure 2A).

Figure 1. 

Literature screening.

Figure 2. 

Characteristics of spatial distribution and risk assessment of PFAS. (A) Detection rates and regulatory status of PFAS; (B) Exposure in point-source pollution and background pollution; (C) Risk assessment.

Abbreviations: PFTeDA=perfluorotetradecanoic acid; 8:2 FTCA=8:2 fluorotelomer carboxylic acid; HFPO-TA=hexafluoropropylene oxide trimer acid; PFUnS=perfluoroundecanesulfonic acid; PFDDA=perfluorododecanedioic acid; TFMS=trifluorome-thanesulfonic acid; PEPA=perfluorinated ether phosphonic acid; TFA=trifluoroacetic acid; PFO2HxA=perfluoro(3,5-dioxahexanoic) acid; PFMOAA=perfluoro-2-methoxyacetic acid; NVHOS=1,1,2,2-tetrafluoro-2-(1,2,2,2-tetrafluoroethoxy) ethane sulfonate; F3-MSA=trifluoromethane sulfonic acid; PFO3OA=perfluoro(3,5,7-trioxaoctanoic) acid; PFBuS=perfluorobutanesulfonic acid; PFO4DA=perfluoro(3,5,7,9-butaoxadecanoic) acid; EtFOSE=N-Ethylperfluorooc tane sulfonamidoethanol; PFPrS=perfluoropropanesulfonic acid.

The study areas were categorized into background contamination zones (104 studies) and point-source zones (18 studies, including contamination from fluorochemical plants, firefighting training areas, paper, textile, and leather industries, or oil and gas-producing regions). Contamination patterns were characterized by nine high-priority PFAS detected in both categories: PFOA, PFOS, PFHxS, PFNA, perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluorobutanesulfonic acid (PFBS), and perfluoroheptanoic acid (PFHpA).

In background contamination zones, research has primarily focused on Asia (particularly China), North America (notably the United States), and parts of Europe. Sixteen countries provided complete concentration data for all nine PFAS (Figure 2B), with the highest levels in the Republic of Korea (26.20 ng/L), the United States (14.34 ng/L), China (13.43 ng/L), and France (13.21 ng/L). In China, the compositional profile was PFBA (67.27%) > PFOA (15.20%) > PFPeA (5.23%) > PFOS (4.26%) (Figure 2B).

In point-source zones, peak geometric mean concentrations were observed in Japan (PFOA, 855.62 ng/L; PFHxA, 46.50 ng/L; PFHpA, 13.52 ng/L; PFNA, 8.39 ng/L), Ghana (PFOS, 86.33 ng/L), China (PFBA, 27.81 ng/L; PFPeA, 3.77 ng/L; PFBS, 7.41 ng/L), and Sweden (PFHxS, 12.24 ng/L). PFOA dominated compositional profiles in China (40.77%) and Pakistan (69.37%), while PFBS was dominant in the United States (18.01%) and the Netherlands (23.26%) (Figure 2B).

China, the Netherlands, the United States, and Burkina Faso reported all nine high-priority PFAS in both background and point-source zones. The mean (range) total concentrations of these nine PFAS across these four countries were 13.25 (1.74–29.20) ng/L in background zones and 30.11 (5.46–83.66) ng/L in point-source zones. As shown in Figure 2B, PFOA, PFBA, and PFBS were dominant in point-source zones, whereas PFBA predominated in background zones.

The HQ values for PFHxA, PFBS, and PFBA were below 1, indicating acceptable health risks. For PFHxS, PFOA, PFOS, and PFNA, the HQ P₅₀ values were 10.30, 0.33, 0.07, and 0.001, respectively, while the HQ P₉₅ values were 698.72, 9.58, 3.30, and 0.03, respectively. The contribution to overall human health risk ranked as follows: PFHxS (80.63%), PFOA (28.01%), PFOS (12.95%), and PFNA (0.07%) (Figure 2C).

The World Health Organization (WHO) recommends localized standards based on actual needs and resources, with regular reviews and timely updates (15). Analysis of regulatory frameworks in several countries (Supplementary Table S2) revealed two major trends: First, PFOA and PFOS remain the primary targets of regulation, with increasingly stringent limits reflecting scientific consensus on their risks even at very low concentrations. Second, regulation is shifting from single-compound limits to combined PFAS limits, broadening the scope of oversight.

Research on PFAS exposure in drinking water is concentrated in the United States, China, and parts of the European Union, with limited studies in most developing countries due to technological, infrastructural, or funding constraints (16). We identified 102 PFAS in drinking water, with significant disparities in research output across compounds (Figure 2A). These differences may reflect variations in usage, environmental persistence, and toxicity. Demand for data on PFAS exposure, toxicity, and population health effects has driven advances in testing technology, which, in turn, facilitates further research. This feedback loop reinforces focus on high-priority PFAS while potentially neglecting others. Notably, low-priority PFAS such as hexafluoropropylene oxide dimer acid, perfluorobutanesulfonamide, and Perfluoropropanesulfonate — detected in ≥30% of samples (n≥400) but currently unregulated (Figure 2A) — require urgent investigation.

Our risk assessment indicates negligible health risks from PFHxA, PFBS, and PFBA, but highlights potential hazards from PFHxS, PFOA, PFOS, and PFNA, ranked as PFHxS > PFOA > PFOS > PFNA. These findings align with previous studies by Thomaidi et al. (10) and Li et al. (17), which identified PFOA and PFOS as significant contributors to global and Chinese drinking water risks. The RfDs used in this study integrate comprehensive toxicological data: PFOA at 3×10⁻⁸ ng/L (pediatric vaccine response, birth weight, adult cholesterol), PFOS at 1×10⁻⁷ ng/L (immune, developmental, cardiovascular, and hepatic effects), PFHxS at 4×10⁻⁸ ng/L (immunotoxic and thyroid effects), and PFNA at 2×10⁻⁹ ng/L (immunotoxic and developmental effects). These precautionary thresholds underscore the need for cautious interpretation of risk estimates.

As toxicological and epidemiological evidence grows, regulatory standards for PFAS in drinking water are becoming more stringent worldwide. However, current Chinese standards for PFOA and PFOS — based solely on developmental endpoints such as reduced osteogenesis and altered puberty in juvenile rodents (18–19) — remain comparatively lenient. In contrast, the U.S. EPA’s 2024 Primary Drinking Water Regulations (PDWR) set a maximum containment level of 4 ng/L for both PFOA and PFOS, based on RfD values (3×10⁻⁸ ng/L for PFOA and 1×10⁻⁷ ng/L for PFOS) derived from multiple endpoints, including immunotoxicity, developmental, hepatic, and cardiovascular effects (20–21). In China, PFOA and PFOS are currently only reference indicators in GB5749-2022 and are not included in routine national monitoring. Most PFAS data derive from small-scale studies, limiting representativeness. Enhancing local exposure data, advancing mechanistic toxicology, and adopting a risk-based, multi-endpoint dose–response approach similar to the U.S. EPA’s framework are essential to support phased standard updates.

This study has limitations. First, variability in the PFAS compounds analyzed across studies limits global comparability of total PFAS exposure. Moreover, emerging contaminant surveys often target suspected contamination zones — even when classified as background — potentially inflating exposure estimates. Second, reliance on self-reported point-source contamination data from primary literature means unreported contamination cannot be excluded. Third, uniform assumptions applied across populations ignore physiological and lifestyle differences due to a lack of region-specific toxicity and exposure data. Finally, heterogeneity in sampling, pretreatment, analytical methods, and quality control across the 122 studies likely contributes to variability (22). Thus, results should be interpreted with caution.

Drinking water safety has become an urgent global health concern (23). Despite these limitations, our findings offer meaningful insights for PFAS management: First, stricter regulatory limits for PFOA and PFOS are needed, incorporating multi-system toxicity endpoints, population-specific exposure factors, technical feasibility, and cost considerations, alongside enhanced monitoring in point-source areas. Second, regulatory expansion to include PFHxS and PFNA, either as individual limits or under a combined standard, should be considered. Implementation of these recommendations requires more comprehensive, targeted exposure assessments and health risk studies. Furthermore, while our analysis focuses on drinking water as an exposure pathway to inform PFAS standards, future high-quality research should address combined risks from diet, inhalation, and dermal contact.

The School of Public Health and the Center for Public Health and Epidemic Preparedness and Response of Peking University for hosting academic visits; and the support from the Hebei Provincial Administration of Disease Control and Prevention and the Shijiazhuang Municipal Center for Disease Control and Prevention.