Environmental exposure to perchlorate, nitrate, and thiocyanate in relation to chronic kidney disease in the general US population, NHANES 2005–2016 : Chinese Medical Journal

Secondary Logo

Journal Logo

Original Article

Environmental exposure to perchlorate, nitrate, and thiocyanate in relation to chronic kidney disease in the general US population, NHANES 2005–2016

Li, Wei1; Wu, Hong2; Xu, Xuewen1; Zhang, Yange1

Editor(s): Li, Jinjiao; Ji, Yuanyuan

Author Information
Chinese Medical Journal ():10.1097/CM9.0000000000002586, April 26, 2023. | DOI: 10.1097/CM9.0000000000002586



Perchlorate is an artificial or naturally occurring contaminant that is commonly used to produce rocket fuel, explosives, highway safety flares, fireworks, and explosives.[1] Perchlorate is pre-dominately found in drinking water and various types of foods.[2,3] Nitrate is widely present in leafy vegetables, processed meats, and contaminated water.[4] For the non-occupationally exposed population, cigarette smoking is likely the main source of exposure to thiocyanate, which is formed during the process of detoxification of the hydrogen cyanide contained in cigarettes.[5] Coal gasification and gold mining also produce thiocyanate in wastewater.[6] Ubiquitous perchlorate, nitrate, and thiocyanate (PNT) in the environment mainly act via the same mechanism. PNT can cause iodide uptake inhibition at the sodium iodide symporter, which functions to transport iodide to the thyroid in vertebrates. Thus, PNT could lead to decreased production of thyroid hormone and could ultimately cause hypothyroidism after prolonged inhibition of iodide uptake.[7-9] Urinary levels of PNT are widely used as biomarkers to evaluate exposure levels in previous studies.[10,11]

Adverse effects of PNT on thyroid function, liver function, obesity, and cancer development have been observed in epidemiological and animal studies.[5,11-13] For instance, some studies suggest that oxidative stress in liver mitochondria can be induced by perchlorate exposure.[14] Perchlorate might also interfere with insulin secretion by affecting calcium channels.[15] Higher urinary perchlorate levels have been found to be associated with an increased incidence of diabetes mellitus.[10] However, the beneficial roles of PNT have also been reported. For example, some studies have demonstrated that dietary nitrate might have a cardiovascular protective effect.[16,17] However, to date, no epidemiological studies have illustrated clear associations between PNT exposure and renal function in the general population.

Chronic kidney disease (CKD) is associated with a substantial public health and economic burden worldwide with its global prevalence of approximately 10%.[18] Except for clinical risk factors such as diabetes and hypertension, some environmental chemicals are also related to the incidence of CKD, including heavy metals, organophosphate esters, and environmental phenols.[19-21] To test the hypothesis that exposure to PNT is also significantly associated with renal function and the prevalence of CKD, we conducted a national cross-sectional study using data from the National Health and Nutrition Examination Survey (NHANES). In this study, we analyzed both linear and non-linear associations of urinary PNT with renal function, as well as the presence of CKD.


Study population

The NHANES, which is a cross-sectional, nationwide study aiming to evaluate the health and nutrition status of the general population in the United States (US), has been conducted since 1960 (https://www.cdc.gov/nchs/nhanes/). All NHANES surveys have been approved by the National Center for Health Statistics Research Ethics Review Board. Informed consent forms were signed by all survey participants. The data used in the present study were publicly accessible on the NHANES website. Based on a multi-stage, stratified, clustered sampling strategy, NHANES enrolls around 5000 participants each year. Our study used data from 70,191 participants enrolled in NHANES 2005 to 2016. We excluded participants aged <20 years (N = 30,441) and pregnant women (N = 708). Participants with missing data of PNT (N = 22,323), covariates (N = 2670), or outcomes (N = 676) were also excluded. Finally, 13,373 participants were included in the present analyses [Figure 1].

Figure 1:
Flowchart of subjects meeting the inclusion criteria and were included in this study. ACR: Albumin-to-creatinine ratio; BMI: Body mass index; eGFR: Estimated glomerular filtration rate; NHANES: National Health and Nutrition Examination Survey; PNT: Perchlorate, nitrate, and thiocyanate.

Measurement of urinary perchlorate, nitrate, and thiocyanate

As previously described,[11] urinary PNT was measured using ion chromatography coupled with electrospray tandem mass spectrometry. Chromatographic separation was performed with an Ion Pac AS16 column (Dionex, USA) with sodium hydroxide as the eluant. The eluant from the column was ionized with an electrospray interface. Individual analyte concentrations were obtained by comparing relative response factors (ratio of native analyte to stable isotope-labeled internal standard) with known standard concentrations. Details related to specimen collection and processing can be found on the NHANES website. For values lower than the limit of detection (LOD), an imputed fill value (LOD/2) was adopted to replace it. Given the screwed distributions of data, the PNT concentrations were ln-transformed before data analysis.

Measurement of chronic kidney disease

In this study, the estimated glomerular filtration rate (eGFR, mL · min−1 · 1.73 m−2) and urinary albumin-to-creatinine ratio (ACR) were calculated to assess kidney function. The ACR was calculated as albumin in spot urine (mg/dL) divided by creatinine in spot urine (g/dL). Thus, the unit of ACR was mg/g. Impaired kidney function (presence of CKD) was defined as an eGFR value <60 mL min−1·1.73 m−2, or ACR in one single-time urine sample >30 mg/g.[20]

The eGFR was computed based on the Modification of Diet in Renal Disease study equation[22]: eGFR (mL min−1 1.73 m−2) = 175 × (serum creatinine [mg/dL])−1.154 × (age, years)−0.203 × (0.742 if female) × (1.212 if African American).

Two methods were used for the adjustment of urine dilution, including traditional creatinine adjustment and covariate-adjusted creatinine standardization.[19,23] Traditional creatinine adjustment was performed by dividing the urinary PNT concentration with the observed urinary creatinine concentration (Ucr) of each spot urine sample: PNT-traditional = UC/Ucr, where PNT-traditional is the urinary PNT concentration (ng/mL) using traditional creatinine adjustment, and UC represents the measured urinary PNT concentration (ng/mL).

PNT-new = UC × (Ucr-fit/Ucr), where PNT-new denotes urinary PNT concentration (ng/mL) after covariate-adjusted creatinine standardization; UC represents the measured urinary PNT concentration (ng/mL). Linear regression models were fitted with the urine creatinine concentration of each individual as the dependent variable, and covariates that could potentially affect urine dilution were included (i.e. age, sex, race/ethnicity, body mass index [BMI], and eGFR). Ucr-fit represents fitted creatinine values obtained using the above model, and Ucr denotes observed urinary creatinine levels.


Based on previous studies, the following covariates were adjusted, including sex, age, race/ethnicity (Mexican American/other Hispanic/non-Hispanic White/non-Hispanic Black/other race, including multi-racial), BMI (kg/m2), family income-to-poverty ratio, physical activity, and cigarettes per day. In sensitivity analyses, three additional variables (healthy eating index, diabetes, and hypertension) were also adjusted. In subgroup analyses, age was divided into 20 to 39, 40 to 59, and 60 to 79 years. The family income-to-poverty ratio was divided into three categories: ≤1.85, 1.85 to 3.50, and >3.50. BMI was divided into <25, 25 to 30, and ≥30 kg/m2. The physical activity represents the total time of moderate leisure-time physical activity (LTPA) in participants (minutes per week). The total moderate LTPA time was calculated by summing the time of moderate LTPA and two times the duration of vigorous LTPA (time of vigorous LTPA × 2). Participants were divided into active (above median) and non-active (below median) groups, according to the median value of LTPA. Details of LTPA calculation have been introduced in a previous study.[24] Participants with zero cigarettes smoked per day were defined as non-smokers, and those with more than zero were defined as smokers. The Healthy Eating Index-2015 (HEI-2015) is a measure for evaluating whether a set of foods is aligned with the Dietary Guidelines for Americans; thus, the HEI-2015 was used to represent the diet quality of participants.[25] Based on the 13 components, the HEI-2015 is scored from zero to 100, with a higher score indicating a healthier diet. Diabetes and hypertension were confirmed by self-reported questionnaires.

Data analysis

Continuous variables are shown as mean ± standard deviation (SD), and/or median (25th–75th percentile). Categorical or ordinal variables are shown as frequency (%). Variations in continuous variables (eGFR and ACR in Table 1) were compared using the Student's t-test (replaced with Mann-Whitney U-test if the data were non-normally distributed) and one-way analysis of variance (ANOVA) test (replaced with Kruskal-Wallis test if the data were non-normally distributed), respectively. Survey-weighted multivariable linear regression models were conducted to calculate β values and corresponding 95% confidence intervals (CIs) to quantitatively evaluate potential associations between PNT exposures and outcomes (eGFR or ACR). Exposures were ln-transformed including ln (PNT), ln (PNT-traditional), and ln (PNT-new). Survey-weighted multivariable logistic regression models were used to calculate odds ratios (ORs) and 95% CIs for associations between PNT exposures and the outcome (presence of CKD). Covariates were adjusted in their continuous forms if possible (e.g. cigarettes per day was used instead of its derived dichotomous variable smoker/non-smoker). To explore potential non-linear relationships, we first used regressions of restricted cubic splines (RCSs) to assess the potentially non-linear relationships between PNT exposure and outcomes. The Akaike information criterion (AIC) was used to choose the most suitable knots that had the smallest AIC. P-non-linear values were computed using the “ANOVA” function in the “rms” package to evaluate the statistical significance of dose–response associations. Additionally, linear trend tests were performed by modeling the categorized PNT as ordinal variables. Stratified analyses were conducted based on age, sex, race/ethnicity, family income-to-poverty ratio, BMI, smoking status, and physical activity. Statistical significances of the interaction between PNT and covariates on outcomes were tested by the Log likelihood ratio test. All tests were two-tailed and the level of statistical significance was set to 0.05. R 4.1.1 (The R Project for Statistical Computing, Vienna, Austria) was used for all analyses.

Table 1 - Demographic characteristics and kidney function state of the participants included in analyses for associations between PNT exposure and the outcome.
Variables Values (N = 13,373) eGFR (mL·min−1·1.73 m−2) P values ACR (mg/g) P values
Age (years) 49.4 ± 17.6 <0.001 <0.001
 20–39 4533 (33.9) 108.3 (94.6–124.6) 5.8 (3.9–9.4)
 40–59 4466 (33.4) 94.3 (80.3–109.6) 6.6 (4.5–12.2)
 60–79 4374 (32.7) 78.4 (63.7–94.1) 9.8 (5.7–23.8)
Sex <0.001 <0.001
 Male 6771 (50.6) 101.6 (85.5–119.1) 6.1 (4.0–12.2)
 Female 6602 (49.4) 87.4 (72.1–104.0) 7.9 (5.3–14.9)
Race/ethnicity (%) <0.001 <0.001
 Mexican American 2190 (16.4) 106.8 (88.9–126.0) 7.4 (4.8–14.3)
 Other Hispanic 1199 (9.0) 100.0 (83.7–118.9) 7.2 (4.8–13.6)
 Non-Hispanic White 6230 (46.6) 89.3 (73.5–106.1) 7.0 (4.5–13.5)
 Non-Hispanic Black 2687 (20.1) 92.5 (77.1–108.7) 6.7 (4.1–14.4)
 Other race, including multi-racial 1067 (8.0) 101.7 (85.8–119.1) 6.9 (4.7–12.5)
Family income-to-poverty ratio 2.6 ± 1.6 <0.001 <0.001
 ≤1.85 5867 (43.9) 97.4 (78.9–117.3) 7.7 (4.8–16.3)
 1.85–3.50 3302 (24.7) 94.5 (77.3–112.7) 7.0 (4.5–14.1)
 >3.50 4204 (31.4) 91.5 (77.5–106.6) 6.2 (4.2–10.9)
Education levels (%) <0.001 <0.001
 <9th grade 1509 (11.3) 97.4 (77.1–119.3) 9.0 (5.4–21.2)
 9th–12th grade or equivalent 5073 (37.9) 95.3 (77.5–114.1) 7.3 (4.7–14.9)
 College or above 6783 (50.7) 93.4 (78.4–109.8) 6.5 (4.3–11.8)
 Unknown 8 (0)
BMI (kg/m2) 29.0 ± 6.7 <0.001 <0.001
 <25 3938 (29.4) 96.4 (79.8–115.1) 7.0 (4.7–13.4)
 25–30 4523 (33.8) 94.6 (77.7–111.1) 6.6 (4.3–12.4)
 ≥30 4912 (36.7) 92.8 (76.3–111.0) 7.5 (4.6–15.5)
Cigarettes per day 0 (0–10.0)/6.6 ± 11.6
Smoking status 0.006 <0.001
 Non-smokers 7203 (53.9) 95.2 (78.3–113.5) 6.9 (4.5–13.1)
 Smokers 6170 (46.1) 93.7 (77.6–111.1) 7.2 (4.6–14.5)
LTPA 120.0 (0–420.0)/341.9 ± 633.2 <0.001 <0.001
 Below median (non-active) 6368 (47.6) 93.3 (75.3–111.3) 7.8 (4.9–16.3)
 Above median (active) 7005 (52.4) 95.8 (80.0–113.2) 6.5 (4.2–11.8)
Alcohol consumption 0 (0–0.4)/0.4 ± 1.1 <0.001 <0.001
 Low (below median) 6340 (47.4) 90.9 (74.4–110.1) 7.7 (4.9–16.2)
 High (above median) 6733 (50.3) 97.6 (81.7–114.5) 6.4 (4.3–11.8)
 Unknown 300 (2.2)
Diet quality (HEI-2015) 53.3 ± 13.6 <0.001 0.553
 Low (below median) 5599 (41.9) 97.3 (80.8–115.0) 6.9 (4.4–12.7)
 High (above median) 5600 (41.9) 92.3 (75.7–109.9) 7.0 (4.6–13.7)
 Unknown 2174 (16.3)
Take medicine <0.001 <0.001
 No 5772 (43.2) 103.5 (88.9–120.4) 6.1 (4.0–10.1)
 Yes 7601 (56.8) 86.8 (71.1–104.3) 8.1 (5.0–17.9)
Serum total cholesterol level (mg/dL) 195.4 ± 42.1
Cardiovascular diseases <0.001 <0.001
 No 498 (3.7) 95.2 (78.8–112.9) 6.9 (4.5–13.2)
 Yes 12,875 (96.3) 76.7 (61.6–92.9) 12.4 (6.2–41.8)
Diabetes <0.001 <0.001
 No 11,512 (86.1) 95.7 (79.7–113.0) 6.6 (4.3–11.8)
 Yes 1861 (13.9) 85.5 (67.1–105.8) 12.8 (6.3–41.0)
Hypertension <0.001 <0.001
 No 10,764 (80.5) 96.6 (80.2–114.1) 6.4 (4.3–11.2)
 Yes 2609 (19.5) 86.1 (69.0–103.2) 12.2 (6.4–33.0)
Current CKD <0.001 <0.001
 No 11,067 (82.8) 97.5 (82.9–114.4) 6.2 (4.2–9.8)
 Yes 2306 (17.2) 67.4 (52.7–95.7) 44.8 (17.1–102.6)
Data were shown as mean ± SD, median (25th–75th percentile), or frequency (%). Minutes per week. ACR: Albumin-to-creatinine ratio; BMI: Body mass index; CKD: Chronic kidney disease; eGFR: Estimated glomerular filtration rate; HEI: Healthy eating index; LTPA: Leisure-time physical activity; SD: Standard deviation.


Baseline characteristics of study participants

The baseline demographic information for all participants are presented in [Table 1]. The median (25th–75th percentile) concentrations for perchlorate, nitrate, and thiocyanate were 1.2 (0.6–1.7) ng/mL, 44,600.0 (26,100.0–69,500.0) ng/mL, and 1100.0 (515.0–2480.0) ng/mL, respectively. Among this US general population, 2306 (17.2%) exhibited eGFR or ACR levels that could be classified as CKD according to either eGFR or ACR criteria. Among 13,373 participants, the mean and SD of age was 49.4 ± 17.6 years, and 50.6% of participants were male. Non-Hispanic White accounted for 46.6% of the total population. Older participants tended to have lower eGFR and higher ACR levels. The non-Hispanic White subgroup had the lowest eGFR level. Participants with a higher income-to-poverty ratio, higher BMI, lower LTPA, higher HEI-2015, co-existing hypertension, diabetes, and those who were smokers tended to have lower eGFR levels [Table 1]. The other information is displayed in Table 1 in detail.

Association between urinary PNT and kidney function

As shown in Tables 2 and 3, for perchlorate without adjustment with urinary creatinine (non-CR-adjusted perchlorate), urinary perchlorate was negatively associated with eGFR (P = 0.011) and ACR (P < 0.001), and no association was found between perchlorate and the prevalence of CKD (P = 0.815) after adjusting for confounders. After covariate-adjusted creatinine standardization, perchlorate (P-new) was negatively associated with ACR (P < 0.001). After traditional creatinine adjustment, perchlorate (P-traditional) was positively associated with eGFR (P < 0.001) and negatively associated with ACR (P = 0.001).

Table 2 - Associations between urinary PNT and eGFR, and risk of CKD.
Items Non-adjusted β (95% CI) P values Adjusted β (95% CI) P values Adjusted OR (95% CI) P values
PNT non-CR-adjusted
Perchlorate 0.64 (0.10, 1.19) 0.022 −0.61 (−1.06, −0.15) 0.011 0.99 (0.93, 1.06) 0.815
Nitrate 4.30 (3.63, 4.96) <0.001 0.67 (0.13, 1.21) 0.017 0.81 (0.75, 0.88) <0.001
Thiocyanate 3.21 (2.75, 3.68) <0.001 0.87 (0.50, 1.24) <0.001 0.89 (0.84, 0.94) <0.001
Perchlorate 1.31 (0.56, 2.05) 0.001 −0.20 (−0.75, 0.36) 0.491 1.01 (0.93, 1.11) 0.775
Nitrate 8.27 (7.29, 9.25) <0.001 2.45 (1.65, 3.25) <0.001 0.74 (0.66, 0.81) <0.001
Thiocyanate 3.56 (3.06, 4.06) <0.001 1.26 (0.83, 1.70) <0.001 0.90 (0.85, 0.94) <0.001
Perchlorate −0.91 (−1.57, −0.25) 0.008 2.75 (2.25, 3.26) <0.001 0.92 (0.84, 1.01) 0.068
Nitrate 4.28 (3.33, 5.23) <0.001 7.37 (6.54, 8.20) <0.001 0.64 (0.58, 0.71) <0.001
Thiocyanate 2.65 (2.15, 3.16) <0.001 2.70 (2.29, 3.11) <0.001 0.86 (0.81, 0.90) <0.001
CI: Confidence intervals; OR: Odds ratio; eGFR: Estimated glomerular filtration rate; PNT: Perchlorate, nitrate, and thiocyanate; CR, creatinine.PNT was ln-transformed. Adjusted model was adjusted for: age, sex, race, family income-to-poverty ratio, body mass index, physical activity, and cigarettes per day.
After covariate-adjusted creatinine standardization.
After traditional creatinine adjustment.
OR and 95% CIs for CKD in logistic regressions.

Table 3 - Associations between urinary PNT levels and ACR.
Items Non-adjusted β (95% CI) P values Adjusted β (95% CI) P values
PNT non-CR-adjusted
Perchlorate −0.06 (−0.08, −0.04) <0.001 −0.04 (−0.06, −0.02) <0.001
Nitrate −0.13 (−0.15, −0.11) <0.001 −0.07 (−0.09, −0.05) <0.001
Thiocyanate −0.07 (−0.08, −0.05) <0.001 −0.03 (−0.05, −0.01) 0.001
Perchlorate −0.09 (−0.11, −0.06) <0.001 −0.05 (−0.08, −0.03) <0.001
Nitrate −0.24 (−0.27, −0.21) <0.001 −0.12 (−0.15, −0.09) <0.001
Thiocyanate −0.07 (−0.09, −0.06) <0.001 1.26 (0.83, 1.70) <0.001
Perchlorate 0.01 (−0.02, 0.03) 0.615 −0.05 (−0.07, −0.02) 0.001
Nitrate −0.08 (−0.11, −0.05) <0.001 −0.10 (−0.14, −0.07) <0.001
Thiocyanate −0.03 (−0.05, −0.02) <0.001 −0.02 (−0.04, −0.01) 0.009
ACR: Albumin-to-creatinine ratio; BMI: Body mass index; CI: Confidence interval; PNT: Perchlorate, nitrate, and thiocyanate; CR, creatinine.PNT and ACR were ln-transformed. Adjusted model was adjusted for: age, sex, race, family income-to-poverty ratio, BMI, physical activity, and cigarettes per day.
After covariate-adjusted creatinine standardization.
After traditional creatinine adjustment.

For non-CR-adjusted nitrate, participants with higher urinary concentrations exhibited higher eGFR levels (P = 0.017), lower ACR levels (P < 0.001), and lower risk of CKD (P < 0.001). After both the traditional (N-traditional) and covariate-adjusted (N-new) creatinine adjustment, urinary nitrates were positively associated with eGFR (all P values <0.001) and negatively associated with ACR (all P values <0.001), and higher nitrate was associated with lower risk of CKD (P < 0.001) [Tables 2 and 3].

For thiocyanate, results from the multivariable models were similar to those observed in nitrate, after adjusting confounding factors. Participants with higher urinary thiocyanate levels showed a higher eGFR, lower ACR, and lower risk of CKD in all regression models that analyzed thiocyanate concentrations after non-CR adjustment, traditional CR adjustment (T-traditional), and covariate-adjusted (T-new) CR standardization (all P values <0.05) [Tables 2 and 3].

Evaluation of dose–response relationship between PNT and kidney function

The curves of RCS regression showed that most of the associations of PNT with eGFR or ACR were non-linear [Figure 2]. The negative association of urinary non-CR-adjusted perchlorate and eGFR was linear [Figure 2A] whereas the negative association of P-new and eGFR [Figure 2D], and the positive association of P-traditional and eGFR [Figure 2G], were non-linear (both P-non-linear values <0.05). The L-shaped associations between non-CR-adjusted perchlorate [Figure 2J], P-new [Figure 2M], or P-traditional [Figure 2P] and ACR were also statistically non-linear (all P-non-linear values <0.05).

Figure 2:
The estimated β values (the red lines) and 95% confidence intervals (the dotted lines) for associations between PNT exposure and eGFR or ACR. (A) association between perchlorate and eGFR; (B) association between nitrate and eGFR; (C) association between thiocyanate and eGFR; (D) association between perchlorate-new (P-new) and eGFR; (E) association between nitrate-new (N-new) and eGFR; (F) association between thiocyanate-new (T-new) and eGFR; (G) association between perchlorate-traditional (P-traditional) and eGFR; (H) association between nitrate-traditional (N-traditional) and eGFR; (I) association between thiocyanate-traditional (T-traditional) and eGFR; (J) association between perchlorate and ACR; (K) association between nitrate and ACR; (L) association between thiocyanate and ACR; (M) association between perchlorate-new (P-new) and ACR; (N) association between nitrate-new (N-new) and ACR; (O) association between thiocyanate-new (T-new) and ACR; (P) association between perchlorate-traditional (P-traditional) and ACR; (Q) association between nitrate-traditional (N-traditional) and ACR; (R) association between thiocyanate-traditional (T-traditional) and ACR. PNT: Perchlorate, nitrate, and thiocyanate; P/N/T-new: PNT After covariate-adjusted creatinine standardization; P/N/T-traditional: After traditional creatinine adjustment; ACR: Albumin-to-creatinine ratio; eGFR: Estimated glomerular filtration rate; RCSs: Restricted cubic splines.

The positive associations between non-CR-adjusted nitrate [Figure 2B], N-new [Figure 2E], or N-traditional [Figure 2H], and eGFR were non-linear (all P-non-linear values <0.05). The negative associations between exposure to nitrate (both non-CR-adjusted and CR-adjusted concentrations) and ACR were L-shaped, and associations among higher concentrations tended to be insignificant [Figure 2K,N,Q]. Similarly, L-shaped non-linear associations between thiocyanate and outcomes were observed in Figure 2.

With respect to the prevalence of CKD, a significant protective role of nitrate and thiocyanate in decreasing CKD incidence was observed [Figure 3], and the associations of thiocyanate and CKD tended to be L-shaped. Moreover, an L-shaped curve was also observed for the associations between P-traditional and the prevalence of CKD [Figure 3G].

Figure 3:
The estimated ORs (the red lines) and 95% confidence intervals (the dotted lines) for associations between PNT exposure and risk of CKD. (A) association between perchlorate and CKD; (B) association between nitrate and CKD; (C) association between thiocyanate and CKD; (D) association between perchlorate-new (P-new) and CKD; (E) association between nitrate-new (N-new) and CKD; (F) association between thiocyanate-new (T-new) and CKD; (G) association between perchlorate-traditional (P-traditional) and CKD; (H) association between nitrate-traditional (N-traditional) and CKD; (I) association between thiocyanate-traditional (T-traditional) and CKD. PNT: Perchlorate, nitrate, and thiocyanate; P/N/T-new: PNT After covariate-adjusted creatinine standardization; P/N/T-traditional: After traditional creatinine adjustment; CKD: Chronic kidney disease; OR: Odds ratio; RCSs: Restricted cubic splines.

In multivariable-adjusted models, for quartiles of PNT, statistically significant dose-response associations were observed in most relationships, except for some associations between perchlorate, P-new or P-traditional and outcomes (eGFR level and the prevalence of CKD) [Supplementary Table 1, https://links.lww.com/CM9/B421]. For example, the value of P for trend in the perchlorate quartile and CKD risk was 0.520 in the adjusted models [Supplementary Table 1, https://links.lww.com/CM9/B421]. All values of P for trend were <0.05 for PNT quartiles and ACR [Supplementary Table 2, https://links.lww.com/CM9/B421].

Stratified and sensitivity analyses

Based on several variables, stratified analyses were performed for PNT, PNT-new, or PNT-traditional and the prevalence of CKD. We found that positive associations between nitrate concentrations (nitrate, N-new, and N-traditional) and the prevalence of CKD were more significant in participants aged >40 years [Supplementary Tables 3–5, https://links.lww.com/CM9/B421]. All P values of interactions for nitrate, N-new, and N-traditional were <0.001 (the interaction P values were tested between exposure and age for the presence of CKD). The positive associations between thiocyanate concentrations (thiocyanate, T-new, and T-traditional) and the prevalence of CKD were more obvious in participants aged >40 years and in non-Hispanic White [Supplementary Tables 3–5, https://links.lww.com/CM9/B421]; significant interactions were found between thiocyanate (P for interaction: <0.001), T-new (P for interaction: 0.003), and T-traditional (P for interaction: <0.001), and age on the outcome. Results of the other stratified analyses are shown in Supplementary Tables 3–5, https://links.lww.com/CM9/B421 in detail.

In the first sensitivity analysis, additional covariates were adjusted including age, sex, race/ethnicity, family income-to-poverty ratio, education, BMI, HEI-2015, alcohol consumption, physical activity, cigarettes per day, and medicine use. Most results were consistent with the primary multivariable regressions [Supplementary Tables 6, 7, https://links.lww.com/CM9/B421]. In the second sensitivity analysis, in addition to variables in the first sensitivity analysis, serum total cholesterol level, diabetes, hypertension, and cardiovascular disease were also adjusted in the multivariable regressions. The results were still consistent with those in the above analyses [Supplementary Tables 8, 9, https://links.lww.com/CM9/B421]. Supplementary Table 8, https://links.lww.com/CM9/B421 shows the associations between exposures and eGFR and the presence of CKD. Supplementary Table 9, https://links.lww.com/CM9/B421 presents the associations between exposures and ACR, respectively. Further, in the sensitivity analysis among a total of 6859 participants with an eGFR within the normal range (60 to 120 mL·min−1·1.73 m−2), we found that associations between PNT and eGFR or ACR were generally similar to those in the total population. Given the small sample size, the absolute values of the effect estimates became smaller [Supplementary Table 10, https://links.lww.com/CM9/B421]. In the final sensitivity analysis, after excluding participants with kidney stones, we still observed some significant associations between exposures and outcomes, especially for nitrate, N-new, and N-traditional [Supplementary Tables 11 and 12, https://links.lww.com/CM9/B421 and 12, https://links.lww.com/CM9/B421].


This study systemically investigated the relationship between urinary PNT and the prevalence of CKD in a nationally representative population. We found that nitrate and thiocyanate were significantly associated with higher levels of eGFR, lower levels of ACR, and lower risk of CKD. Perchlorate was also observed to be associated with ACR (all concentrations of perchlorate with or without CR adjustment) or eGFR (only P-traditional). Overall, in quartile analysis, we observed dose–response associations between PNT and kidney function in most relationships in this population. Notably, some non-linear relationships were found between exposures and outcomes, such as the L-shaped associations between thiocyanate and ACR or CKD prevalence.

To increase shelf life and to avoid bacterial growth, nitrates and nitrites are widely used as food additives in processed meats, and high consumption of processed meats is associated with a higher incidence of upper gastrointestinal tract cancers.[26-28] Based on recommendations of the World Health Organization, the acceptable daily intake of nitrate is 0 to 3.7 mg/kg body weight (http://www.inchem.org/documents/jecfa/jecmono/v50je07.htm). However, previous studies have suggested the existence of a possible protective effect of nitrate on human health.[17,29-31] Some evidence has shown that the beneficial effects of nitrates on cardiovascular, renal, and metabolic functions might be associated with the nitrate–nitrite–nitric oxide (NO) pathway alongside the more classical L-arginine–NO synthase (NOS) pathway.[32,33] NO could regulate vascular homeostasis, neurotransmission, and host defense. The nitrite reduction to NO becomes more pronounced when oxygen availability is limited and NOS activity is decreased. Consequently, in situations where systemic and regional ischemia prevails, it is assumed to be beneficial to increase nitrate and nitrite stores pharmacologically or via dietary nitrate intake.[4] This is supported by the results of some animal studies. In mouse models, Zhang et al[34] found that acute treatment with inorganic nitrate before renal ischemia may function as a novel therapeutic approach to prevent acute kidney injury. Li et al[35] demonstrated that inorganic nitrate and nitrite treatment attenuates the risk of kidney fibrosis by targeting oxidative stress and lipid metabolism. In the current study, consistent with the effects reported in previous studies, we illustrated the beneficial role of nitrates for kidney function in the general population.

We also found that higher urinary thiocyanate was associated with better kidney functions. The results remained unchanged when the thiocyanate concentration was adjusted by the urinary CR level. Notably, similar to nitrate, age may be an interactive factor for associations between thiocyanate and kidney functions, which indicates that older individuals might benefit more from thiocyanate exposure. Thiocyanate was a major indirect exposure marker of cyanide in smokers. In non-smokers, thiocyanate is sourced exogenously from the digestion of vegetables, especially cruciferous vegetables.[36] In this study, we found that the protective effect of thiocyanate was more obvious in non-smokers [Supplementary Table 5, https://links.lww.com/CM9/B421], which suggests that urinary thiocyanate might be mainly derived from food for these participants. Recent animal studies have demonstrated the anti-inflammatory and antimicrobial roles of thiocyanate in cardiovascular and pulmonary animal models, which has implications for future treatment of atherogenesis and infectious lung disease.[37-39] However, the anti-inflammatory effect of thiocyanate remains controversial. For example, a study found that high doses of hypothiocyanite (hypothiocyanite is generated by the catalysis of peroxidases using thiocyanate transported via several anion transporters) causes necrosis, and low doses can activate the NF-κB pathway, which enables the allergic inflammation of the airway.[409] However, limited animal or epidemiological studies have focused on the effect of hypothiocyanite on renal functions. In this study, the L-shaped associations between thiocyanate and CKD indicated that the dose-dependent beneficial role of thiocyanate had an optimal range of concentration, which suggests that high concentrations of thiocyanate do not have a protective effect on the kidneys. Further prospective studies should be conducted to confirm our findings and to explore the optimal thiocyanate concentrations, and potential mechanisms of the beneficial effect of thiocyanate on renal function.

Food is the primary source of perchlorate for most US adults.[3] The thyroid gland is the major target of perchlorate toxicity in humans. However, except for its adverse effect on thyroid function,[8] the other health effects of perchlorate have rarely been studied. The impact of perchlorate on human health remains important but it is not fully comprehended.[41] Future studies should be performed to evaluate its health effects, such as the effect on the kidney function of perchlorate at environmental exposure levels. In this study, significant associations between perchlorate and the prevalence of CKD were only found when the perchlorate concentration was adjusted by creatinine, and a significant L-shaped association was observed between perchlorate and ACR for all perchlorate concentrations before and after CR adjustment. These results indicate that environmental exposure to perchlorate may also have a beneficial role in kidney function.

Both strengths and limitations exist in this study. First, the analyses were conducted in a large and nationally representative population. Second, we applied multiple statistical models, including traditional linear regression and RCSs, to examine the associations of exposure to PNT with renal functions after adjusting multiple confounding factors; consistent results were observed in most of the subgroup and sensitivity analyses. Third, we adjusted concentrations of PNT by urinary creatinine based on two types of methods, which makes the results more reliable. Several limitations of this study should be mentioned as well. First, causal associations cannot be concluded because of the cross-sectional nature of this research. In this study, reverse causality may occur, given that reduced glomerular filtration may decrease urinary excretion of chemicals.[42] Second, some unmeasured confounding factors could not be adjusted in this study. Third, the non-linear associations in this study were difficult to clearly explain. Fourth, the diagnosis of CKD in this study was based on biological parameters, and a pathological diagnosis was unavailable to confirm the diagnosis. Finally, given the large temporal variation in individual exposure and the short biological half-lives of PNT, single measurements of urinary PNT may not represent long-term exposure.

To conclude, our results indicated possible associations between PNT exposure and kidney function among the general adult population in the US. It should be noted that our results were based on the levels of environmental exposure to PNT in the general population of the US, and outside validations were necessary to verify our conclusions. Notably, different levels of PNT exposure may lead to distinct outcomes, and the optimal levels of PNT exposure to human health should be analyzed further. Additionally, to identify individuals more sensitive to PNT exposure, interactions between PNT exposure and the other clinicopathological characteristics need to be explored in future studies. Thus, more epidemiological studies with distinct conditions or populations should be conducted in the future. Besides, more prospective cohort studies and well-designed toxicological experiments should be performed to elucidate whether these relationships are causal.

Availability of data and materials

The National Health and Nutrition Examination Survey (NHANES) data are publicly available at https://www.cdc.gov/nchs/nhanes/index.htm.

Conflicts of interest



1. Pleus RC, Corey LM. Environmental exposure to perchlorate: a review of toxicology and human health. Toxicol Appl Pharmacol 2018;358:102–109. doi: 10.1016/j.taap.2018.09.001.
2. Renner R. Perchlorate in food. Environ Sci Technol 2008;42:1817. doi: 10.1021/es0870552.
3. Kumarathilaka P, Oze C, Indraratne SP, Vithanage M. Perchlorate as an emerging contaminant in soil, water and food. Chemosphere 2016;150:667–677. doi: 10.1016/j.chemosphere.2016.01.109.
4. Babateen AM, Fornelli G, Donini LM, Mathers JC, Siervo M. Assessment of dietary nitrate intake in humans: a systematic review. Am J Clin Nutr 2018;108:878–888. doi: 10.1093/ajcn/nqy108.
5. Leung AM, Braverman LE, He X, Schuller KE, Roussilhes A, Jahreis KA, et al. Environmental perchlorate and thiocyanate exposures and infant serum thyroid function. Thyroid 2012;22:938–943. doi: 10.1089/thy.2012.0058.
6. Maiti D, Ansari I, Rather MA, Deepa A. Comprehensive review on wastewater discharged from the coal-related industries characteristics and treatment strategies. Water Sci Technol 2019;79:2023–2035. doi: 10.2166/wst.2019.195.
7. Eguchi A, Kunisue T, Wu Q, Trang PT, Viet PH, Kannan K, et al. Occurrence of perchlorate and thiocyanate in human serum from e-waste recycling and reference sites in Vietnam: association with thyroid hormone and iodide levels. Arch Environ Contam Toxicol 2014;67:29–41. doi: 10.1007/s00244-014-0021-y.
8. Tarone RE, Lipworth L, McLaughlin JK. The epidemiology of environmental perchlorate exposure and thyroid function: a comprehensive review. J Occup Environ Med 2010;52:653–660. doi: 10.1097/JOM.0b013e3181e31955.
9. Bahadoran Z, Mirmiran P, Ghasemi A, Kabir A, Azizi F, Hadaegh F. Is dietary nitrate/nitrite exposure a risk factor for development of thyroid abnormality? A systematic review and meta-analysis. Nitric Oxide 2015;47:65–76. doi: 10.1016/j.niox.2015.04.002.
10. Liu G, Zong G, Dhana K, Hu Y, Blount BC, Morel-Espinosa M, et al. Exposure to perchlorate, nitrate and thiocyanate, and prevalence of diabetes mellitus. Int J Epidemiol 2017;46:1913–1923. doi: 10.1093/ije/dyx188.
11. Zhu F, Huang M, Jiao J, Zhuang P, Mao L, Zhang Y. Environmental exposure to perchlorate, nitrate, and thiocyanate in relation to obesity: a population-based study. Environ Int 2019;133:105191. doi: 10.1016/j.envint.2019.105191.
12. Shiue I. Urinary thiocyanate concentrations are associated with adult cancer and lung problems: US NHANES, 2009–2012. Environ Sci Pollut Res Int 2015;22:5952–5960. doi: 10.1007/s11356-014-3777-8.
13. King L, Huang Y, Li T, Wang Q, Li W, Shan Z, et al. Associations of urinary perchlorate, nitrate and thiocyanate with central sensitivity to thyroid hormones: a US population-based cross-sectional study. Environ Int 2022;164:107249. doi: 10.1016/j.envint.2022.107249.
14. Zhao X, Zhou P, Chen X, Li X, Ding L. Perchlorate-induced oxidative stress in isolated liver mitochondria. Ecotoxicology 2014;23:1846–1853. doi: 10.1007/s10646-014-1312-9.
15. Larsson-Nyrén G, Sehlin J, Rorsman P, Renström E. Perchlorate stimulates insulin secretion by shifting the gating of L-type Ca2+ currents in mouse pancreatic B-cells towards negative potentials. Pflugers Arch 2001;441:587–595. doi: 10.1007/s004240000426.
16. Wang L, Fu Z, Zheng J, Wang S, Ping Y, Gao B, et al. Exposure to perchlorate, nitrate and thiocyanate was associated with the prevalence of cardiovascular diseases. Ecotoxicol Environ Saf 2022;230:113161. doi: 10.1016/j.ecoenv.2022.113161.
17. Bondonno CP, Dalgaard F, Blekkenhorst LC, Murray K, Lewis JR, Croft KD, et al. Vegetable nitrate intake, blood pressure and incident cardiovascular disease: Danish diet, cancer, and health study. Eur J Epidemiol 2021;36:813–825. doi: 10.1007/s10654-021-00747-3.
18. GBD Chronic Kidney Disease Collaboration. Global, regional, and national burden of chronic kidney disease, 1990–2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2020;395:709–733. doi: 10.1016/s0140-6736(20)30045-3.
19. Kang H, Lee JP, Choi K. Exposure to phthalates and environmental phenols in association with chronic kidney disease (CKD) among the general US population participating in multi-cycle NHANES (2005-2016). Sci Total Environ 2021;791:148343. doi: 10.1016/j.scitotenv.2021.148343.
20. Kang H, Lee J, Lee JP, Choi K. Urinary metabolites of organophosphate esters (OPEs) are associated with chronic kidney disease in the general US population, NHANES 2013-2014. Environ Int 2019;131:105034. doi: 10.1016/j.envint.2019.105034.
21. Luo J, Hendryx M. Metal mixtures and kidney function: an application of machine learning to NHANES data. Environ Res 2020;191:110126. doi: 10.1016/j.envres.2020.110126.
22. Levey AS, Coresh J, Greene T, Stevens LA, Zhang YL, Hendriksen S, et al. Using standardized serum creatinine values in the modification of diet in renal disease study equation for estimating glomerular filtration rate. Annals of internal medicine 2006;145:247–254. doi:10.7326/0003-4819-145-4-200608150-00004.
23. Guo J, Wu C, Zhang J, Chang X, Zhang Y, Cao Y, et al. Associations of melamine and cyanuric acid exposure with markers of kidney function in adults: results from NHANES 2003-2004. Environ Int 2020;141:105815. doi: 10.1016/j.envint.2020.105815.
24. Zhang YB, Chen C, Pan XF, Guo J, Li Y, Franco OH, et al. Associations of healthy lifestyle and socioeconomic status with mortality and incident cardiovascular disease: two prospective cohort studies. BMJ 2021;373:n604. doi: 10.1136/bmj.n604.
25. Jayanama K, Theou O, Godin J, Cahill L, Shivappa N, Hébert JR, et al. Relationship between diet quality scores and the risk of frailty and mortality in adults across a wide age spectrum. BMC Med 2021;19:64. doi: 10.1186/s12916-021-01918-5.
26. Song P, Wu L, Guan W. Dietary nitrates, nitrites, and nitrosamines intake and the risk of gastric cancer: a meta-analysis. Nutrients 2015;7:9872–9895. doi: 10.3390/nu7125505.
27. Picetti R, Deeney M, Pastorino S, Miller MR, Shah A, Leon DA, et al. Nitrate and nitrite contamination in drinking water and cancer risk: a systematic review with meta-analysis. Environ Res 2022;210:112988. doi: 10.1016/j.envres.2022.112988.
28. Kotopoulou S, Zampelas A, Magriplis E. Dietary nitrate and nitrite and human health: a narrative review by intake source. Nutr Rev 2022;80:762–773. doi: 10.1093/nutrit/nuab113.
29. Eriksson KE, Eidhagen F, Liska J, Franco-Cereceda A, Lundberg JO, Weitzberg E. Effects of inorganic nitrate on ischaemia-reperfusion injury after coronary artery bypass surgery: a randomised controlled trial. Br J Anaesth 2021;127:547–555. doi: 10.1016/j.bja.2021.06.046.
30. Li X, Zhang W, Laden F, Curhan GC, Rimm EB, Guo X, et al. Dietary nitrate intake and vegetable consumption, ambient particulate matter, and risk of hypertension in the Nurses’ Health study. Environ Int 2022;161:107100. doi: 10.1016/j.envint.2022.107100.
31. Liu AH, Bondonno CP, Russell J, Flood VM, Lewis JR, Croft KD, et al. Relationship of dietary nitrate intake from vegetables with cardiovascular disease mortality: a prospective study in a cohort of older Australians. Eur J Nutr 2019;58:2741–2753. doi: 10.1007/s00394-018-1823-x.
32. Lundberg JO, Carlström M, Weitzberg E. Metabolic effects of dietary nitrate in health and disease. Cell Metab 2018;28:9–22. doi: 10.1016/j.cmet.2018.06.007.
33. Lundberg JO, Weitzberg E, Gladwin MT. The nitrate-nitrite-nitric oxide pathway in physiology and therapeutics. Nat Rev Drug Discov 2008;7:156–167. doi: 10.1038/nrd2466.
34. Zhang G, Han H, Zhuge Z, Dong F, Jiang S, Wang W, et al. Renovascular effects of inorganic nitrate following ischemia-reperfusion of the kidney. Redox Biol 2021;39:101836. doi: 10.1016/j.redox.2020.101836.
35. Li X, Zhuge Z, Carvalho L, Braga VA, Lucena RB, Li S, et al. Inorganic nitrate and nitrite ameliorate kidney fibrosis by restoring lipid metabolism via dual regulation of AMP-activated protein kinase and the AKT-PGC1α pathway. Redox Biol 2022;51:102266. doi: 10.1016/j.redox.2022.102266.
36. Felker P, Bunch R, Leung AM. Concentrations of thiocyanate and goitrin in human plasma, their precursor concentrations in brassica vegetables, and associated potential risk for hypothyroidism. Nutr Rev 2016;74:248–258. doi: 10.1093/nutrit/nuv110.
37. Chandler JD, Day BJ. Biochemical mechanisms and therapeutic potential of pseudohalide thiocyanate in human health. Free Radic Res 2015;49:695–710. doi: 10.3109/10715762.2014.1003372.
38. Chandler JD, Min E, Huang J, Nichols DP, Day BJ. Nebulized thiocyanate improves lung infection outcomes in mice. Br J Pharmacol 2013;169:1166–1177. doi: 10.1111/bph.12206.
39. Morgan PE, Laura RP, Maki RA, Reynolds WF, Davies MJ. Thiocyanate supplementation decreases atherosclerotic plaque in mice expressing human myeloperoxidase. Free Radic Res 2015;49:743–749. doi: 10.3109/10715762.2015.1019347.
40. Suzuki S, Ogawa M, Ohta S, Nunomura S, Nanri Y, Shiraishi H, et al. Induction of airway allergic inflammation by hypothiocyanite via epithelial cells. J Biol Chem 2016;291:27219–27227. doi: 10.1074/jbc.M116.746909.
41. Soldin OP, Braverman LE, Lamm SH. Perchlorate clinical pharmacology and human health: a review. Ther Drug Monit 2001;23:316–331. doi: 10.1097/00007691-200108000-00002.
42. Williams JK, Smallwood MJ, Benjamin N, D'Souza RJ, Shore AC, Winyard PG, et al. Renal nitrate clearance in chronic kidney disease. Nitric Oxide 2020;97:16–19. doi: 10.1016/j.niox.2020.01.011.

Perchlorate; Nitrate; Thiocyanate; Chronic kidney disease; National Health and Nutrition Examination Survey

Supplemental Digital Content

Copyright © 2023 The Chinese Medical Association, produced by Wolters Kluwer, Inc. under the CC-BY-NC-ND license.