Introduction

A recent study analyzed the hotspot evolution and research trends of a variety of emerging pollutants over the past 20 years and found that perfluoroalkyl and polyfluoroalkyl substances (PFAS) are among currently the most important emerging pollutants1. PFAS are organic pollutants. These chemicals are lipophilic, hydrophilic, and thermochemically stable2. The chemical stability of single-carbon and carbon–fluorine bonds makes them difficult to degrade in the environment3. They are commonly used in industrial and commercial processes as organic surfactants in electronics, surface coatings, and firefighting foams4. Among the various PFAS that have been identified so far, National Health and Nutrition Examination Survey (NHANES) regularly detects the concentrations of more than 10 serum PFAS (containing isomers) in American household populations, including perfluorooctanoic acid (PFOA), perfluorononaic acid (PFNA), perfluorodecanoic acid (PFDA), perfluoroundecanoic acid (PFUnDA), perfluorohexane sulfonic acid (PFHxS), and perfluorooctane sulfonic acid (PFOS) on a 2-year cycle.

As an important part of the justice movement, the environmental justice theory proposes that there are racial and class differences in poison damage5. Compared to other groups, residents in poor areas tend to be more exposed to environmental pollution, making them more vulnerable to the adverse effects of the pollutants. Perhaps unequal exposure to toxic substances is not only a matter of distance from the pollutant sources but also involves factors such as community, race, industrial change, class composition, and suburbanization6,7. However, Glynn et al.’s study showed the opposite conclusion, showing that participants born in high-income countries had higher serum concentrations of PFAS than participants born in other income levels8. Interestingly, current research on PFAS and population health has focused on vulnerable populations, such as children9,10, adolescents10,11, pregnant women12,13, and older adults14. In addition, it is easy to overlook that people who suffer from a chronic disease or even suffer from multiple chronic diseases are also vulnerable to environmental pollutants, and they are often considered to be more vulnerable to the negative effects of environmental pollutants due to their reduced resistance to diseases15. Hypertension and hyperlipidemia are often associated and occur simultaneously. Comorbidities (≥ 2 chronic diseases) are common among middle-aged and older patients in community health centers, with the most common patterns being hyperlipidemia-hypertension16. Comorbidities are often associated with greater disability and a higher treatment burden. Simultaneously, adults with complex and sick patterns have difficulty in performing self-management tasks. However, only a few studies have focused on this topic.

Obesity is closely related to the economy. The economic burden of obesity is enormous17. Previous research has shown an association between higher socioeconomic levels in adulthood and lower levels of fat in high-income countries. However, an association in the opposite direction has only been observed in middle-income countries18. PFAS have a long half-life and can easily cross the placental barrier from the pregnant woman to the fetus19,20,21. There is evidence that prenatal exposure of a pregnant woman is positively associated with PFAS and the body mass index (BMI) or waist circumference of her child during adolescence22,23. Therefore, we hypothesized that there would be a positive association between serum PFAS concentration and obesity. Moreover, owing to the close relationship between obesity and the economy, obesity may have a certain degree of mediating effect on the process of influencing serum PFAS concentrations by economic level.

In this study, we aimed to measure the association between serum PFAS concentration and the ratio of family income to poverty (PIR) in hypertensive and hyperlipidemic populations and to assess the potential influence of sex and age on these associations. Simultaneously, we established a mediating effect model to explore the mediating effect of BMI and waist circumference on the possible influence of PIR on serum PFAS concentration, to further explore the internal reasons for the possible influence of socioeconomic status on serum PFAS concentration.

Methods

Research design and population

This study summarizes the cross-sectional data from the 2015 to 2018 NHANES cycles. The National Health and Nutrition Examination Survey (NHANES) is a population-based national survey conducted every 2 years that measures the health and nutritional status of non-institutionalized adults, civilians, children, and adolescents in the United States. The Centers for Disease Control and Prevention have described the detailed research procedures for NHANES24. To avoid growth and development disturbances, we included only survey participants aged over 20 years while screening for complete data on serum PFAS concentrations, PIR (defined as the ratio of a family's self-reported income to the appropriate poverty line for a family25), and covariates (see the Covariates section for details). Based on the study design, we excluded missing information and age-ineligible participants and included 2665 adults. Further details are shown in Fig. 1.

Fig. 1
figure 1

Study design flowchart and reasons for excluded participants.

Full size image

Population grouping and disease definition

The participants were divided into four groups based on their health status: no disease (without hypertension and hyperlipidemia), hypertension (with hypertension but without hyperlipidemia), hyperlipidemia (with hyperlipidemia but without hypertension), and comorbidities (with hypertension and hyperlipidemia). The presence of one of the following conditions was considered hypertension: systolic blood pressure > 140 mmHg, diastolic blood pressure > 90 mmHg, or self-reported current use of blood pressure medications. The presence of one of the following conditions is considered to be hyperlipidemia: TC ≥ 5.18 mmol/L or HDL-C ≤ 1.04 mmol/L or use of high cholesterol drugs.

Measurement of PFAS in serum

PFAS were quantitatively detected using online solid-phase extraction coupled with high-performance liquid chromatography-turboionspray ionization-tandem mass spectrometry (SPE-HPLC-TIS-MS/MS). Six common PFAS were used as outcome measures: PFOA, PFNA, PFDA, PFUnDA, PFHxS, and PFOS. The PFOA and PFOS concentrations were calculated as the sum of their isomer concentrations. In each cycle, concentrations below the detection limit were replaced by the detection limit divided by the square root of 2. Detailed quality control information is available from the NHANES and Nutrition Examination Survey Laboratory Procedure Manual.

Covariates and mediators

Demographic covariates included in this study were obtained through a report questionnaire, including age (continuous), sex (male or female), race or ethnicity (Mexican American, Other Hispanic, Non-Hispanic White, Non-Hispanic Black, or Other Race-Including Multi-Racial), Six month time period (November 1 through April 30 or May 1 through October 31), and adult educational level (below secondary school, secondary school, high school graduate, college degree, and college graduate or above). Physical parameters such as height, weight, and waist circumference were measured by trained health technicians. The mediating variables included in this study were BMI and waist circumference, which reflect the obesity status of the participants.

Statistical analysis

Descriptive statistics were calculated for serum PFAS concentrations in the participants divided into four groups based on their health status. The Kruskal–Wallis test was used to determine whether participants in the four groups had the same serum PFAS concentrations. The concentrations between the two groups were further compared based on the average rank of the samples in each analysis group. Multivariate linear regression models were used to examine the relationship between the PIR and individual serum PFAS concentrations in patients with different diseases. After logarithmic conversion of the independent and dependent variables, the results of the regression analysis were expressed as the percentage change in serum PFAS concentration, which may have been caused by a 1% increase in PIR. For the overall analysis, we fitted 24 regression models, each combining four different health conditions and six different PFAS. The independent variables used in each model were sex (continuous variable), age (continuous variable), PIR (continuous variable), race (categorical variable), inspection time (categorical variable), and educational level (categorical variable). Subsequently, to explore the potential effects of different gender and age, we conducted a subgroup analysis, which was classified into four subgroups: males, females, older adults (≥ 60 years old), and young adults (20–59 years old). Similarly, each subgroup was fitted to 24 regression models, each with independent variables similar to those used in the overall analysis. Simultaneously, models with three different covariates were fitted (no common covariates were included, some common covariates were included, and all common covariates were included) to test the influence of model covariates on the regression results.

Mediation analysis is an ideal strategy for elucidating the ways in which independent variables are used against dependent variables and for elucidating statistical evidence for mechanism analysis26,27. In this study, we quantitatively calculated the mediating effect of obesity indicators (BMI or waist circumference) in the regression process through a mediating analysis. The direct effect represents the direct relationship between the PIR and serum PFAS concentrations in different groups of patients. Indirect effects represent the relationship between the PIR mediated by obesity (BMI or waist circumference) and serum PFAS concentrations. The mediated proportion represents the percentage of mediating effects. The combined effect was expressed as the sum of the direct and indirect effects. These can be represented statistically by a set of linear regressions.

$$ Y_{i} = \, \alpha_{1} + \, cX_{i} + \, \beta_{1} Z_{i} + \, \varepsilon_{i} $$
(1)
$$ M_{i} = \, \alpha_{2} + \, aX_{i} + \, \beta_{2} Z_{i} + \, \varepsilon_{i} $$
(2)
$$ Y_{i} = \, \alpha_{3} + \, bM_{i} + \, cX_{i} + \, \beta_{3} Z_{i} + \, \varepsilon_{i} $$
(3)

Estimates of indirect effects rely on the values of the remaining two independent regression models, so the results of significance tests and standard error can be obtained by bootstrapping. Therefore, bootstrapping with 1000 replications was performed for each expected mediating variable and PFAS concentration, controlling for confounding factors, to examine the mediating effect9.

A sensitivity analysis was used to further evaluate the robustness of the regression models. Participants with extreme PFAS concentrations of 1% were excluded from each fitted model. In the remaining participants, the association between the PIR and serum PFAS concentrations was assessed again. In this study, R 4.3.1 software was used for statistical analysis. P < 0.05 was considered statistically significant.

Results

Participant characteristics

As shown in Table 1, a total of 2665 participants were included in this study. Of these, 789 (29.6%) did not have hypertension or hyperlipidemia (no disease group), 224 (8.4%) had hypertension but no hyperlipidemia (hypertension group), 909 (34.1%) had hyperlipidemia but no hypertension (hyperlipidemia group), and 743 (27.9%) had hypertension and hyperlipidemia (comorbidity group). The Kruskal–Wallis test results indicated that the concentrations of PFAS in the four groups were different. Subsequently, according to the analysis of the average rank of PFAS concentrations in each group of samples, the concentrations of PFAS in the serum were the highest in the comorbidity group. For PFOA, PFNA, and PFHxS, there were no statistically significant differences in serum concentrations between participants in the hypertension and hyperlipidemia groups; however, they were higher than those in the no-disease group. For PFDA and PFUnDA, there were no statistically significant differences in serum concentrations among the no disease, hypertension, and hyperlipidemia groups. Serum PFOS concentrations in the hypertension group were higher than those in the hyperlipidemia group, and both were higher than those in the no disease group.

Table 1 The median concentration of PFAS (ng/mL) in the serum of the participants.
Full size table

Correlation analysis

The results of the pairwise spearman correlation coefficient showed that the correlation coefficient between the selected PFAS ranged from 0.136 to 0.756 (Fig. 2).

Fig. 2
figure 2

Correlation analysis of PFAS.

Full size image

Multiple linear regression models

Overall analysis

As shown in Table 2, serum PFAS concentrations were positively correlated with the PIR in the no disease, hyperlipidemia, and comorbidity groups. More specifically, for every 1% increase in PIR in the no-disease group, serum concentrations of PFOA, PFNA, and PFHxS increased by 0.10% (95%CI 0.04–0.16%), 0.08% (95%CI 0.01–0.16%), and 0.11% (95%CI 0.03–0.18%), respectively. The serum concentrations of PFOA, PFNA, PFDA, PFUnDA, and PFHxS increased by 0.10% (95%CI 0.04–0.15%), 0.09% (95%CI 0.02–0.16%), 0.10% (95%CI 0.03–0.17%), 0.08% (95%CI 0.01–0.15%), and 0.09% (95%CI 0.02–0.15%), respectively, when PIR increased by 1% in hyperlipidemia group. In comorbidity group, the serum concentrations of PFOA, PFNA, PFDA, PFUnDA, PFHxS, and PFOS increased by 0.09% (95%CI 0.03–0.15%), 0.07% (95%CI 0.01–0.14%), 0.13% (95%CI 0.04–0.21%), 0.12% (95%CI 0.04–0.19%), 0.10% (95%CI 0.02–0.18%), and 0.11% (95%CI 0.03–0.19%), respectively, for every 1% increase in PIR.

Table 2 After adjusting for age, sex, race, survey time, and education level, the percentage change in the concentration of PFAS as the income poverty ratio increased by 1%.
Full size table

Regression models of different covariates

As shown in Fig. 3, three linear regression models were established using different covariates. Associations between the PIR and serum PFAS concentrations in the no disease, hyperlipidemia, and comorbidity groups were observed in all three models. More specifically, for model 1, serum perfluorooctanoic acid concentrations increased by 0.16% (95%CI 0.10–0.21%) for every 1% increase in PIR in the disease-free population. In Models 2 and 3, the effect size decreased from 0.16 to 0.15% (95%CI 0.09–0.20%) and 0.10% (95%CI 0.04–0.16%), respectively. For more detailed data, see Tables A1 and A2 in the supplementary files.

Fig. 3
figure 3

Percentage change in serum PFAS with each 1% increase in PIR, considering the different covariate models. PFOA, perfluorooctanoic acid; PFNA, perfluorononaic acid; PFDA, perfluorodecanoic acid; PFUnDA, perfluoroundecanoic acid; PFHxS, perfluorohexane sulfonic acid; PFOS, perfluorooctane sulfonic acid; PIR, the ratio of family income to poverty. Model 1 did not adjust for confounding factors, Model 2 adjusted for age and sex, and Model 3 adjusted for age, sex, race, survey time, and education level.

Full size image

Gender difference

Association between income and PFAS concentrations in different sexes and ages: Serum PFAS concentrations were positively correlated with PIR in men in the no-disease and comorbidity groups. In the no-disease group, for every 1% increase in the PIR in men, the serum concentration of PFHxS increased by 0.11% (95%CI 0.01–0.22%). In comorbidity group, for every 1% increase in PIR in men, the serum concentrations of PFOA, PFNA, PFDA, PFUnDA, PFHxS, and PFOS increased by 0.16% (95%CI 0.07–0.24%), 0.13% (95%CI 0.04–0.22%), 0.17% (95%CI 0.05–0.28%), 0.16% (95%CI 0.05–0.27%), 0.18% (95%CI 0.06–0.29%), and 0.21% (95%CI 0.10–0.32%), respectively.

Serum PFAS concentrations were positively correlated with the PIR in women in the no disease and hyperlipidemia groups. In no disease group, for every 1% increase in PIR in women, the serum concentration of PFOA increased by 0.11% (95%CI 0.03–0.20%). In the hyperlipidemia group, for every 1% increase in the PIR in women, the serum concentrations of PFOA and PFDA increased by 0.11% (95%CI 0.03–0.19%) and 0.13% (95%CI 0.02–0.23%), respectively.

Age difference

A positive correlation between serum PFAS concentrations and PIR was observed in young adults in the no disease, hypertension, hyperlipidemia, and comorbidity groups. In no disease group, for every 1% increase in PIR in young adults, the serum concentrations of PFOA, PFNA, and PFHxS increased by 0.12% (95%CI 0.06–0.18%), 0.10% (95%CI 0.03–0.18%), and 0.10% (95%CI 0.02–0.18%), respectively. In hypertension group, for every 1% increase in PIR in young adults, the serum concentration of PFDA increased by 0.23% (95%CI 0.03–0.44%). In hyperlipemia group, for every 1% increase in PIR in young adults, the serum concentrations of PFOA, PFNA, PFDA, PFHxS, and PFOS increased by 0.10% (95%CI 0.03–0.16%), 0.09% (95%CI 0.01–0.17%), 0.10% (95%CI 0.02–0.18%), 0.22% (95%CI 0.10–0.35%), and 0.23% (95%CI 0.10–0.36%), respectively. In comorbidity group, for every 1% increase in PIR in young adults, the serum concentrations of PFOA, PFNA, PFDA, PFHxS, and PFOS increased by 0.19% (95%CI 0.09–0.29%), 0.13% (95%CI 0.02–0.24%), 0.16% (95%CI 0.03–0.29%), 0.22% (95%CI 0.10–0.35%), and 0.23% (95%CI 0.10–0.36%), respectively.

A positive correlation between serum PFAS concentration and PIR was observed in older adults in the no disease, hyperlipidemia, and comorbidity groups. In no disease group, for every 1% increase in PIR in the older adults, the serum concentration of PFHxS increased by 0.30% (95%CI 0.03–0.57%). In the hyperlipidemia group, for every 1% increase in the PIR in the older adults, the serum concentrations of PFOA, PFNA, PFDA, and PFUnDA increased by 0.14% (95%CI 0.04–0.23%), 0.17% (95%CI 0.04–0.29%), 0.17% (95%CI 0.02–0.31%), and 0.16% (95%CI 0.02–0.31%), respectively. In comorbidity group, for every 1% increase in PIR in the older adults, the serum concentrations of PFDA and PFUnDA increased by 0.11% (95%CI 0.01–0.21%) and 0.14% (95%CI 0.04–0.24%), respectively.

Mediating effect

Mediation analyses were performed to examine whether the association between PIR and PFOA and PFNA was mediated by BMI (Figs. 4 and A1). In hyperlipidemia group, BMI was found to mediate 21.5% of the association between PIR and PFOA, and 11.7% of the association between PIR and PFNA. However, the significant mediating role of BMI was not observed in either no disease group, hypertension group or comorbidities group. Subsequently, an empirical test of the mediating effect was conducted (Table A3).

Fig. 4
figure 4

Mediating effects of obesity. *P < 0.05. **P < 0.01. ***P < 0.001. (A) represented no disease group, (B) represented hypertension group, (C) represented hyperlipidemia group, (D) represented comorbidities group. PIR, the ratio of family income to poverty. BMI, body mass index; PFOA, perfluorooctanoic acid.

Full size image

Sensitivity analysis

As shown in Table 3, the results remained robust in the sensitivity analysis. Specifically, the statistical significance of only 1 outcome changed from that of the 24 regression models fitted to the overall analysis. In the sensitivity analysis, the increase in serum PFDA concentration in the hypertension group changed from 0.14% (95%CI 0.00–0.29%) to 0.15% (95%CI 0.01–0.30%) with a 1% increase in the PIR.

Table 3 Sensitivity analysis of the overall analysis.
Full size table

Discussion

This study, which included 2665 U.S. adults, aimed to investigate the association between income and the emerging environmental pollutant PFAS in the serum of participants with hypertension and hyperlipidemia, to explore the potential effects of age and gender, and to analyze the mediating effect of obesity on income and PFAS concentrations.

This study observed that in people without hypertension and hyperlipidemia, serum PFAS concentrations increased by 0.06–0.16%, with a 1% increase in the income-poverty ratio. This finding is similar to that of previous studies. Using the NHANES, Jessica Tyrrell et al. observed a positive association between polyfluorinated compounds (PFOA, PFOS, and PFOA) and socioeconomic status in adults28. A meta-analysis showed that higher income was associated with higher PFAS concentrations, with an average increase in PFAS concentrations ranging from 10 to 14% when the household income of the study population doubled29. For PFOA, serum concentrations of PFAS were 21% (95%CI − 26 to − 14%) lower in the lowest quartile of adjusted household income than in the highest quartile30. However, a study that analyzed the relationship between serum PFAS concentrations and monthly income in 319 participants from suburban Seoul (Yeongseong and Suwon), Yeshan, and Busan in South Korea between 2006 and 2007 found no such association31. This may be due to differences in sample size, sampling location, and exposure history.

Perhaps, there are several studies that can be used to explain the relationship between income status and PFAS. Diet is a major source of human exposure to PFAS, with individuals of higher socioeconomic status tending to consume more fresh meat, fish, and vegetables, which are potential sources of PFOA32. Even within the same high-income countries, the underlying factors contributing to this growth may differ between the United States and Europe. In the United States, beef consumption may be the culprit, whereas in Europe, fish consumption may play a major role. However, differences in diet do not fully explain the potential for differences in PFAS concentrations in groups with different income levels. In addition, perfluorinated compounds tend to be found in expensive waterproof fabrics, which may explain their association with higher socioeconomic status. The serum concentrations of polyfluorinated compounds are also correlated with the levels of polyfluorinated compounds in office air33. People with a higher social status are more likely to work in white-collar offices, and higher exposure may explain the existence of a positive association to some extent. To uncover the relationship between social, economic, environmental, and health factors, and reduce human health inequalities, it is necessary to study the root causes of differences in PFAS exposure due to income levels. With this knowledge, we can raise awareness and encourage people to make informed choices to reduce their daily exposure to environmental pollutants.

In this study, the concentrations of PFOA, PFNA, PFHxS, and PFOS were higher in participants with hypertension than in those without hypertension. Among participants with hypertension, men had higher concentrations of PFOA, PFHxS, and PFOS than women. This is similar to the results of a recently published systematic review34 that included three cohort studies, two case–control studies, and eight cross-sectional studies for a meta-analysis evaluating the association between PFAS and hypertension. The study found that PFNA (OR = 1.11, 95%CI 1.04–1.19), PFOA (OR = 1.12, 95%CI 1.02–1.23), PFOS (OR = 1.19, 95%CI 1.06–1.34), and PFHxS (OR = 1.03, 95%CI 1.00–1.06) were significantly associated with hypertension. In addition, this study proposes that a clear gender difference is observed in individuals exposed to PFAS. Specifically, men exposed to PFNA, PFOA, and PFOS had a higher risk of developing hypertension than women. However, the underlying mechanism between PFAS exposure and hypertension remains poorly understood, and larger, higher-quality epidemiological studies, as well as in vivo and in vitro experiments, are needed to further elucidate this relationship. In addition to hypertension, PFAS may be associated with hyperlipidemia. Aker et al.35 examined the relationship between perfluoroalkyl acid mixtures and several cardiometabolic health outcomes and found that for every one-quarter increase in perfluoroalkyl acid mixtures, the total cholesterol, LDL, and HDL levels increased. The long-chain homologues PFNA, PFDA, and PFUnDA may be major contributors to lipid changes in vivo.

To our knowledge, this is the first study to report the potential mediating role of obesity in the relationship between income status and concentrations of PFAS in vivo. Our study also found that BMI may play a mediating role in the association of income status with concentrations of PFAS in vivo in hyperlipidemia group. There may be several studies that can be used to explain the mediating role of obesity. There may be several studies that can be used to explain the mediating role of obesity. On the one hand, studies have confirmed a significant negative correlation between obesity and PFAS36. On the other hand, BMI and income levels in adulthood are inversely related in high-income countries18. However, the mediating effect of obesity was not significant in people with hypertension. There is currently no plausible explanation for this result, which may be due to the positive association between PFAS and hyperlipidemia, but not the significant association between PFAS and hypertension36.

As emerging pollutants, PFAS have attracted considerable research attention in recent years, and the number of related studies is quite large. However, compared with the relatively concentrated focus on pregnant women and newborns in the field, the attention paid to adults with chronic diseases, such as hypertension and hyperlipidemia, is insufficient. Therefore, as a relatively rare study, this study examined the association between economic status and PFAS concentration in the disease population and attempted to reveal the potential influence of sex and age while exploring the mediating effect of obesity in the association using mediating models. A major advantage of this study is its rigorous sampling design, weighting scheme, and quality control based on NHANES. This allowed the study to obtain nationally representative cross-sectional data with large sample sizes to answer multiple association hypothesis questions of interest covering multiple subgroups. Our findings are robust and consistent across multiple sensitivity analyses.

This study has several limitations. First, the cross-sectional study design made it impossible to determine the timeliness of exposure and establish a causal relationship between the household income-poverty ratio and serum PFAS concentrations. Although we adjusted for as many demographic covariates and confounding variables as possible in our regression analysis, the effects of other unmeasured confounding variables could not be excluded. Finally, this study analyzed only six commonly investigated persistent PFAS in the population. Because the NHANES had little data on this, we were unable to assess exposure to other short chains and other PFAS compounds, which requires further research; however, our study provides a basis for future studies focusing on socioeconomic status and PFAS exposure in subhealthy chronic disease populations. Although our study revealed a significant positive association between household income and PFAS exposure in a vulnerable population with hyperlipidemia and comorbidities, our understanding remains incomplete due to existing studies. Further large longitudinal and experimental studies are needed to determine whether these associations are causal and to elucidate their potential mechanisms of action.

Conclusion

In summary, we observed a positive association between the PIR and serum PFAS levels in American adults with hyperlipidemia and comorbidities. Simultaneously, the potential effects of age and sex should not be ignored. In patients with hyperlipidemia, BMI significantly mediated the effect of the income-poverty ratio on PFOA. These results are similar to those of previous studies showing that income in the general population is positively correlated with PFAS concentrations. The environmental justice hypothesis states that groups with lower socioeconomic status tend to be more exposed to environmental pollutants. These studies clearly suggest that the environmental justice hypothesis is not a good description of PFAS exposure patterns. Given the widespread exposure, long half-lives, and health risks of PFAS, these findings may have important implications for interventions aimed at reducing their physical burden. Further prospective and experimental studies are needed to confirm or rule out a causal relationship between income and PFAS concentration in individuals with hypertension or hyperlipidemia and to investigate the mechanisms involved. This is an urgent and emerging topic in the fields of clinical medicine, public health, environmental science, and economics.